Manual on methodologies and criteria for Modelling and Mapping ...

Manual on methodologiesand criteria for Modellingand Mapping CriticalLoads & Levels and AirPollution Effects, Risksand Trends

Texte

5204

ISSN0722-186X

TEXTE

On behalf of the Federal Environmental Agency

UMWELTBUNDESAMT

Texte

5204

ISSN

0722-186X

Manual on methodologies and criteria for Modelling and Mapping Critical Loads & Levels and Air Pollution Effects, Risks and Trends

This Publication is also available as Download under http://www.umweltbundesamt.de The publisher does not accept responsibility for the correctness, accuracy or completeness of the information, or for the observance of the private rights of third parties. The contents of this publication do not necessarily reflect the official opinions. Publisher: Federal Environmental Agency (Umweltbundesamt) Postfach 33 00 22 14191 Berlin Tel.: +49/30/8903-0 Telex: 183 756 Telefax: +49/30/8903 2285 Internet: http://www.umweltbundesamt.de Edited by: Section II 4.4 Dr. Till Spranger (Chairman ICP Modelling & Mapping) Ullrich Lorenz Dr. Heinz-Detlef Gregor (Head of Section) Layout: Dr. Hans-Dieter Nagel Öko-Data GmbH, Strausberg Berlin, December 2004

This Manual is the basic guideline for modelling and mapping critical levels and loads and theirexceedances, and for dynamic modelling of acidification. It helps Parties to the UNECEConvention on Long-range Transboundary Air Pollution (CLRTAP) to fulfil their obligations touse harmonized methods to derive data for effects and risk assessments. This happens in theframework of the International Cooperative Programme on Modelling and Mapping CriticalLoads&Levels and Air Pollution Effects, Risks and Trends (ICP M&M). This programme wasestablished in 1988 under the leadership of Germany; it was chaired from the beginning until2002 by Heinz Gregor who strongly influenced the development of critical levels and loads, andthe role they play in European emission abatement policy.

The new and completely revised version of this Manual was mostly discussed and adopted bythe 19th meeting of the ICP M&M Task Force in Tartu, Estonia (May 2003); some additional partswere discussed and adopted by the 20th meeting in Laxenburg, Austria (May 2004). TheWorking Group on Effects in 2003 appreciated the Manual review and revision and recommended its future use to Parties. This version replaces the previous (1996) Manual, itsinterim updates, and all other guidelines published in the context of the modelling and mappingwork.

This update contains new scientific information and methods developed in the course of themodelling and mapping exercise, coordinated by the Coordination Center for Effects (CCE), andfrom recent UNECE workshops. The scientific background and relevant sources of additionalinformation to the methods described in the manual are mentioned or included in the respective chapter or annex.

National Focal Centres have strongly contributed to the method development and tested theirapplication on a national level. The complete revision of the Manual itself has only been possible due to numerous contributions by many scientists. The main authors/editors are

• Till Spranger for Ch. 1; • a revision group led by Ron Smith and David Fowler for Ch. 2; • an ICP Vegetation revision group led by Gina Mills for Ch. 3; • ICP Materials for Ch. 4; • Maximilian Posch for Chapters 5.1 – 5.4 (together with Jane Hall and many others); • the Expert Panel on Critical Loads of Heavy Metals led by Gudrun Schütze for Ch. 5.5;• Maximilian Posch, Jean-Paul Hettelingh and Jaap Slootweg for Chapter 6; • Maximilian Posch for Ch.s 7 and 8

The attractive new layout and the internet presentation has been developed by Hans-Dieter Nagel and his team. The main pathway of distribution is meant to be the internet;revisions and extensions are available at http://www.icpmapping.org .

Thanks to all contributors!

(Till Spranger, Chairman of the Task Force on Modelling and Mapping)

PREFACE

Mapping Manual 2004

1 Introduction I-1

1.1 THE CRITICAL LOAD AND LEVEL CONCEPT IN THE UNECE CONVENTION ON

LONGRANGE TRANSBOUNDARY AIR POLLUTIONI-1

1.2 AIMS OF THE ICP ON MODELLING AND MAPPING I-5

1.3. DIVISION OF TASKS WITHIN THE PROGRAMME I-61.3.1 Mandate for the Task Force of the ICP on Modelling and Mapping I-61.3.2 Mandate for the Coordination Center for Effects I-61.3.3 Responsibilities of the National Focal Centres I-7

1.4. OBJECTIVES OF THE MANUAL I-7

1.5. STRUCTURE AND SCOPE OF THE MANUAL I-8

REFERENCES I-9

2 Guidance on Mapping Concentration Levels and Deposition Loads II-12.1 GENERAL REMARKS AND OBJECTIVES II-1

2.1.1 Mapping resolution and application of Critical Loads II-2

2.2 MAPPED ITEMS II-2

2.3 METHODS OF MAPPING, THEIR UNDERLYING ASSUMPTIONS AND DATA REQUIREMENTS II-32.3.1 Linkage to emission inventories II-32.3.2 Quantification and mapping methods: Scales of time and space II-32.3.3 Mapping meteorological parameters II-72.3.4 Mapping ozone (O3) concentrations and deposition II-82.3.5 Mapping sulfur dioxide (SO2) concentrations and oxidised sulfur (SOx) deposition II-102.3.6 Mapping nitrogen oxides (NOx) concentrations and oxidised nitrogen (NOy) deposition II-12

2.3.7 Mapping ammonia (NH3) concentration, reduced nitrogen (NHx) deposition and total nitrogen deposition

II-13

2.3.8 Mapping base cation and chloride deposition II-152.3.9 Mapping total potential acid deposition II-152.3.10 Uncertainties of quantification and mapping methods II-16

2.4 USE OF DEPOSITION LOAD AND CONCENTRATION MAPS II-19

REFERENCES II-22

3 Mapping Critical Levels for Vegetation III-13.1 GENERAL REMARKS AND OBJECTIVES III-13.2 CRITICAL LEVELS FOR SO2, NOX, NH3 AND O3 III-2

3.2.1 SO2 III-23.2.2 NOx III-23.2.3 NH3 III-33.2.4 O3 III-3

3.3 SCIENTIFIC BASES OF THE CRITICAL LEVELS FOR OZONE III-93.3.1 Crops III-93.3.1.1 Crop sensitivity to ozone III-93.3.1.2 Stomatal flux-based critical levels for yield reduction in wheat and potato III-93.3.1.3 AOTX-based critical levels for crop yield reduction III-123.3.1.4 VPD-modified AOT30 - based critical level for visible injury on crops III-143.3.2 (Semi-) natural vegetation III-163.3.3 Forest trees III-18

CONTENT

Mapping Manual 2004 • Content C1

Cha

pter

IC

hapt

er II

Cha

pter

III

Cha

pter

IVC

hapt

er V

Cha

pter

VI

Cha

pter

VII

Cha

pter

VIII

3.3.3.1 Provisional flux-based critical levels III-183.3.3.2 AOTX-based critical levels for forest trees III-19

3.4 CALCULATING EXCEEDANCE OF STOMATAL FLUX-BASED CRITICAL LEVELS FOR OZONE III-213.4.1 Stages in calculating AFstY and CLef exceedance III-213.4.2 Ozone concentrations at the canopy height III-223.4.3 Accumulation period III-243.4.4 The stomatal flux algorithm III-253.4.5 Calculation of AFstY and exceedance of CLef for agricultural crops III-323.4.5.1 Ozone concentration at the top of the wheat and potato canopy III-323.4.5.2 Estimating the time period for ozone flux accumulation for wheat and potato III-323.4.5.3 Parameterisation of stomatal flux models for wheat and potato III-363.4.6 Calculation of AFstY and exceedance of CLef for forest trees III-383.4.6.1 Ozone concentration at the top of the forest canopy III-383.4.6.2 Parameterisation of the stomatal flux models for beech and birch III-38

3.5 CALCULATING EXCEEDANCE OF CONCENTRATION-BASED CRITICAL LEVELS FOR OZONE III-403.5.1 Stages in calculating AOTX and CLec exceedance III-403.5.2 Calculation of AOTX and CLec exceedance for agricultural crops III-413.5.2.1 Ozone concentrations at the canopy height for agricultural crops III-413.5.2.2 Accumulation period for agricultural crops III-413.5.3 Calculation of AOTX and CLec exceedance for horticultural crops III-423.5.3.1 Ozone concentrations at the canopy height for horticultural crops III-423.5.3.2 Accumulation period for horticultural crops III-423.5.4 Calculation of AOTXVPD and exceedance of the short-term critical level for ozone injury III-423.5.5 Calculation of AOTX and CLec exceedance for (semi-) natural vegetation III-433.5.5.1 Ozone concentrations at canopy height III-433.5.5.2 Time-window for calculating AOT40 for (semi-) natural vegetation III-43

3.5.5.3 Mapping (semi-) natural vegetation communities at risk from exceedance of the critical level III-44

3.5.6 Calculation of AOTX and CLec exceedance for forest trees III-443.5.6.1 Ozone concentrations at canopy height III-443.5.6.2 Time-window for calculating ozone exposure for forest trees III-44

ANNEX I MPOC APPROACH TO RISK ASSESSMENT FOR FOREST TREES III-45

ANNEX II GUIDANCE FOR ASSESSING IMPACTS OF OZONE ON VEGETATION USING THE MODELS

DESCRIBED IN THIS CHAPTERIII-46

REFERENCES III-48

4 Mapping of Effects on Materials IV-1

4.1 INTRODUCTION - OBJECTIVES, DEFINITIONS AND GENERAL REMARKS IV-1

4.2 MAPPING "ACCEPTABLE CORROSION RATES" AND "ACCEPTABLE LEVELS/LOADS" FOR

POLLUTANTSIV-2

4.2.1 Introduction IV-24.2.2 Dose-response functions IV-24.2.3 Calculation and mapping of acceptable levels/loads and their exceedances IV-44.2.4 Calculations and mapping of costs resulting from corrosion IV-64.2.5 Sources of uncertainty IV-7

4.3 DIRECT EFFECTS OF OZONE IV-7

CONTENT

Mapping Manual 2004 • ContentC2

Cha

pter

IC

hapt

er II

Cha

pter

III

Cha

pter

IVC

hapt

er V

Cha

pter

VI

Cha

pter

VII

Cha

pter

VIII

4.4 DATA AND MAPPING PROCEDURE IV-84.4.1 Data and scale of mapping IV-94.4.2 Urban and rural areas IV-9

REFERENCES IV-10

5 Mapping Critical Loads V-1

5.1 INTRODUCTION V-1

5.2 EMPIRICAL CRITICAL LOADS V-25.2.1 Empirical Critical Loads of nutrient nitrogen V-25.2.1.1 Introduction V-25.2.1.2 Data V-25.2.1.3 Ecosystem classification V-35.2.1.4 Use of empirical critical loads V-65.2.1.5 Recommendations V-65.2.2 Empirical Critical Loads of acidity V-8

5.3 MODELLING CRITICAL LOADS FOR TERRESTRIAL ECOSYSTEMS V-105.3.1 Critical loads of nutrient nitrogen (eutrophication) V-11

5.3.1.1 Model derivation V-115.3.1.2 The acceptable leaching of nitrogen V-125.3.1.3 Sources and derivation of input data V-135.3.2 Critical loads of acidity V-15

5.3.2.1 Model derivation: the Simple Mass Balance (SMB) model V-155.3.2.2 Chemical criteria and the critical leaching of Acid Neutralising Capacity V-185.3.2.3 Sources and derivation of input data V-215.3.2.4 Possible extensions to the SMB model V-24

5.4 CRITICAL LOADS FOR AQUATIC ECOSYSTEMS V-295.4.1 The Steady-State Water Chemistry (SSWC) model V-295.4.1.1 Model derivation V-295.4.1.2 The F-factor V-305.4.1.3 The non-anthropogenic sulphate concentration V-315.4.1.4 The ANC-limit V-325.4.2 The empirical diatom model V-335.4.3 The First-order Acidity Balance (FAB) model V-345.4.3.1 Model derivation V-345.4.3.2 Systems of lakes V-375.4.4 Input data V-37

5.5 CRITICAL LOADS OF CADMIUM, LEAD AND MERCURY V-395.5.1 General methodological aspects of mapping critical loads of heavy metals V-395.5.1.1 Calculation of different types of critical loads V-39

5.5.1.2 Limitations in sites that allow critical load calculations V-415.5.1.3 Definitions and symbols/abbreviations used in critical load calculations V-415.5.1.4 Stand-still approach versus calculation of critical limit exceedance V-435.5.2 Terrestrial ecosystems V-445.5.2.1 Simple steady-state mass balance model and related input data V-44

5.5.2.2 Critical dissolved metal concentrations derived from critical limits in terrestrial ecosystems

V-49

CONTENT


Cha

pter

IC

hapt

er II

Cha

pter

III

Cha

pter

IVC

hapt

er V

Cha

pter

VI

Cha

pter

VII

Cha

pter

VIII

5.5.3 Aquatic ecosystems V-565.5.3.1 Critical loads of cadmium and lead V-565.5.3.2 Critical levels of mercury in precipitation V-605.5.4 Limitations in the present approach and possible future refinements V-63

REFERENCES V-65

ANNEX I TRANSFER FUNCTIONS FOR LEAD AND CADMIUM FOR THE CONVERSION OF METAL

CONCENTRATIONS IN DIFFERENT SOIL PHASESV-A1

ANNEX II CALCULATION OF TOTAL METAL CONCENTRATION FROM FREE METAL ION

CONCENTRATIONS USING THE WHAM MODELV-A5

ANNEX III CALCULATION OF THE CRITICAL TOTAL AQUEOUS CONCENTRATION FROM THE CRITICAL

DISSOLVED CONCENTRATION USING THE WHAM MODELV-A7

6 Dynamic Modelling VI-1

6.1 INTRODUCTION VI-16.1.1 Why dynamic modelling? VI-16.1.2 Constraints for dynamic modelling under the LRTAP Convention VI-3

6.2 BASIC CONCEPTS AND EQUATIONS VI-36.2.1 Charge and mass balances VI-36.2.2 From steady state (critical loads) to dynamic models VI-46.2.3 Finite buffers VI-56.2.3.1 Cation exchange VI-56.2.3.2 Nitrogen immobilisation VI-66.2.3.3 Sulphate adsorption VI-76.2.4 Biological response models VI-76.2.4.1 Terrestrial ecosystems VI-76.2.4.2 Aquatic ecosystems VI-9

6.3 AVAILABLE DYNAMIC MODELS VI-106.3.1 The VSD model VI-116.3.2 The SMART model VI-116.3.3 The SAFE model VI-116.3.4 The MAGIC model VI-12

6.4 INPUT DATA AND MODEL CALIBRATION VI-126.4.1 Input data VI-136.4.1.1 Averaging soil properties VI-146.4.1.2 Data that also used for critical load calculations VI-146.4.1.3 Data needed to simulate cation exchange VI-166.4.1.4 Data needed for balances of nitrogen, sulphate and aluminium VI-206.4.2 Model calibration VI-21

6.5 MODEL CALCULATIONS AND PRESENTATION OF MODEL RESULTS VI-226.5.1 Use of dynamic models in integrated assessment VI-226.5.2 Target load calculations VI-246.5.3 Presentation of model results VI-28

REFERENCES VI-29

CONTENT

Mapping Manual 2004 • ContentC4

Cha

pter

IC

hapt

er II

Cha

pter

III

Cha

pter

IVC

hapt

er V

Cha

pter

VI

Cha

pter

VII

Cha

pter

VIII

7 Exceedance Calculations VII-17.1 BASIC DEFINITIONS VII-1

7.2 CONDITIONAL CRITICAL LOADS OF N AND S VII-27.3 TWO POLLUTANTS VII-3

7.4 SURFACE WATERS VII-77.4.1 The SSWC model VII-77.4.2 The empirical diatom model VII-7

7.4.3 The FAB model VII-7

7.5 TARGET LOADS VII-7

REFERENCES VII-8

8 General Mapping Issues VIII-1

8.1 THE EMEP GRID VIII-18.1.1 The polar stereographic projection VIII-18.1.2 The EMEP grid VIII-28.1.3 The area of an EMEP grid cell VIII-5

8.2 PERCENTILES AND PROTECTION ISOLINES VIII-68.2.1 Cumulative distribution function VIII-68.2.2 Quantiles and percentiles VIII-78.2.3 Percentile functions and protection isolines VIII-10

8.3 THE AVERAGE ACCUMULATED EXCEEDANCE (AAE) VIII-12

8.4 CRITICAL LOAD EXCEEDANCE AND GAP CLOSURE METHODS VIII-12

REFERENCES VIII-15

CONTENT


Cha

pter

IC

hapt

er II

Cha

pter

III

Cha

pter

IVC

hapt

er V

Cha

pter

VI

Cha

pter

VII

Cha

pter

VIII

��

��

��

��

The critical loads and levels concept is aneffect-based approach that has been usedfor defining emission reductions aimed pro-tecting ecosystems and other receptors.Sustainability indicators are defined for spe-cific combinations of pollutants, effects, andreceptors (see definitions of critical levels inchapter 3, of critical loads in chapter 5.1, andof indicators and criteria used in dynamicmodels in chapter 6). Critical loads and lev-els provide a sustainable reference pointagainst which pollution levels can be com-pared. They can further be used for calculat-ing emission ceilings for individual countrieswith respect to acceptable air pollutionlevels (e.g., defined reductions of criticalload/level exceedances).

The development of critical loads/levels andtheir application in an emission reductionpolicy framework can be seen as an environ-mental design process, using the samemodels and methods as causal research butin a reverse sequence (Figure 1).

During the 1970s it was recognised thattransboundary air pollution has ecologicaland economic consequences (e.g. for the

forest and fish industries) caused by acidify-ing air pollutants. In response to this, thecountries of the UN Economic Commissionfor Europe (UNECE) developed a legal,organisational and scientific framework todeal with this problem. The UNECEConvention on Long-range TransboundaryAir Pollution (LRTAP)) was the first interna-tional legally binding instrument to deal withproblems of air pollution on a broad regionalbasis (see www.unece.org/env/lrtap). Signedin 1979, it entered into force in 1983.

The LRTAP Convention requires that itsParties cooperate in research into the effectsof sulphur compounds and other major airpollutants on the environment, includingagriculture, forestry, natural vegetation,aquatic ecosystems and materials (Article7(d) of the Convention). The Convention alsocalls for the exchange of information on thephysico-chemical and biological data relating to the effects of LRTAP and theextent of damage which these data indicatecan be attributed to LRTAP (Article 8(f) of theConvention). To this end the Executive Bodyfor the Convention established a WorkingGroup on Effects (WGE) that is supported bya number of International CooperativeProgrammes (ICPs).

1 Introduction

1.1 The critical load and level con-cept in the UNECE Conventionon Long-range TransboundaryAir Pollution

Mapping Manual 2004 • Chapter I Introduction Page I - 1

Figure 1: Environmental research vs. environmental design (adapted from Harald Sverdrup)

1 Introduction

Mapping Manual 2004 • Chapter I IntroductionPage I - 2

In 1986 a work programme under the NordicCouncil of Ministers (Nilsson 1986) agreedscientific definitions of critical loads for sul-phur and nitrogen. This provided the neces-sary stimulus to work under the Conventionand in March 1988 two Convention work-shops were held to further evaluate the criti-cal levels and loads concept and to provideup-to-date figures. The Bad Harzburg(Germany) workshop dealt with critical levelsfor direct effects of air pollutants on forests,crops, materials and natural vegetation, andthe Skokloster (Sweden) workshop (Nilssonand Grennfelt 1988) on critical loads for sul-phur and nitrogen compounds.Furthermore, at the Bad Harzburg workshopthe first discussions took place on the possi-ble use of critical level/loads maps for defin-ing areas at risk. It was foreseen that these

could play an important role in the develop-ment of policy.

As a result of these workshops, in 1988 theExecutive Body for the Convention approvedthe establishment of a programme for map-ping critical loads and levels (Task Force on

Mapping) under the Working Group forEffects (WGE) with Germany as the leadcountry (www.icpmapping.org). In 1989 theExecutive Body welcomed the offer of theNetherlands to host a Coordination Centerfor Effects (CCE) that was established at theRIVM in Bilthoven, The Netherlands(www.rivm.nl/cce).

The mandates of the Task Force of theInternational Cooperative Programme onModelling and Mapping of Critical Loads andLevels and their Air Pollution Effects, Risksand Trends (ICP M&M)1 , the CCE and theNational Focal Centres2 are describedbelow.

The structure of the Programme within theConvention is shown in Figure 2.

Figure 2: LRTAP Organogram

1 established by the Executive Body in 1999 to replace the Task Forceon Mapping, see chapter. 1.3

2 in 2003, 24 National Focal Centres are actively participating in the ICPM&M

The Convention has been extended by eightprotocols, which identify specific obligationsor measures to be taken by Parties. The firstsubstance-specific protocols were negotiat-ed on the basis of economic and technolog-ical information (e.g. best available technolo-gies). They set the same emission targets forall Parties, either in terms of a percentagereduction, or as a decrease to an emissionlevel of a former year. They took no directaccount of the effects of the emissions.

A second generation of protocols came intobeing when, in June 1994, a second protocolfor reducing sulphur emissions (the 'Osloprotocol') was signed by 30 countries. Thisidentified effects-based, cost-effectiveabatement measures based on the analysis

of impacts using critical loads. The long-term objective for negotiating national emis-sion reductions was to eliminate the excesssulphur deposition over critical loads for sul-phur, i.e. avoid future exceedances.However, cutting sulphur dioxide emissionsto achieve deposition levels below criticalloads was not feasible for all ecosystems inEurope. Even so, the negotiations werebased on the assessment of environmentaleffects and the protection of ecosystems aswell as technical and economic considera-tions.

The role of critical levels/loads maps for thedevelopment and implementation of air pol-lution control strategies is shown in Figure 3.

1 Introduction


Figure 3: Critical Loads and Abatement strategies

Summarising this figure, the following "crucial steps" are involved:

• Define methods and criteria to determine and map critical loads and levels(Convention workshops);

• Obtain international approval (Working Group on Effects and Executive Body);• Perform a mapping exercise (based on this Manual and on the proceedings of critical

levels/loads and mapping workshops);• Define excess deposition/concentration per unit area;• Use the results for developing strategies and negotiating agreements.

1 Introduction


Figure 4: describes in more detail and on anational level how spatial information onreceptors, pollutants and critical levels/loadsare combined in the mapping process.

In practice, maps of critical loads have beenused as yardsticks to assess the need forreducing depositions in each grid cell of theCo-operative Programme for Monitoring andEvaluation of the Long-Range Transmissionof Air Pollutants in Europe (EMEP). An emis-sion scenario may be viewed by comparinga computed European deposition map withthe European critical loads map. In supportof the Oslo protocol the negotiators startedto consider the use of computer models toassess the costs and effectiveness of emis-sion reduction scenarios. One of these "inte-grated assessment models" is the RegionalAcidification INformation and Simulation(RAINS) model. Figure 5 illustrates the mod-ules of the RAINS model. The first moduleaddresses energy use, agricultural and otherproductive activities, while related emissionsand control costs are reflected in the follow-ing two modules. The fourth module handlesatmospheric dispersion, while the last mod-ule addresses environmental effects. TheRAINS model can be operated in two distinct

modes. In the scenario analysis mode, themodel is run forward: it can predict theregional pattern of concentrations/deposi-tions that will result from the particular com-binations of economic activities, along withthe costs and environmental benefits ofalternative emission control strategies. Inoptimisation mode, RAINS can determinethe least expensive way to arrive at a prede-termined deposition level.

The model user can specify environmentaltargets in various ways: a certain percentagereduction of excess over critical loads, adeposition pattern of a past year, etc. Thismode of RAINS has been used extensively inpolitical negotiations in Europe. Please notethat this is a specification and extension ofthe general concept as shown in Figure 1.

Figure 4: Production of critical levels/loads exceedance maps (adapted from GAUGER et al. 2002)

Since the Oslo Protocol negotiations, thecomplexity of the work under the ICP M&Mhas increased.

First, a more complex formulation of criticalloads was developed and used in support ofthe 1999 Protocol to Abate Acidification,Eutrophication and Ground-level Ozone (theGothenburg Protocol). It recognizes that:

a) sulphur as well as oxidized and reduced nitrogen contribute to acidifi-cation. Therefore, two critical loads for acidity had to be distinguished, the critical load of sulphur-based acidity and the critical load of nitrogen-based acidity (see chapter 5.1 - 5.4).

b) both oxidized nitrogen and volatile organic compounds contribute to the formation of tropospheric ozone, for which critical levels are identified for forests, crops and natural vegetation (see chapter 3.2.4).

c) a small deposition rate of nitrogen which can be taken up by vegetation or immobilized is essential for ecosys-tems (see chapter 5.3).

d) deposition of both oxidized and reduced nitrogen exceeding the critical

load for nutrient nitrogen contribute to eutrophication (see chapter 5.1 to 5.3).

Second, there have been major activities todevelop an effects-based approach also forheavy metals for the preparation of thereview and possible revision of the 1998Århus Protocol on Heavy Metals. Critical lim-its, transfer functions and methods to deter-mine and apply critical loads of heavy metalsare being developed and are listed in chap-ter 5.5.

The aims and objectives of the ICP onModelling and Mapping were approved bythe WGE at its nineteenth session in 2000(Annex VII of document EB.AIR/WG.1/2000/4):

"To provide the Working Group on Effects and the Executive Body for the Convention and its subsidiary bodies with comprehensive information on critical loads and levels and their exceedances for selected pollutants, on the develop-ment and application of other methods for

1 Introduction

1.2 Aims of the ICP on Modellingand Mapping


Figure 5: Scheme of the RAINS model (adapted from Amann et al.)

1 Introduction


effect-based approaches, and on the modelling and mapping of the present status and trends in impacts of air pollu-tion."

Short-term and specific aims are agreedannually at sessions of the Working Groupand approved by the Executive Body. Thereader is invited to view the documentsrelating to long- and medium-term strate-gies, and relating to the work-plan of theExecutive Body, through the web pages ofthe Convention (www.unece.org/env/wge/documents and www.unece.org/env/ebrespectively).

A network of National Focal Centres (NFCs)under the ICP M&M is responsible for thegeneration of national data sets. NFCscooperate with the Coordination Center forEffects to develop modelling methodologiesand European databases for critical loads.CCE reports on this work to the Task Forceof the ICP M&M.

The Programme´s organization and divisionof tasks between its subsidiary bodies, asapproved by the WGE (EB.AIR/WG.1/2000/4)are as follows:

"The International Cooperative Programmeon Modelling and Mapping was establishedin 1999 (ECE/EB.AIR/68, para. 52 (f)) to fur-ther develop and expand activities so farcarried out by the Task Force on Mapping ofCritical Levels and Loads and theirExceedances (led by Germany) and by theCoordination Center for Effects (at theNational Institute of Public Health and theEnvironment at Bilthoven, Netherlands),pursuant to their original mandates(EB.AIR/WG.1/18), amended to reflect thepresent structure of the Executive Body andthe new requirements:

1.3.1 Mandate for the Task Force of the ICP on Modelling and Mapping

(a) The Programme Task Force supports the Working Group on Effects, the

Working Group on Strategies and Review and other subsidiary bodies under the Convention by modelling, mapping, reviewing and assessing the critical loads and levels and their exceedances and by making recommendations on the further development of effect-based approaches, and on future modelling and mapping requirements;

(b)The Task Force plans, coordinates and evaluates the Programme's activities, it is responsible for updating the Programme Manual, as well as for quality assurance;

(c) The Task Force prepares regular reports, presenting, and, where appro-priate, interpreting Programme data.

1.3.2 Mandate for the Coordination Center for Effects

(a) The Coordination Center for Effects (CCE) assists the Task Force of the ICP on Modelling and Mapping, and gives scientific and technical support, in col-laboration with the Programme Centres under the Convention, to the Working Group on Effects and, as required, to the Working Group on Strategies and Review, as well as to other relevant subsidiary bodies under the Convention, in their work related to the effects of air pollution, including the practical development of methods and models for calculating critical loads and levels and the application of other effect-based approaches;

(b) In support of the critical loads/levels mapping and modelling exercise, CCE:

(i) Provides guidance and documenta-tion on the methodologies and data used in developing critical loads and critical levels of relevant pollu-tants, and their exceedances;

(ii) Collects and assesses national and European data used in the model-ling and mapping of critical loads and levels of relevant pollutants. The Center circulates draft maps and modelling methodologies for

1.3 Division of Tasks within theProgramme

review and comment by National Focal Centres, and updates model-ling methodologies and maps as appropriate;

(iii)Produces reports and maps on crit-ical loads/levels documenting map-ping and modelling methodologies,with the assistance of the National Focal Centres and in cooperation with the Task Force on ICP on Modelling and Mapping;

(iv)Provides, upon request, the Working Group on Effects and the Task Force on ICP on Modelling and Mapping, the Working Group on Strategies and Review and the Task Force on Integrated Assessment Modelling, with scientific advice regarding the use and interpretation of data and modelling methodolo-gies for critical loads and levels;

(v) Maintains and updates relevant databases and methodologies, and serves as a clearing house for data collection and exchange regarding critical loads and levels among Parties to the Convention, in consul-tation with the International Cooperative Programmes and EMEP;

(vi)Conducts periodic training sessions and workshops to assist National Focal Centres in their work, and to review activities and develop and refine methodologies used in con-junction with the critical load andcritical level mapping exercise;

(c) While the Coordination Center for Effects reports to the Working Group on Effects and the Task Force of ICP on Modelling and Mapping, and receives guidance and instruction from them concerning tasks, priorities and timeta-bles, it also assists the Working Groupon Strategies and Review, the Task Force on Integrated Assessment Modelling, and other bodies under the Convention, when appropriate.

1.3.3 Responsibilities of the National Focal Centres

The tasks of the National Focal Centres havebeen defined previously in the precedingversion of the Mapping Manual:

The National Focal Centres are responsiblefor:

(a) the collection and archiving of data needed to obtain maps in accordance with the Manual guidelines and in col-laboration with the Coordination Center for Effects,

(b) the communication of national map-ping procedures (data, formats, mod-els, maps) to the Coordination Center for Effects,

(c) the provision of written reports on the methods and models used to obtain national maps,

(d)organising training facilities for nation-al experts in collaboration with theCoordination Center for Effects

(e) making the necessary provisions to obtain national maps in accordance with the resolution and standards (measurement units, periodicity, etc.) described in the Manual,

(f) collaborating with the Coordination Center for Effects to permit assess-ment of the methods applied in order to perform multinational mapping exer-cises (e.g. using GIS) and model com-parisons,

(g) updating the Mapping Manual as appropriate, in collaboration with the Task Force on Mapping and the Coordination Center for Effects.

The principal objectives of this Manual are todescribe the recommended methods to beused by the Parties to the Convention, repre-sented by their National Focal Centres, to:

(a) Model and map critical levels and loads in the ECE region;

1 Introduction

1.4. Objectives of the Manual


1 Introduction


(b)Model and map areas with air pollution values exceeding critical levels or loads;

(c) Develop, harmonize and apply meth-ods and procedures (including dynam-ic modelling) to assess recovery and risk of future damage;

(d)Determine and identify sensitive recep-tors and locations.

Thus it provides a scientific basis for theapplication of critical levels and loads, theirinterrelationships, and the consequences forabatement strategies, e.g. for the assess-ment of optimized allocation of emissionreductions.

This Manual includes methodologies usedby ICP Materials (chapter 4) and (concerningozone) ICP Vegetation (chapter 3.2.4). Incontrast to manuals (or comparable method-ological documents) of other ICPs andEMEP CCC this manual does not containinformation on methods of measurementand specific details on data generation. Thisreflects the aims and tasks of the ICPModelling and Mapping within theConvention.

Specific technical information as well asdetailed results and other information byNational Focal Centres can be found in thebiannual Status Reports of the CCE (see alsochapter 1.6).

Chapter 2 describes methods to map pollu-tant concentrations and depositions. Thesemay be used to generate exceedance mapsby subtracting critical levels/loads withthem. At the European scale, EMEP modelresults are used to construct such maps.The modelled pollutant concentrations anddepositions are derived from national emis-sions which provide the link to negotiationson emission controls. In addition, NFCs areencouraged to produce high resolutionmaps which can be used for effects assess-ments in specific ecosystems at the nationaland local level. This chapter was producedby experts including those from EMEP.

Chapter 3 describes the methods developedfor the quantification and mapping of criticallevels / fluxes of gaseous pollutants for veg-etation. It is largely based on conclusionsand recommendations of Convention work-shops and, for ozone, on intensive workcoordinated by ICP Vegetation in coopera-tion with EMEP.

Chapter 4 describes derivation and applica-tion of acceptable levels for effects on mate-rials. It constitutes an informal Manual of theICP on Materials (www.corr-institute.se/ICP-Materials).

Chapter 5 describes how to quantify criticalloads of nutrient nitrogen, potential acidityand heavy metals. Since this is still the cen-tral task of ICP Modelling and Mapping, thisis the most detailed chapter. The structurehas been changed from the 1996 Manual inthat the main distinction is by method(empirical vs. modelled), and not by effect(eutrophication vs. acidity). The chapterstarts with an overview including definitions(5.1), followed by a subchapter on empiricalcritical loads (5.2) with sections on nutrient N(results of a workshop in Berne in 2002) andacidity. Chapter 5.3 describes methods tomodel critical loads for terrestrial ecosys-tems (SMB model), again divided into subchapters on nutrient N (eutrophication)and acidity. Compared to the previousManual, it has been updated taking intoaccount a.o. the results of workshops inCopenhagen (1999) and York (2000), as wellas various CCE workshops. Chapter 5.4deals with critical loads for surface waters(developed in close cooperation with ICPWaters). Finally, chapter 5.5 describes meth-ods to model and map critical loads of heavymetals5.

Chapter 6 describes dynamic models andthe use of their results. The authors devel-oped it in cooperation with the Joint ExpertGroup on dynamic modelling.

Chapter 7 describes how to identify criticalload exceedance and parameters derivedfrom exceedance (protection isolines, [aver-age] accumulated exceedances).

Chapter 8 describes procedures needed toproduce maps, including map geometry /projections, spatial generalisation and repre-

1.5. Structure and scope of theManual

5 to be finalised by 2004 with assistance from the ad hoc WG on criticalloads of heavy metals

sentativity, and the estimation of uncertaintyand bias.

Annex I lists definitions and descriptions ofthe most important parameters, processes,indicators and criteria.

Annex II describes land use data andecosystem classification systems used with-in the Modelling and Mapping programme.

Annex III gives an overview on unit conver-sions.

In addition, the ICP M&M website(www.icpmapping.org) lists "related docu-ments" (e.g. on deposition methods, onempirical critical loads for nutrient nitrogen)describing certain methodological aspectsin more detail.

For historical details on the establishment ofthe Task Force on Mapping and the man-dates of the cooperating partners in themapping exercise see EB AIR/R.18/Annex IV,Section 3.6 and EB AIR/WG.1/R.18/Annex I.

The historical development of the pro-gramme and the approaches used for calcu-lating critical loads and levels can be fol-lowed by consulting the following back-ground material:

(a) Report of the Initial ECE Mapping Workshop, Bad Harzburg 1989, Draft Manual on Methodologies and Criteria for Mapping Critical Loads/Levels 1990 with scientific Annexes I to IV, both available at the Federal Environmental Agency, Berlin, Germany

(b)Mapping Vademecum 1992, available at the Coordination Center for Effects, Bilthoven, The Netherlands, RIVM Report No. 259101002.

(c) Manual on Methodologies and Criteria for Mapping Critical Loads/Levels (First Edition); Texte Umweltbundesamt 25/93, Federal Environmental Agency (UBA)(ed.), Berlin, Germany

(d)Manual on Methodologies and Criteria for Mapping Critical Loads/Levels and Geographical Areas Where They Are Exceeded (fully revised in 1995/1996); Texte Umweltbundesamt 71/96, Federal Environmental Agency (UBA)(ed.), Berlin, Germany

(e) various interim revisions to (d), e.g. for the mapping of critical levels for ozone

(f) numerous scientific articles referenced in the following chapters.

Status, results and agenda of the ICPModelling and Mapping are described in var-ious documents to be found on theConvention's web site (www.unece.org/env/wge/documents). Various aspects concern-ing technical and scientific background anddetailed results also of National FocalCentres can be found in CCE publications,

especially the biannual CCE Status Reports(www.rivm.nl/cce).

Amann et al. IIASA, The RAINS model,www.iiasa.ac.at/rains

Gauger T, Anshelm F, Schuster H, ErismanJW, Vermeulen AT, Draaijers GPJ, BleekerA, Nagel HD (2002) Mapping of ecosys-tem specific long-term trends in deposi-tion loads and concentrations of air pollu-tants in Germany and their comparisonwith Critical Loads and Critical Levels,Final Report 299 42 210Umweltbundesamt Berlin

Nilsson J (ed) (1986) Critical Loads ofNitrogen and Sulphur. EnviromentalReport 1986:11, Nordic Council ofMinisters, Copenhagen, 232 pp.

Nilsson J, Grennfelt P (eds.) (1988) Criticalloads for sulphur and nitrogen. Reportfrom a workshop held at Skokloster,Sweden 19-24 March 1988. Miljorapport15, 1-418.

1 Introduction


References

The purpose of this chapter is to provideguidance to the participating countries in thegeneration of concentration level and depo-sition load maps for a range of pollutants forcomparison with critical level/load maps.The document is intended as a general refer-ence, with links into the recent literature.While specific recommendations are provid-ed, the procedures are described in outlineand the reader is referred to specialist publi-cations for the measurenent and modellingapproaches described here.

Total deposition is the sum of dry (turbulentflux of gases and particles to the surface),wet (via rain, snow or hail) and fog and cloudwater deposition. All three pathways shouldbe accounted for, but these deposition path-ways differ so fundamentally that it is pro-posed to determine them separately andcombine the quantifications to total deposi-tion estimates (Hicks et al. 1993).

There are two main objectives for Europe-wide mapping of concentrations and deposi-tions:

The first aim is to construct exceedancemaps relative to critical levels and loads,which are then allocated to emissions in dif-ferent countries. The thus obtained transfercoefficients between emission in allEuropean countries and exceedances ineach grid cell of the Co-operative Pro-gramme for Monitoring and Evaluation of theLong-Range Transmission of Air Pollutantsin Euerope (EMEP) is particularly well suitedto provide scientific results for a) implemen-tation of, and compliance with, existingLRTAP Convention protocols and b) theirreview and extension. Such data are crucialfor Integrated Assessment Modelling, andconcentration and deposition maps fromEMEP model calculations are designed forthis purpose. Considering all the uncertain-ties inherent in the discussions of futureemission-deposition relationships, the coun-tries´ economies and energy demands, andour knowledge of environmental effects,maps provided at a scale of 150 x 150 km²have proved adequate for the developmentof international protocols. The new EMEP

Eulerian model provides higher resolutionconcentration and deposition fields at 50 x 50 km².

The second aim is to map concentrationsand depositions which can be used foreffects assessments in specific ecosystems.Such data are needed with a much betterspatial resolution than required for theIntegrated Assessment Modelling. NationalFocal Centers should aim at a sufficient spa-tial resolution for the assessment process,making use of national models and measure-ment networks. EMEP will continue to pro-vide background, long-range transported aircomponents which can be used as boundaryconditions for such national models. Thechapter provides a range of different tech-niques for the provision of maps of concen-tration and deposition, depending on theresources and ambition of the country, withthe EMEP values regarded as default dataallowing the assessment process to be com-pleted everywhere.

Within the countries of Europe, the expertiseand facilities for measurement of concentra-tions and fluxes of pollutants is very variable.The extent to which the methods presentedcan be applied is therefore variable and it isnecessary to show the range of optionsavailable. It is important to stress thatinvolvement in the measurement and model-ling activities is highly desirable as a part ofthe cooperation in the assessment process.The development of satisfactory strategiesfor control of pollutant emissions requiresfull participation in the underlying science aswell as the political process.

The spatial scale aimed at within theMapping Programme differs from the site-oriented approach in the ECE InternationalCooperative Programmes (ICPs) onIntegrated Monitoring and on Forests (LevelII): what is needed is not data for isolatedsites but ecosystem-specific regional esti-mates for all of Europe. Therefore, the focusof this chapter is on methods that are capa-ble of producing critical level/loadexceedance maps for whole countries, usinglong-range transport models in combinationwith concentration measurements, small-scale dry deposition modelling and interpo-lated wet deposition measurement data from

2 Guidance on Mapping Concentration Levels and Deposition Loads

2.1 General remarks and objectives

Mapping Manual 2004 • Chapter II Guidance on Mapping Concentration Levels and Deposition Loads Page II - 1


Mapping Manual 2004 • Chapter II Guidance on Mapping Concentration Levels and Deposition LoadsPage II - 2

(ICP, EMEP, national, regional) monitoring sites. Site-level measurements of dry depo-sition and canopy throughfall for exampleare intended for use for deriving processparametrisations (mainly micrometeorologi-cal measurements) and for independentmodel validation (mainly throughfall meas-urements; see Chapter 2.3.1, 2.3.2 and2.3.10).

Consequently, this chapter is much lessdetailed in its description of field and labora-tory methods than the respective sections inthese ICPs´ Manuals and in the EMEP publi-cations on monitoring methodology (EDC1993 (ICP Integrated Monitoring); UNECEICP Forests 1999; EMEP/CCC 1996).

A range of publications is available withdetailed descriptions of the underlying theo-ry and methods as well as modelling. Theseinclude proceedings from several ECEWorkshops dealing with this subject, mostnotably the 1992 “Workshop on Deposition”in Göteborg, Sweden (Lövblad et al. 1993),the 1993 “Workshop on the Accuracy ofMeasurements” with WMO sponsored ses-sions on “Determining theRepresentativeness of MeasuredParameters in a Given Grid Square asCompared to Model Calculations” in Passau,Germany (Berg and Schaug 1994), and inErisman and Draaijers (1995), Sutton et al.(1998), Slanina (1996), Fowler et al. (1995a,2001a) and ICP Forests Manual (UNECE1999). Supplementary information can befound in other workshop proceedings and inscientific journals.

NFCs are strongly advised to ensure that themonitoring and modelling methodologiesdescribed in the publications listed aboveare documented in the development or vali-dation of a database for national concentra-tion and deposition (and critical level andload exceedance) maps. Compatibility ofthese maps with other national maps withinthe Mapping Programme as well as withmonitoring methods employed within otherdeposition monitoring programmes underthe Convention (ICPs on Forests and onIntegrated Monitoring; EMEP) is very impor-tant.

2.1.1 Mapping resolution and applica-tion of Critical Loads

The use of deposition data with CriticalLoads data very often involves differentscales of the different datasources. In mostcases the Critical Loads data are provided ata finer resolution than the deposition data.To avoid misleading the reader it is essentialthat the different scales be noted in the leg-end. However, it is important to note that theapplication of deposition data at a coarsescale relative to the high resolution CritialLoad data usually gives exceedances whichare systematically underestimated (see2.3.2).

The following items are to be mapped foreach country:

For critical level exceedance maps:-ozone concentration (AOT40 values) and ozone flux,

-sulfur dioxide concentration,

-nitrogen dioxide concentration,

-ammonia concentration.

As input into deposition and critical loadcomputations:

-precipitation amount and other meteo-rological parameters (as required),

-wet, dry, cloudwater/fog and aerosol deposition.

For critical load exceedance maps:-oxidized sulfur (SOx) deposition (total and non-sea-salt),

-oxidized nitrogen (NOy) deposition,

- reduced nitrogen (NHx) deposition,

-base cation and chloride deposition (total and non-sea-salt),

- total nitrogen deposition,

- total potential acid deposition.

2.2 Mapped items

Heavy Metal deposition (pending agreementon critical loads):

-aerosol and wet deposition of lead (Pb), cadmium (Cd), zinc (Zn), and copper (Cu) (suggested as a minimum),

- total deposition of mercury (Hg).

For all maps, the most recent available datashould be used, not going further back intime than five years.

The mapping items concerning gaseous pol-lutant levels to be mapped are listed in detailin Chapter 3 of this manual. The concentra-tions and averaging periods are based uponthe findings of the ECE workshops on criticallevels held at Bad Harzburg (Germany) in1988, in Egham (U.K.) in 1992, Berne(Switzerland) in 1993, St. Gallen(Switzerland) in 1995, Kuopio (Finland) in1996, Harrogate (UK) in 2002, and Göteborg(Sweden) in 2002.

2.3.1 Linkage to emission inventories

Several methods are available to estimateboundary layer atmospheric concentrationsand wet, dry, and cloud-water/fog deposi-tion on different scales of time and space.

Only some of these methods are linked toemission inventories (see Table 2.2 / PageII-20): those that are based on emissioninventories (group A: EMEP and nationallong-range transport modelling, also com-bined with small-scale dry deposition mod-els) can be distinguished from those that areindependent from emission inventories(group B: EMEP and national monitoring ofair concentration and wet deposition, site-level micrometeorological and throughfallmeasurements).

The objectives of methods in group A (linkedto emissions) are (1) regional present andpast situation analysis, and (2) providing abasis for scenario analysis and thereforeemission reduction negotiations. The objec-

tives of the measurement activities in groupB (not linked to emissions) are (1) modelevaluation, and (2) site-specific effectsanalysis (see Chapter 2.1 and Table 2.2).

2.3.2 Quantification and mappingmethods: Scales of time and space

For the time scale, annual deposition ratesare sufficient in order to determine criticalload exceedances, whereas for critical levelexceedances, short-term information issometimes needed (see Chapters 3 and 4).However, there can be substantial variabilityfrom year to year with deposition, for exam-ple with changes in rainfall amount, and it isrecommended that a 3 year average deposi-tion is an appropriate time average for calcu-lation of critical load exceedances.

Three groups of methods exist for variousspatial scales:

Long-range transport models - the most widely used source of deposition data, providing inputs at a range of spatial scales (50 x 50 km² for EMEP, 5 x 5 km² for country scale models)

Nested high-resolution models - are used to provide higher spatial resolution(1 x 1 km²)

Site or catchment specific measure-ments - providing local deposition esti-

mates (1 to 1000 ha); methods include:

Wet deposition collectors (point or field scale)

Micrometeorology (field scale)

Throughfall methods (canopy scale)

Catchment mass balance (landscape scale).

Long-range transport (LRT) modelsprovide average estimates of concentrationsand deposition rates for large grid squares(typically 5 x 5 km² to the EMEP Eulerianmodel´s 50 x 50 km²). They belong to groupA as defined above: they are based on emis-sion inventories and are therefore most suit-able for scenario analyses and country tocountry budgets (‘blame matrices’) used inemission reduction negotiations (if the



2.3 Methods of mapping, their under-lying assumptions and data requirements



model domain is more than one country).Long-range transport model results areavailable for the European UNECE regionfrom EMEP and can be taken as default orreference model outputs.

Standard multiannual concentrations fromEMEP are given as one number per year percomponent per 50 x 50 km² grid squareand deposition fluxes are provided either asaverage deposition to the 50 x 50 km² gridsquare or as ecosystem specific depositionestimates. EMEP model output can be pro-vided for shorter time periods, but with theoverall constraint that one of the majorinputs, the emission inventory, is often pro-vided only as an annual total.

The scale at which critical levels/loads andconcentration/deposition are mapped great-ly influences the magnitude of exceedancevalues (Spranger et al. 2001, Bak 2001,Lövblad 1996, Smith et al. 1995). For exam-ple, if the average value from a 50 x 50 km²grid square is matched to critical loads onthe 250 squares of 1 x 1 km² within the 50 x 50 km² grid square, there will be generally be less critical load exceedancethan if the deposition were available at the 1 x 1 km² scale. The only circumstance inwhich this underestimate would not occurwould be if the high deposition locationsmatched the high critical load locations.Over many areas of Europe, exactly theopposite occurs. In many areas of complexterrain the parts of the landscape receivingthe largest deposition, such as the higherareas in the mountains of North-WestEurope, are also the most sensitive to theeffects of deposition, for example acidifica-tion. The same holds true for forested areas,which tend to correlate with poor soils inlarge parts of Europe. This problem is worsefor components with low local sources (NH3,NOx) because the within-grid distribution ofsources is not reflected in the grid averageestimation from a LRT model, but doesmarkedly increase the within-grid variabilityof deposition and hence increases the criti-cal load exceedances. As the current depo-sition estimates from EMEP are provided ata scale which is much larger than the scaleof this spatial variability, the critical loads

exceedances for these areas are underesti-mated.

These effects are minimised by estimatingdeposition to the smallest spatial scale pos-sible. However, there is an underlying rela-tionship that the critical load exceedanceswill increase as the spatial resolution of thedeposition gets closer to that of the criticalloads.

High resolution modelling.A second group of methods tries to over-come these scale problems by applyingsmaller-scale “inferential” models usinglarge-scale meteorology and concentrationfields either obtained from LRT models (seeabove) or by interpolation of sufficientlydense measurement networks (see below).

Dry deposition is inferred by multiplying theconcentration with the deposition velocity ofthe component of interest (Hicks et al. 1987,1993). The latter is calculated using a resist-ance model in which the transport to andabsorption or uptake of the component bythe surface is described. Resistances aremodelled using observations of meteorolog-ical parameters and parametrisation of surface exchange processes for differentreceptor surfaces and pollution climates asdescribed in Erisman et al. (1994a), Smith et al. (2000), Nemitz et al. (2001), Embersonet al. (2000), Grünhage and Haenel (1997),Gauger et al. (2003). The deposition veloci-ties of cloudwater/fog droplets can be simi-larly estimated by modelling momentumtransfer (Fowler et al. 1993) and a similartechnique has been used to estimate basecation deposition (Draaijers et al. 1995).Parameters determining the depositionvelocity include atmospheric parameters(e.g. wind speed, temperature, radiation, rel-ative humidity, atmospheric stability, cloudand/or fog frequency) and surface condi-tions (e.g. roughness, wetness, stomatalresponse, soil water). Unfortunately, up tonow no reliable European-widecloudwater/fog concentration fields areavailable, hampering cloudwater/fog deposi-tion estimation on a European scale.

The land use maps used for this depositionmodelling should be identical to the stock-

at-risk maps used for critical levels/loadsmapping (see Chapter 6). In addition to thegeographical position of sensitive ecosys-tems, land use type/vegetation type, vegeta-tion height and crown coverage should bemapped as well on a scale that allows forcorrect allocation of deposition to allecosystem types in the model domain.

Uncertainties of inferential deposition mod-els are described in Chapter 2.3.10.

Site or catchment specific measure-ments. All methods based on pointmeasurements (e.g., wet deposition meas-urements, micrometeorological dry deposi-tion measurements, throughfall measure-ments) belong to group B, as they cannot bedirectly connected to emission inventories.Maps can only be produced directly fromthese measurements if the network is denseenough to account for spatial (and temporal)variations. This may be the case for net-works measuring air concentrations of com-pounds with little spatial variation or formeasurements of wet deposition in areas ofsimple terrain. Network (point) measure-ments should be interpolated using the krig-ing technique and it may be helpful toinclude monitoring data from neighbouringcountries for interpolation. For some air con-centrations, such as ammonia, or for rainconcentrations in complex terrain, therequired density of the measurement net-work could be too dense for practical appli-cation. In these cases, it is recomended thatconcentrations are obtained from less densenetworks and that simple models are used toassist the interpolation, e.g. using altitudedependences. It is preferable to interpolateconcentrations in rain or in air and then cal-culate the deposition at the receptor siteusing local estimates of rainfall and land-usespecific ground-level dry deposition rates(see above).

Additional monitoring of air concentrationsof gases in order to create a dense networkas a basis for mapping can be made with dif-fusive samplers. These samplers, can beused for a number of gases (ozone, sulphurdioxide, nitrogen dioxide, nitrogen oxides(NOx), ammonia, nitric acid, mercury, hydro-gen chloride, etc.) The sampler providesaverage concentrations over a time period,

normally from one week to one month. It is asimple and cheap complementary method,to be used in parallel to other methods pro-viding also the temporal variability (Ferm andSvanberg 1998, Ferm 2001, Sjöberg et al.2001).

Wet deposition. In most cases,the long-term spatial variation in wet deposi-tion within regions is determined mostly byvariations in precipitation amount and lessby variations in concentrations in rain orsnow. In addition, precipitation amounts aremostly available from relatively dense mete-orological networks. Therefore, if concentra-tion variations are small, maps of annual wetdeposition rates should not be drawn byinterpolating measured wet deposition rates;it is recommended to interpolate measuredsolute concentrations and estimate the wetdeposition as the product of the mappedsolute concentration and the precipitationamount, the latter provided by the meteoro-logical service for the country. This is animportant step because the precipitationfields are defined by dense networks of col-lectors. An additional and very importantenhancement of wet deposition occurs inthe uplands of Northern Europe, due to washout of orographic cloud by falling rain orsnow. As networks do not generally measureat high elevation in complex terrain, theseeffects are generally omitted from networkmeasurements. The underlying physicalprocess is well documented and the effectsmay be modelled using the network data(Dore et al. 1992, Fowler et al. 1995b).

Dry and cloudwater/fog deposition canbe estimated from concentration measure-ments of airborne substances by microme-teorological measurements at the processlevel (for SO2: Fowler et al. 2001c; for NH3:Flechard and Fowler 1998; for cloud:Beswick et al. 1991). During the last decadeit has become possible to make micromete-orologically based long-term flux measure-ments (i.e. continuous flux measurementsover more than a year). This has beendemonstrated for O3, NOx and SO2 (LIFE proj-ect, Erisman et al. 1998a) and for CO2 andH2O (Aubinet et al. 2000). Such measure-ments give information about the seasonal





and interannual variability in the fluxes.

These measurements of deposition fluxeshave therefore become straightforward andcan be applied to all pollutant gases.However, the primary purpose of the meas-urements is to provide the parameters formodelling as the tool for extrapolation overthe landscape as the measurement stationsare expensive to operate. Thus the numberof dry deposition stations could not realisti-cally be sufficiently large to interpolate flux-es over spatial scales directly. There are lowcost micrometeorological methods, such asthe Time Averaged Gradient (TAG) system(Fowler et al. 2001b). These will provide themeans of obtaining deposition parametersfor many more representative terrestrial sur-faces in Europe.

The methods mentioned here only work ifstringent prerequisites concerning microme-teorological variables (e.g., surface homo-geneity) are fulfilled. They cannot be directly extrapolated, but the process knowl-edge obtained from such measurements canbe parametrized in inferential models andfluxes can be mapped using this information(see high resolution modelling above).

Throughfall and stemflow measurementscan be used to estimate the site-specifictotal deposition of sulfur to plant canopies,mainly forests (wet plus dry plus cloudwa-ter/fog). The data are useful for paralleleffect studies, in order to estimate deposi-tion rates on the basis of field data availablefrom existing monitoring programmes and tovalidate other deposition estimates. Theywill also provide knowledge on the seasonalvariation and the trends of deposition. Inmany cases throughfall monitoring is con-sidered to be sufficient, and stemflow is onlymeasured for some tree species, for which itis known to be of importance (e.g. beechtrees) (UNECE 1999). In practice, it is notgenerally possible to determine the totaldeposition of substances for which uptakeor leaching within the canopy is large relativeto the deposited amounts (e.g. nitrate,ammonia, and calcium, potassium and magnesium) from throughfall measurements.

Throughfall measurements are cheaper andgenerally easier to perform than micromete-

orological measurements. They also give agood overview of the deposition situation inthe forest, not only for sulfur but also fornitrogen compounds. Recent Swedish expe-riences have highlighted the problems withcomparing throughfall measurements withwet deposition when the dry deposition con-tribution to the total is very low (Westlingpers. comm.), as is now the case for sulphurin many areas of Europe. Large uncertaintiesin wet deposition at wind-exposed siteshave been shown with field intercomparisonstudies (Draaijers et al. 2001). Even if it is notpossible to estimate the total deposition ofnitrogen with this method, a lower limit canbe set. Sampling considerations (e.g. loca-tion of collectors, species composition, spa-tialy variability) are very important for achiev-ing good results and sampling requirementsare described in detail in the ICP ForestsManual (UNECE 1999) and in review articlessuch as Draaijers et al. (1996a) and Erismanet al. (1994b).

In order to interpret the data, the relationbetween total deposition and throughfall canbe expressed:

(2.1)

where:

THF = Flux in throughfall (plus stemflow)

DRY, WET, Cl/Fog = dry, wet, cloudwater/fog

deposition

CEX = canopy exchange; CEX > 0 for leach-ing, CEX < 0 for uptake

When CEX=0, the dry deposition can be esti-mated as the difference between total flux inthroughfall and independent measurementsof wet and cloudwater/fog deposition. If CEXdiffers from 0, dry deposition cannot be dis-tinguished from internal cycling. Thismethod can give large overestimates of thetrue deposition flux (CEX>0), due to canopyleaching (for some base cations), or largeunderestimates of the true deposition flux(CEX<0), due to canopy uptake (e.g., fornitrogen compounds and protons).Therefore, throughfall plus stemflow fluxes

should be interpreted as upper bounds oftotal base cation deposition and as lowerbounds of total nitrogen and proton deposi-tion.

In some cases, the total deposition to plantcanopies can be deduced from throughfalland precipitation measurements in the openfield using empirical canopy budget models.These models are not always applicable andshould if used be applied with care. Themethod by Ulrich (1983), which uses the rela-tion of dry vs. total deposition of sodium asan indicator of dry vs. total deposition ofother elements, is used most widely.However, some of its assumptions, such asthe constant (in time and with respect tosubstances) ratio of dry particulate deposi-tion vs. wet deposition rates, are question-able. For several reasons (e.g., differentdeposition and canopy uptake processes), itcannot be applied for determining total nitro-gen deposition to the forest ecosystem(Bredemeier 1988, Spranger 1992). A modi-fication (Beier et al. 1992) and an extensionof the model (Draaijers and Erisman 1995)mitigate some of the methodological prob-lems, even though they have not yet beenproperly evaluated under all circumstances(see Draaijers et al. 1996a,b), but thereremains an issue with canopy modellingthat, in many cases, a cation surplus existsin the throughfall solutions indicating thatmajor substances have not been measured.

The calibrated watershed method inte-grates deposition fluxes over a scale com-patible to critical load computations forexample for lakes and surface waters.However, major fluxes to the groundwaterand soil exchange have to be accounted for.It is most useful for conservative elements(e.g., S, Na, Cl) in areas with clearly delineat-ed watersheds. The data are useful to vali-date deposition estimates derived frommodelling.

Concluding remarks. LRT Modelswill normally be used to calculate patterns ofconcentration and deposition acrossEurope. High resolution inferential modelsmay be used to calculate patterns withincountries or regions. The role of measure-ments in the process of mapping is twofold.Low cost devices, such as passive samplers

or bulk samplers, may be used to observeregional patterns of concentrations and wetdeposition. If the measurement strategy isgood, these measured patterns will be morereliable than those calculated. However fordry deposition the situation is more com-plex. The deposition rate depends on atmos-pheric conditions as well as on ecosystemtype. Recent developments of low cost drydeposition monitors (Fowler et al. 2001b) willmake it possible to determine regional pat-terns of dry deposition in the near future. Butstill the application in dense networks isexpected to be limited. The limitations ofusing throughfall measurement networks forestimating regional patterns of total deposi-tion have been described above.

The results of measurements on the otherhand are necessary to test, validate andimprove model parameterisations.Defendable estimates of the deposition needto be validated by measurements to relevantEuropean ecosystems. It is recommended toestablish a network across Europe wheredetailed measurements of dry and wet dep-osition will be made. This may be linked toexisting monitoring networks such as theEMEP network and the Level II Programmeof ICP Forests.

2.3.3 Mapping meteorological parame-ters

Meteorological parameters are requiredinputs for most critical levels or critical loadscalculations. The data requirements anddata provision will vary from country tocountry. Data are generally avaialable fromnational weather services. European datacan be obtained from European Centre forMedium-Range Weather Forecasts (ECMWF,website: www.ecmwf.int), who provide mod-elled data based on observations withinEurope, and there are other sources forsome data such as the US EPA/NCAR glob-al precipitation database.

Precipitation amounts are needed for criticalload computations, for wet deposition map-ping (see Chapter 2.3.2), and for surfacewetness parametrisations (also for materials:“time of wetness”; see Chapter 4).





Fog and cloud occurence is needed forcloudwater/fog deposition estimates.

Wind speed, temperature and radiation arebasic requirements for the inferential model-ling of dry deposition. Additionally, relativehumidity, soil water deficit, and atmosphericstability are often required.

The availability of accurate local meteoro-logical data is often a constraint to detailledlocal high resolution modelling, and there-fore the success of models in improvingdeposition estimates to specific ecosystemsmay depend as much on the availability ofquality meteorological data as on the qualityof the local concentration estimates ormeasurements.

2.3.4 Mapping ozone (O3) concentra-

tions and deposition

Ozone concentration data are required togenerate maps showing areas where thecritical level is exceeded (see Chapter 3.2.4).Data may be available from photochem-istry/transport modelling (see (a.) and (b.)below) or from monitoring networks (see (c.)below).

The concentrations of ozone close to terres-trial surfaces (e.g. within 1 m) show a largespatial variability in both rural and urbanareas. For urban areas, this variability ismainly caused by chemical consumption ofozone by NO, which is locally emitted. Forrural areas away from local sources, thisvariability is largely caused by spatial andtemporal changes in the degree to whichindividual sites are vertically ‘connected’ tothe main reservoir of ozone in the boundarylayer. Like in urban areas O3 might be consumed by the reaction with NO whichcan be emitted from bacterial processes inthe soil (PORG 1997)To provide a spatial resolution of the ozoneexposure on a horizontal scale whichreflects the variations in the orography it ishelpful to produce the ozone concentrationfield at a grid of 1 x 1 km² cell-size at least(see (b.) and (c.) below). As the critical levelsare based on the concentration measured inthe turbulent layer near the receptor, ozone

levels modelled or measured at higher distances from the ground are not directlyrelated to the observed effects. Therefore asurface-type specific correction should beapplied for assessing exceedances of critical levels, but it is hardly possible toquantify the correction from monitoring datato dose-effect data at present (Fuhrer 2002).

The supply of ozone to vegetation is provid-ed by atmospheric turbulence and hencewind speed and the thermal structure of airclose to the ground. The deposition ofozone on terrestrial surfaces and vegetationcauses a vertical gradient of the ozone concentration, which is largely determinedby the sink activity of the soil-vegetationsystem. Maps of O3 deposition can be produced from inferential modelling basedon parameters obtained from long-termmeasurements and land-use information(Emberson at al 2000).

(a.) LRT model results

EMEP/MSC-W has available calculations ofozone concentrations in 50 x 50 km² grid-squares with hourly time-resolution and alsodeposition to specified land cover types, e.g.forests, arable crops, etc. The model calcu-lations are available for a notional height of45 m above ground and scaling algorithmsare available to provide output at lowerheights, particularly 3 m and 1 m. AOT40for crops and forests is also available, aswell as their changes in each grid square perunit of changed country-emissions of VOCand NOx. The calculations are based on newEMEP/CORINAIR emissions with 11 sourcesectors and VOC speciation specified foreach sector. The model also includes bio-genic VOC emissions from forests.

(b.) High resolution modelling

The low spatial resolution of LRT modelsdoes not match with the resolution requiredfor the evaluation of ozone exposure offorests ecosystems, and estimates of ozoneexposure can be improved by local scalemodelling within the 50 x 50 km² EMEPsquare. The necessary concentration values

at receptor level can be obtained at high re-solution from the large scale model averagevalues by correcting them for local emissionof nitrogen oxides, orography and deposi-tion. The computation of deposition veloci-ties and deposition fluxes requires land-usemaps (see Annex II), as well as meteorologi-cal data.

One method of adjusting for local scaleeffects is to adjust the diurnal cycle in con-centrations from the LRT by accounting forthe dependence of ozone concentrations onlocal orography (PORG 1997, Coyle et al.2002). The elevation of a particular locationdetermines the extent to which it experi-ences the influence of air from the free tro-posphere and from the boundary layer.Based on data from the EUROTRAC-TORand EMEP monitoring programs as well ason results in literature, this dependence canbe modelled and combined with small scaleorographical data.

High-resolution AOT40 maps can also becomputed by using other atmospheric trans-port and photochemistry models, providedthat the output data are high-resolutionresults modelled over a long (for AOTF: April-September), continuous time period, andthat the model results are evaluated withmeasurements as well as with MSC-Wmodel results.

(c.)Monitoring and interpolating ozone con-centrations and fluxes

The monitoring of ozone is necessary toestablish or to validate exceedance maps aswell as for the verification of the long rangetransport and chemistry models.

Over large parts of Europe and particularly inSouth and East Europe there are very fewavailable data. Efforts should be made toobtain data where monitoring stations exist.Elsewhere, the establishment of a network ofmonitoring stations is strongly recommend-ed. Stations should be linked to the EMEPnetwork.

To provide data from such a network that arerepresentative for an extensive area it is rec-ommended that the monitoring stations besited at rural locations avoiding local

sources of the oxides of nitrogen such asroads. Anshelm and Gauger (2001) devel-oped a method to classify the suitability ofmonitoring sites for mapping concentra-tions.

Some countries have already a sufficiently‘dense’ network. If the number of monitoringstations needs to be increased, it is recom-mended to install them at various altitudesand/or at various distances from the emittersof ozone precursors. Stations at urban loca-tions are not representative for extensiveareas, but they may be required for differen-tiating areas with rural and with urban pollu-tion conditions and for population exposureassessments through mapping exceedancesof air quality guidelines based on humanhealth.

The preferred sampling height is 3-5 m andthe monitoring station requires an openaspect without the presence of trees or othertall vegetation in the proximity of the sampleintake. Appropriate recommendations forsampling and calibration are available fromthe Chemical Coordinating Centre of EMEP.

More or less simple interpolation proceduresexist to obtain an estimate of the exposureof terrestrial surfaces to ozone using topo-graphic and other information. Altitude maybe used here as an indicator for the degreeto which areas are ‘connected’ to the ozonereservoir in the boundary layer (see also b.)).

One possibility is to interpolate the meas-ured hourly ozone values and then to com-pute seasonal dose-parameters such asAOT40 (see chapter 3.2.4) on the basis ofthose hourly maps (Loibl and Smidt 1996).But in general it will be easier to calculatethe values of the required dose-parametersfor each monitoring station first and then touse these values for the spatial applicationof a regression model. For the interpolationof AOT40 values relationships with altitude(height above mean sea level) have beenused in the UK (Fowler et al. 1995c) and inthe Nordic countries (Lövblad et al. 1996).Relationships with relative height (the heightof the site of interest above the valley groundwithin a certain distance) are applied inAlpine regions (Loibl and Smidt 1996). Such





relationships are reasonably consistent onthe regional scale (100 to 500 km) with theexception of coastal and urban areas, butmust be established from relatively densemonitoring networks. Dividing the studyarea into subregions and evaluating region-specific influences on the O3 concentrationor exposure may enhance the quality of theinterpolation. Data assimilation, combiningobserved and LRT modelled concentrations,is another method recently applied to pro-vide improved ozone concentration fields(Flemming 2003).

2.3.5 Mapping sulfur dioxide (SO2) concentrations andoxidised sulfur (SOx) deposition

Data on SO2 gas concentrations, sulfate(SO4²-) aerosol concentrations and SO4²-concentrations in rain are required to gener-ate maps showing areas where the criticallevel is exceeded. Data are available fromlong-range transport modelling, possiblycoupled to small-scale modelling (see (a.)and (b.) below), or from monitoring networks(see (c.) below).SO2, in contrast to ozone or sulfate aerosol,is a primary pollutant. It is emitted by bothhigh (e.g. power plants) and low (e.g. house-holds) sources. Therefore the spatial vari-ability of concentrations tends to be higherthan that of ozone and sulfate aerosol butlower than that of ammonia. Close to urbanareas, the concentrations of rural sulfur dioxide are elevated and this effect shouldbe modelled explicitly where possible, forexample by using urban concentrationmeasurements and areas of urbanisation tomodel the urban effect (a similar method wasused by Stedman et al. (1997) to model NOxand NO2 near roads).

For rural areas away from local sources, spatial variability is largely caused by spatialand temporal changes in the degree towhich individual sites are vertically ‘connect-ed’ to the main reservoir in the boundarylayer (see preceding subchapter on ozone).

As for ozone, the SO2 levels measured 3-5 mabove ground are not directly related to the

observed effects, since dry deposition caus-es a systematic vertical concentration gradi-ent towards the surface, while the criticallevels are based on the concentration meas-ured close to the receptor. However, sur-face-type specific corrections are not gener-ally applied and measured/modelled valuesusually taken uncorrected.Non-sea-salt inputs of sulfur are needed inthe critical loads framework, since criticalloads are generally compared with anthro-pogenic S (and N) (see Chapter 5.3.2).Consequently, the base cation and chloridedeposition in the charge balance - fromwhich critical loads are derived with the SMBmodel - have to be corrected for sea saltcontributions as well. Natural marine emis-sions of reduced sulfur compounds (espe-cially Dimethylsulfate, DMS) are includedinto the EMEP emission data base (andtherefore EMEP model results), whereassea-salt emissions are not.

Depositions of base cations, sulphur andchloride (given in equivalents) are correctedby assuming that either all sodium or allchloride is derived from sea salts, and thatthe relations between ions are the same as insea water (after Lyman and Fleming 1940,cited in Sverdrup 1946):

(2.2)

where X = Ca, Mg, K, Na, Cl or SO4,Y = Na or Cl,rXY = is the ratio of ions X to Y in

seawater and the star denotes the sea-salt correct-ed deposition. Ratios rXY areshown in Table 2.1 with3-decimal accuracy.

Note that for arbitrary ions X, Y and Z therelationships rYX = 1/rXY and rXY·rYZ = rXZhold. If Na (Cl) is chosen to correct for seasalts, Na*

dep=0 (Cl*dep=0).

Using such a correction will only yield reli-able estimates on non-sea-salt S, Mg, Ca, K,and Cl in areas where sea-salt is the onlysource of Na in ambient air. This generallywill be the case in western and northernEurope. In some parts of southern andsouth-eastern Europe, however, significantquantities of Na in the atmosphere originatefrom wind-blown evaporates and applyingthe sea-salt correction there will result inunderestimated non-sea-salt concentra-tions.


As for ozone, EMEP/MSC-W has availablecalculations of sulfur dioxide concentrationsand sulfate aerosol concentrations in 50 x 50 km² grid-squares with hourly time-resolution and also deposition to specifiedland cover types, e.g. forests, arable crops,etc. The model calculations are available fora notional height of 45 m above ground andscaling algorithms are available to provideoutput at lower heights, particularly 3 m and1 m. Daily concentrations and deposition ofsulfate in rain are also available.Concentrations allocated to emissions inseparate countries are available with monthly time-resolution.


Procedures as the ones described for ozonein the preceding subchapter can be appliedfor SO2. However, since SO2 is a primary

pollutant, the most important factor causingvariability is local emission, which has to beaccounted for in the high-resolution model.SO2 deposition velocities depend mostly onstomatal opening (stomatal pathway: to beparametrized using vegetation type/landuse, and meteorology data), on surface wet-ness, and on NH3 concentrations. When sur-faces are wet, and at humidities >90%, sur-face resistances to deposition become verylow and the flux is mostly determined byatmospheric resistances (Erisman et al.1994a).

For SO42- aerosol, dry deposition is highest

for forests or other rough surfaces that arefar from SO2 emission sources (Gallagher etal. 1997). Sulfate deposition velocities canbe estimated using Slinn´s (1982) or a similarsimple particle deposition model (e.g.,Erisman et al. 1995, Ruijgrok et al. 1996) butthere is still significant research effortfocussed on improving these estimates andit is reasonable to use site specific models toimprove on the LRT sulfate aerosol deposi-tion.Wet deposition maps can be produced frommonitoring data according to the methodsdescribed in Chapter 2.3.2, including oro-graphic effects where applicable. The mostimportant factor in improving wet depositionestimation at the local scale is the availabili-ty of rainfall maps derived from dense net-works of rainfall collectors. If the density ofthe concentration monitoring network is notsufficient, EMEP/MSC-W modelled concen-tration data can be combined with local rain-fall maps for improved estimates of wet de-position.



Table 2.1: Ion ratios rXY=[X]/[Y] (in eq/eq) in seawater



c.) Monitoring and interpolating concentra-tions

Measurement stations for SO2 should be setup in areas that are not directly affected bylocal emitters. As for the EMEP network,and contrary to regional, national or EUhealth-related programmes, the main objec-tive of a measurement network is not neces-sarily to determine the highest ambient con-centrations (leading to high local criticallevel exceedances) but rather the large-scaleconcentrations that are due to long-rangetransport. Individual measurement stationsshould be representative for a maximumarea, thus making interpolation and mappingof concentrations unaffected by localsources possible. Criteria for siting meas-urement stations are listed in, e.g.,EMEP/CCC (1996).

Due to the fact that SO2 is a primary pollu-tant, the measurement network density,especially in emitter areas such as CentralEurope, has to be high so that the interpola-tion error (determined, e.g., by variogramanalysis when using kriging) is minimal rela-tive to the measured values. The same istrue for mountainous areas due to the verti-cal gradients present. As secondary pollu-tants with relatively slowly varying concen-trations, the measurement network densityfor SO42- aerosol and for SO4²- in rain can bemuch lower than for SO2. Methods to deter-mine (and data on) representativity of meas-urements, as well as their precision andaccuracy, are listed in Berg and Schaug(1994).Maps can be produced from measurementsby interpolation if all criteria mentionedabove (accuracy/precision, representativity)are fulfilled. For some applications a blend-ing height approach is appropriate, whereground-level observations are extrapolatedto 50 m height (the blending height) abovethe ground, using a resistance model. At thisheight the concentration is less dependenton the surface processes and can be inter-polated over larger areas (Erisman andDraaijers 1995). The preferred interpolationprocedure is kriging, which also provides anestimated interpolation error.

2.3.6 Mapping nitrogen oxides (NOx)concentrations and oxidised nitro-gen (NOy) deposition

Data are available from long-range transportmodelling, possibly coupled to small-scalemodelling (see (a.) and (b.) below), or frommonitoring networks (see (c.) below).

NOx (=NO+NO2), like SO2, is emitted by bothhigh (e.g. power plants) and low (e.g. traffic)sources, mostly as NO. The spatial variabili-ty of NOx concentrations tends to be higherthan that of ozone and nitrate but lower thanthat of, e.g., ammonia, due to conversion ofNO by reaction with O3. In rural areas emis-sion of NO from soils (both agricultural andsemi-natural) can likewise contribute to localNO2 levels. Many national modelling activi-ties are able to provide estimates of surfaceconcentrations of NO2 at a higher resolutionthan 50 x 50 km², for example at 5 x 5 km² in the Netherlands and in the UK(the UK Air Quality data base is atwww.airquality.co.uk/archive/) and these canincorporate models to adjust concentrationsfor local emissions, for example by usingdistance to major roads.


As for SOx, EMEP/MSC-W has available con-centrations and deposition of NOx (NO andNO2), NO3

- in aerosol and in rainfall, andHNO3.


Procedures as the ones described for SOxcan be applied with the following comments.For inferential dry deposition modelling, NO3

-aerosol and HNO3 (ideally also HONO) concentration maps are needed besides theNOx concentration maps. Since measure-ments are too scarce to be interpolated inmost countries, they generally will have to beestimated from atmospheric models. Themost important factor causing variability inNOx is local emission which has to beaccounted for in the high-resolution model.NO2 deposition velocities depend almost

exclusively on stomatal opening (stomatalpathway: to be parametrized using vegeta-tion type/land use, and meteorology data)and the importance of surface wetness toSO2 deposition does not hold for NO2. Forthe aerosol fraction, HNO3 deposition isdetermined by atmospheric resistancesbecause the surface resistance is very low(surface roughness and windspeed are mostimportant) and the NO3

- aerosol depositionvelocities are estimated similarly to SO4²-

aerosol.

As was metioned above, ozone gradientsmay be affected by fast chemical reactionsbetween O3, NO2 and NO. These reactionswill also affect the gradients and hence theuptake of NO2. In LRT models this effect isnot taken into account. Although no firm evidience is available, it is felt that theuptake by low vegetation is only marginallyaffected. (Duyzer et al. 1995). On the otherhand, the effect on uptake of NO2 and O3 byforests could be influenced by chemicalreactions taking place in the canopy (Waltonet al. 1997).

(c.)Monitoring and interpolating concentra-tions

Measurement stations for NOx should be setup in areas that are not directly affected bylocal emitters (most importantly not nearroad traffic). Criteria for siting measurementstations are similar to those for SOx. Due tothe fact that NO is a primary pollutant andthe reaction to NO2 is relatively fast, themeasurement network density has to behigh, as for SO2. The other nitrogen com-pounds are assumed to have slowly varyingconcentrations and the network density canbe reduced accordingly.

2.3.7 Mapping ammonia (NH3) concen-tration, reduced nitrogen (NHx) deposition and total nitrogen deposition

Ammonia is emitted primarily from low levelagricultural sources with varying sourcestrengths. Gaseous NH3 has a short atmos-pheric residence time (Erisman and Draaijers1995) and as a result its concentrations in airmay show steep horizontal and vertical gradients (Asman et al. 1988). Even in areasnot affected by strong local sources, theambient concentration of ammonia may varyby a factor of three to four on scales lessthen a few kilometres.The very localised pattern of ammonia con-centration, and also of ammonia dry deposi-tion, has consequences for mapping proce-dures. Mapping of ammonia concentrationsby interpolation from measurements alone isnot treated explicitly here, as the requiredmeasurement network density would beextremely high and the method is only feasi-ble over small areas. However, the criticallevel of ammonia is so high that except verynear sources (farms) exceedances are notvery likely.

A long-range transport model with, forexample a 50 x 50 km² spatial resolution,will not resolve these large variations eitherfor ammonia concentrations or for the drydeposition of ammonia which will be themajor fraction of total reduced nitrogen deposition close to an ammonia source. Soassessments of the exceedances of criticalloads will be biased when using LRT models.In the absence of very detailed emissiondata (on the level of the individual farm),measurements in a dense network are need-ed to obtain accurate exceedance levels(Asman et al. 1988).

It is also important to note that ammoniamay be emitted as well as deposited ontovegetation, and therefore surface–atmos-phere exchange modelling must be used toquantify the net exchange over the land-scape. The background developments toallow these processes to be simulated use acompensation point approach (Schjorring etal. 1998; Sutton et al. 2000).





(a.) Long-range transport modelling

As for SOx, EMEP/MSC-W has available concentrations and deposition of NH3 andNH4

+ in aerosol and in rainfall. However, theinterpretation of the concentration and drydeposition of ammonia estimates for ammonia must be qualified because of thespatial resolution of the model and theeffects of local sources. Improvements inthe ammonia component of the EMEP/MSC-W LRT model are currently being developed.

(b.) High-resolution modelling

In the LRT model results from EMEP/MSC-W, it is assumed that the concentration dis-tribution of ammonia within a grid cell ishomogeneous, whereas generally sub-gridconcentration variations will be present. Thespatial resolution on which concentrationscan be modelled strongly depends on theresolution of the available emission esti-mates. Where these are available at spatialscales of the order of 1 x 1 km², consider-able improvements to the estimates from theEMEP/MSC-W LRT 50 x 50 km² model arepossible. Models are available for thesemore detailed calculations, such as the OPSmodel for the Netherlands (see below) andthe FRAME model for the UK (Singles et al.1998).

The Operationele Prioritaire Stoffen model(OPS) developed at RIVM is able to calculatedispersion (and deposition) of NHx on a 5 x 5 km² grid over The Netherlands(Asman and Van Jaarsveld 1992; Erisman etal. 1998b). The model is able to describeboth short- and long-distance transport,average concentrations (and depositions)can be computed for time scales from 1 dayto more than 10 years, and it can account forboth point sources of various heights andarea sources of various shapes and heights.The basis for the model on the local scale isformed by the Gaussian plume formulationfor a point source. Computations are madefor a limited number of meteorological situa-tions (classes) with a representative meteor-ology for each class derived from actual

observations. The uncertainty in emissionvalues appeared to be the most importantfactor determining the uncertainty in con-centrations. Model results and air concen-tration measurements in the Netherlands arereasonably well correlated, but substantialdifferences in the absolute values have yetto be explained quantitatively (Duyzer et al.2001). Given the concentrations on a 5 x 5 km² scale, deposition of NHx in theNetherlands was estimated and showedgood agreement with results of throughfall(corrected for canopy exchange) andmicrometeorological measurements(Erisman et al. 1995).

For aerosol deposition and wet deposition,the procedures as described for SOx can beapplied to NH4

+.

(c.)Monitoring and interpolating concentra-tions

Accurate representative measurement ofNH3 concentrations, especially in high emission density areas, requires manymeasuring sites. Typically, most of the concentration gradient is present within afew km of the source and local scale moni-toring is a valuable tool to understand theprocesses. With the developments in passive samplers, large-scale monitoring ofammonia concentrations has become possible and there are national ammoniamonitoring networks in the Netherlands andin the UK (Sutton et al. 2001a,b). The localsite conditions must be noted for NH3 moni-toring so that the data can be correctly interpreted. The main use of the measure-ment networks for ammonia is to support themodels used to predict the local scale variations in concentration, as neither models nor measurements on their own canadequately predict concentrations.

Criteria for siting measurement stations forammonium are similar to those for SOx.Ammonium is assumed to have slowly vary-ing concentrations and the network densitycan be relatively low.

The deposition of total nitrogen is neededfor many applications in the critical loadframework. It is defined as the sum of totaldeposition of reduced (NHx) nitrogen [NH3dry deposition, NH4

+ aerosol deposition,NH4

+ wet deposition, NH4+ cloudwater/fog

deposition] and oxidised (NOy) nitrogen [NO2dry deposition, HNO3 dry deposition, NO3

-aerosol deposition, NO3

- wet deposition,NO3

- cloudwater/fog deposition]. Themethodological considerations concerningNHx and NOy depositon mapping applyaccordingly.

2.3.8 Mapping base cation and chlo-ride deposition

The deposition of physiologically activebasic cations (Bc = Ca+Mg+K; i.e. the sum ofcalcium, magnesium and potassium) coun-teracts impacts of acid deposition and canimprove the nutrient status of ecosystemswith respect to eutrophication by nitrogeninputs. Sodium (Na) fluxes are needed forestimating the sea-salt fraction of sulfur,chloride, and Bc inputs, and as a tracer forcanopy and soil budget models. In addition,inputs of Bc as well as Na and chloride (Cl)determine the potential acidity of deposition.

As the aim of the Convention is to minimizeacid deposition irrespective of other man-made emissions, base cation inputs notlinked to emissions of acidifying compounds(for example from emissions of Sahara dust,large-scale wind erosion of basic topsoilparticles, etc.) should in principle not beaccounted for within the critical loads frame-work. The non-anthropogenic, non-sea-saltatmospheric input of base cations is definedas a property of the receptor ecosystem andindirectly enters the critical load equation foracidity (see Chapter 5). However, at presentthere is no method to differentiate anthro-pogenic from non-anthropogenic depositionof base cations due to the lack of emissioninventories and long-range transport modelsfor this task.

There are so far no emission inventoriesavailable and therefore base cation andchloride deposition is not yet estimatedusing “classical” LRT models. Work is going

on to provide base cation and chloride dep-osition on a European scale with 50 x 50 km² resolution. As soon as EMEPmodels based on (anthropogenic) emissioninventories are applied to base cations, thedifferent sources of base cations should beidentified and only those relevant for controlof acid deposition included in critical loadscalculations.

The wet deposition of non-sea-salt chloride(Cl*) can be estimated by correcting site fluxes of Cl for the sea-salt fraction (seeChapter 2.3.5, eq. 2.2), then interpolating thederived Cl* concentration (see the wet depo-sition section of Chapter 2.3.2). Similar procedures are applied to map the concen-trations of the non-sea-salt base cations Ca*,Mg* and K* to produce non-sea-salt basecation wet deposition.

Base cation particle deposition can beenestimated from concentrations in wet depo-sition and empirical scavenging ratios (Ederand Dennis 1990, Draaijers et al. 1995). Drydeposition velocities can be inferred as forSO4

2- aerosol and the obtained dry depositionestimates added to measured and interpo-lated wet deposition estimates (e.g., Gaugeret al. 2003). A similar approach has beenused for the UK (RGAR 1997, CLAG 1997).Deposition of base cations have been esti-mated for the Nordic countries based onmonitoring data on concentrations of basecations in precipitation and air-borne parti-cles.

2.3.9 Mapping total potential acid de-position

Total Potential Acid Deposition is defined asthe sum of total deposition of strong acidanions plus ammonium minus non-sea-saltbase cations.

As stated in the preceding subchapter, mostchloride inputs are assumed to be of sea-salt origin, and these are removed from theequation by removing all other sea-saltinputs (i.e. of base cations incl. Na and sul-fate) using a “sea-salt correction” with Na asa tracer. The implicit assumption is that sea-salt is neutral and containing no carbonates.





Surplus chloride inputs (Cl*dep) are assumed

to be due to anthropogenic HCl emissions.

The sum of critical load (for sulfur) and back-ground (non-anthropogenic) base cationdeposition has formerly been defined as crit-ical (sulfur) deposition, as used for the nego-tiations for the Second Sulfur Protocol (Oslo,1994). For comparison to CL(S+N), asdefined in Chapter 5.3.3 (eq. 5.16), only deposition values of S and N are needed.However, if the amount of total acid input isof interest (e.g., for comparison to CL(Acpot),as defined in Chapter 5.3.2), non-sea-saltbase cation and chloride deposition has tobe included into the input side of the poten-tial acidity exceedance equation:

(2.3)

where:

SO*x, dep = non-sea-salt sulfate

deposition

NOy dep, NHx dep = total oxidized/reduced

nitrogen deposition

BC*dep, Cl*

dep = non-sea-salt base

cation / chloride depo-sition

In areas strongly affected by sea spray (highsea-salt Na, Cl, S inputs), the “total potentialacid” definition of Eq. 2.3 becomes problem-atic, since base cations have a beneficialnutrifying effect irrespective of their chemi-cal form (e.g., CaCl vs. CaCO3). At theGrange-over-Sands Workshop 1994 it wasconcluded that total Mg+Ca+K depositionrates should be used for the determination ofcritical loads for acidity (Sverdrup et al. 1995)(see Chapter 5.3.2).

As stated in Chapter 5.3.2, Eq. 2.3 assumesthat deposited NHx is completely nitrifiedand exported from the system as NO3

-, there-by acidifying the system. Thus, with respectto soil acidification it is assumed that 1 molof SO*

x is forming 2 moles of H+, and 1 mol of

NOy, NHx and Cl* each 1 mol of H+.

It is important to be consistent when deter-mining total acid inputs: If results are deter-mined on a site and process level, and if H+

deposition rates are determined separately,NH4

+ inputs (max. 2 eq H+ per mol) have to bedistinguished from NH3 inputs (max. 1 eq H+

per mol). The same applies to SO2 (2 eq H+

per mol) vs. SO42- (0 eq H+ per mol). On a larg-

er scale, this may be neglected: Note thatthe emission and subsequent deposition of 1mol and 2 mol NH3 yields the same potentialacid deposition as the deposition of 1 mol oftheir reaction product (NH4)2SO4, namely 4eq.

2.3.10 Uncertainties of quantification and mapping methods

The errors concomitant with the differentmethods are strongly dependent on thescale considered and the availablity of data.The following analysis is focussed on themapping of concentrations and depositionsfrom the EMEP/MSC-W LRT model, inferen-tial models and interpolated measurements.

Although it is expected that the revisedEMEP/MSC-W 50 x 50 km² Eulerian LRTmodel will become the standard model from2003, the results from the model have notbeen generally available for analysis.However, preliminary indications are that thediscrepancies between model output andmeasurements will be no larger than thosefrom the previous EMEP/MSC-W 150 x150 km² Lagrangian model and significantimprovements are expected with some com-ponents. From a critical loads/levels per-

spective, the change from a 150 x 150 km²scale to a 50 x 50 km² represents a majorimprovement in mapping the concentrationsand deposition. The other notable change inthe move from the Lagrangian model to theEulerian model has been the inclusion ofvegetation specific dry deposition fluxeswithin the EMEP/MSC-W model. Thischange takes the EMEP/MSC-W model partof the way towards a full local inferentialmodel, as the output pollutant fluxes arenow vegetation specific and the issue ofgrid-average values being inappropriately

applied to specific vegetation types, e.g.forests, should no longer occur. However,the issue of scale is important as the con-centrations and the meteorological inputsare still ‘average’ values for the whole 50 x

50 km² grid square. Where concentrationsare expected to be slowly varying, e.g. sul-phate aerosol, the ‘average’ concentrationconcept should not be an important issue,but the approach is still inadequate for therapidly varying concentration fields associ-ated with some primary pollutants. Therewill still be issues to resolve with local mete-orology as grid ‘average’ values, e.g. windspeed, will not be correct for many of theecosystems, e.g. forests on the higher alti-tude areas within the grid square. Therefore,the uncertainty in deposition from theEulerian model should be improved from theLagrangian model but will not match the lev-els of uncertainty which could be achievedby a local scale inferential model. A fullanalysis of uncertainty in the EMEP/MSC-WEulerian model would be a substantial task.

There are a number of references comparingthe EMEP/MSC-W Lagrangian model withavailable EMEP measurements (e.g. Barrettet al. 1995), and with the results of othermodels (Iversen 1991) and the previous ver-sion of this manual (UBA 1996). When eval-uating model-measurement intercompar-isons, it is important to recall that (a) thereare also uncertainties with the measure-ments and (b) the model may be estimatingsomething rather different from what is beingmeasured, e.g. the NO2 concentration at asingle site in a 50 x 50 km² grid square isonly an estimate from a sample of size one ofthe ‘average’ NO2 concentration in thesquare, which is the value the EMEP/MSC-Wmodel is attempting to match. An evaluationof the overall uncertainty of the modelrequires that some further information isavailable on the effects of the spatial distri-bution of measurement sites.

Inferential deposition models treat deposi-tion as a one-dimensional (vertical) transferto homogeneous surfaces with infinitelength, assuming a constant flux layer. Thismeans that the flux from the 50 m referenceheight is assumed to be equal to the flux at

the surface. The reference height must behigh enough so that the concentration is notseverely affected by dry deposition, but itmust be below the surface layer height. Fastchemical reactions as well as the impact ofenhanced turbulent exchange induced bylocal roughness transistions (forest edges,hills, mountains) are not taken into account.Components whose deposition stronglydepends on the aerodynamic resistance(e.g. HNO3, aerosols and cloud water/fogdroplets) will show higher deposition ratesthan modelled. The impact of transitions ondry deposition rates of components like NO2,whose deposition is mainly determined bystomatal conductance, will be relativelysmall.

However, the main uncertainties in dry depo-sition of sulfur and nitrogen compounds inthe inferential framework are due to (1)uncertainties in surface resistance para-metrisations which are not always availablefor all vegetation species and surface types,and (2) uncertainties in the concentrationestimates, which for all the reactive gasesshow a scale of spatial variation which is toogreat to quantify from measurement activi-ties. Surface wetness, which is one of themajor factors determining dry deposition ofsoluble gases (NO2, NH3), is up to nowparametrized very roughly only. The overalluncertainty in surface resistance variesbetween 20% and 100%, depending oncomponent and surface type (van Pul et al.1995).

For the deposition of nitrogen compounds,the main sources of uncertainty weredescribed by Lövblad and Erisman (1992) tobe uncertainties in emissions, concentra-tions and surface resistances to dry deposi-tion, as well as surface wetness and particledeposition for NHx especially in NW andCentral Europe.

Using error propagation methods andassuming that the above mentioned uncer-tainties represent random errors, the totaluncertainty in dry deposition of acidifyingcompounds for an average 10x20 km² grid isestimated at 50-100%. Systematic errors in





dry deposition may arise from neglectingcomplex terrain effects in the parametrisa-tion of the deposition velocity, and fromother simplifications (see Chapter 2.3.2). Asthese calculations are scale dependent andare based on a relatively coarse spatial res-olution, the uncertainty in reality may bemuch larger. For NH3, for example, muchlarger uncertainties in even small areas havebeen shown (Dragosits et al. 2002).Additional uncertainties arise for estimatesof dry deposition for base cations (i) in para-metrising the deposition velocity, (ii) in theprecipitation concentration maps and (iii) inthe scavenging ratios used to derive sur-face-level airborne particulate concentra-tions from concentrations in precipitation.

The overall uncertainty in modelled dry dep-osition velocities integrated over the particlesize distribution representative for alkalineparticles at the Speulder forest (TheNetherlands) was found to equal 60%. Forother sites (and for regions) additional uncer-tainties will arise due to limited availabilityand accuracy of relevant information on landuse and on meteorology. The uncertainty indeposition velocity caused by variations inthe size distribution of alkaline particlesamounts to 30-50%, assuming a mass medi-an diameter (MMD) of 5 µm and taking ageometric standard deviation of 2-3 to repre-sent the variation (Ruijgrok et al. 1996). TheMMD of particles depends on the distanceto sources and on relative humidity.

Also scavenging ratios vary with particlediameter, and assuming the same MMD andstandard deviation as above, the uncertaintyin estimated ambient air concentrationscaused by variation in size distribution isestimated at 50-100%. Large uncertaintiesin air concentrations can be expected veryclose or far from major sources and/or inareas with strongly deviating precipitationclimatology.

Systematic errors arise from (i) using scav-enging ratios based on a limited set of simul-taneous ambient air and precipitation con-centration measurements, (ii) neglectingcomplex terrain effects in the parametrisa-tion of the deposition velocity, (iii) usingannual mean air concentrations and deposi-tion velocities for flux calculation, thereby

neglecting temporal correlations. The totaluncertainty in dry deposition of base cationsfor an average 10x20 km² grid, caused byrandom errors in deposition velocities andair concentrations, is estimated to be 80-120% (Draaijers et al. 1995).

Errors on wet deposition maps are due to (i)limited accuracy of measurements and (ii)the non-representativity of measurementsites. An in-depth analysis of methods tominimize and to quantify these errors can befound in the proceedings of an EMEPWorkshop on these topics (Berg and Schaug1994) and will not be repeated here. Theuncertainty of the wet deposition rate for a50x50 km² grid square, based on interpolat-ed measurements of precipitation concen-tration and precipitation amount, is estimat-ed at 50% on average. Larger uncertainties(about 70%) were found by van Leeuwen etal. (1995) in mountaineous areas and com-plex terrain; the same is to be expected if themeasurement network is not as dense as inNorthwest Europe (Schaug et al. 1993).Using a spatial scale of 5 x 5 km² over theUK, uncertainties were reported as ±35%,reflecting the improved information availableby using a detailed rainfall map with anappropriate orographic model (Smith andFowler 2001).

The uncertainty in total deposition is deter-mined by the uncertainty in wet, dry andcloud and fog deposition. The latter is nottaken into account by most models andrarely measured. As described above, theuncertainty of dry deposition is generallymuch larger than that of wet deposition.Total deposition estimates are more uncer-tain in areas with complex terrain or withstrong horizontal concentration gradients.The uncertainty in total deposition (gridsquare average) of acidifying compounds (N + S) and base cations can be estimated tobe 70-120% and 90-140%, respectively forgrid sizes of the order of 10 x 10 km².

Errors of sulfur deposition rates determinedfrom throughfall measurements vs. inferen-tial models for a forest in the Netherlandswere estimated by Draaijers and Erisman(1993). Error estimates for sulfur and othersubstances are provided by Erisman andDraaijers (1995).

These large uncertainties illustrate the needto validate model results by measurementsof airborne concentrations, wet deposition,dry deposition and throughfall, as stated inChapter 2.3.2.

This, and the scale-specific qualities of themethods, should be kept in mind when read-ing the substance-specific listing of individ-ual methods in Chapters 2.3.3-2.3.9.

These maps are designed to be used in com-bination with critical loads and critical levelsmaps to show where and by how much crit-ical loads and critical levels are exceeded.The use of deposition data with critical loadsdata very often involves different scales ofthe different data sources and, in mostcases, the critical loads data are provided ata finer resolution than the deposition dataresulting in an underestimation of the criticalload exceedance. These issues have beendiscussed above and improved depositionestimates, for example by using nationalmodels at a finer spatial resolution, canimprove the quality of the critical loadexceedances. One important point re-iterat-ed here is that it is essential to note any dif-ferent scales in the legends to figures andmaps.

National maps and model outputs may becompared with data from the EMEP model,since the EMEP data are used for theIntegrated Assessment Modelling activitiesand for the protocol developments. If thenational datasets deviate strongly fromEMEP model data, EMEP should be notifiedand scientifically based improvements of theEMEP models should be developed.National datasets may always be used fornational purposes, and national model out-puts should be calibrated with monitoringresults at international (e.g., EMEP), nationaland subnational levels. This should alsoensure that it is possible to compare nation-al maps where they meet at internationalborders, and these activities at various levels(regional, national, international) should be

coordinated by comparing results at EMEPor other workshops.



2.4 Use of deposition load and con-centration maps



Table

2.2

:S

umm

ary

of

met

hod

s fo

r at

mo

sphe

ric

conc

entr

atio

n an

d d

epo

sitio

n m

app

ing

(see

Cha

pte

r 2.

3), t

heir

link

age

to e

mis

sio

n in

vent

ori

es, a

nd t

heir

sui

tab

ility

fo

r m

ap-

pin

g a

nd s

ite-l

evel

val

idat

ion

(1=

go

od

, 2=

usef

ul, 3

=sh

oul

d n

ot

be

used

) (se

e te

xt!)



Anshelm F, Gauger Th (2001) Mapping ofecosystem specific long-term trends indeposition loads and concentrations ofair pollutants in Germany and their com-parison with Critical Loads and CriticalLevels. Part 2: Mapping Critical Levelsexceedances. Final Report of Project29942210. Inst. for Navigation, StuttgartUniversity and Umweltbundesamt.

Asman WAH, Van Jaarsveld HA (1992) A vari-able-resolution transport model appliedfor NHx in Europe. Atmos. Environ. PartA-General Topics 26: 445-464.

Asman WAH, Drukker B, Janssen AJ (1988)Modeled historical concentrations anddepositions of ammonia and ammoniumin Europe. Atmos. Environ. 22: 725-735.

Aubinet M, Grelle A, Ibrom A, Rannik U,Moncrieff J, Foken T, Kowalski AS, MartinPH, Berbigier P, Bernhofer C, Clement R,Elbers J, Granier A, Grunwald T,Morgenstern K, Pilegaard K, Rebmann C,Snijders W, Valentini R, Vesala T (2000)Estimates of the annual net carbon andwater exchange of forests: The EURO-FLUX methodology. Adv. Ecol. Res. 30:113-175.

Bak J (2001) Uncertainties in large scaleassessments of critical loadsexceedances. Water, Air and SoilPollution Focus, Vol. 1 Nos.: 1-2, 265-280.

Barrett K, Seland Ø, Foss A, Mylona S,Sandnes H, Styve H, Tarrasòn L (1995)European transboundary acidifying airpollution. EMEP/MSC-W Report 1/95, TheNorwegian Meteorological Institute, Oslo,Norway.

Beier C, Gundersen P, Rasmussen L (1992)A new method for estimation of dry depo-sition of particles based on throughfallmeasurements in a forest edge. Atmos.Environ. 26A: 1553-1559.

Berg T, Schaug J (eds.) (1994) Proceedingsof the EMEP Workshop on the Accuracyof Measurements with WMO sponsoredsessions on Determining the

Representativeness of MeasuredParameters in a Given Grid Square asCompared to Model Calculations,Passau, Germany, 22-26 Nov. 1993.

Beswick KM, Hargreaves K, Gallagher MW,Choularton TW, Fowler D (1991) Sizeresolved measurements ofcloud droplet deposition velocity to a for-est using an eddy correlation technique.Roy QJ Meteor. Soc. 117: 623-645.

Bredemeier M (1988) forest canopy transfor-mation of atmospheric deposition. WaterAir Soil Poll. 40: 121138.

CLAG (1997) Deposition fluxes in the UnitedKingdom: A compilation of the currentdeposition maps and mapping methods(1992-1994) used for Critical Loadsexceedance assessment in the UnitedKingdom. Critical Loads Advisory GroupSub-group report on Depositon Fluxes. 45pp. Penicuik: Institute of TerrestrialEcology.

Coyle M, Smith RI, Stedman JR, Weston KJ,Fowler D (2002) Quantifying the spatialdistribution of surface ozone concentra-tion in the UK. Atmos. Environ. 36: 1013-1024.

Dore AJ, Choularton TW, Fowler D (1992) Animproved wet deposition map of theUnited Kingdom incorporating the topo-graphic dependence of rainfall concentra-tions. Atmos. Environ. 26A: 1375-1381.

Draaijers GPJ, Erisman JW (1993)Atmospheric sulfur deposition to foreststands - throughfall estimates comparedto estimates from inference. Atmos.Environ. 27A: 43-55.

Draaijers GPJ, Erisman JW (1995) A canopybudget model to assess atmosphericdeposition from throughfall measure-ments. Water Air Soil Pollut. 85: 2253-2258.

Draaijers GPJ, van Leeuwen EP, Potma C,van Pul WAJ, Erisman JW (1995) Mappingbase cation deposition in Europe on a10x20 km grid. Water Air Soil Pollut. 85:2389-2394.

References

Draaijers GPJ, Erisman JW, Spranger T,Wyers GP (1996a) The application ofthroughfall measurements. Atmos.Environ. 30: 3349-3361.

Draaijers GPJ, Erisman JW, van LeeuwenNFM, Römer FG, te Winkel BH, VeltkampAC, Vermeulen AT, Wyers GP (1996b) Theimpact of canopy exchange on differ-ences observed between atmosphericdeposition and throughfall fluxes. Atmos.Environ. 31: 387-397.

Draaijers GPJ, Bleeker A, van der Veen D,Erisman JW, Möls H, Fonteijn P,Geusenbrock M (2001) Field intecompar-ison of throughfall, stemflow and precipi-tation measurements performed withinthe framework of the Pan EuropeanIntensive Monitoring Programme ofEU/ICP Forests TNO report R2001/140.

Dragosits U, Theobald MR, Place CJ, Lord E,Webb J, Hill J, ApSimon HM, Sutton MA(2002) Ammonia emission, depositionand impact assessment at the field scale:a case study of sub-grid spatial variablili-ty. Environ. Pollut. 117: 147-158.

Duyzer JH, Deinum G, Baak J (1995) Theinterpretation of measurements of surfaceexchange of nitrogen oxides; correctionsfor chemical reactions. PhilosophicalTransactions of the Royal Society ofLondon A 351: 1-18.

Duyzer J, Nijenhuis B, Weststrate H (2001)Monitoring and modelling of ammoniaconcentrations and deposition in agricul-tural areas of the Netherlands. Water, Airand Soil Pollution Focus, Vol 1 Nos. 5-6:131-144.

EDC (Environmental Data Centre) (1993)Manual for Integrated Monitoring.Programme Phase 1993-1996. NationalBoard of Waters and the Environment,Helsinki.

Eder BK, Dennis RL (1990) On the use ofscavenging ratios for the inference of sur-face level concentration. Water Air SoilPollut. 52: 197-215.

Emberson LD, Ashmore MR, Cambridge HM,Simpson D, Tuovinen JP (2000) Modellingstomatal ozone flux across Europe.Environ. Pollut. 109: 403-413.

EMEP/CCC (1996) EMEP Manual forSampling and Analysis. EMEP/CCCReport 1/95.

Erisman JW, Draaijers G (1995) Atmosphericdeposition in relation to acidification andeutrophication. Studies in EnvironmentalScience 63, Elsevier, Amsterdam.

Erisman JW, van Pul A, Wyers P (1994a)Parameterization of dry deposition mech-anisms for the quantification of atmos-pheric input to ecosystems. Atmos.Environ. 28: 2595-2607.

Erisman JW, Beier C, Draaijers GPJ,Lindberg S (1994b) Review of depositionmonitoring methods. Tellus 46B: 79-93.

Erisman JW, Draaijers G, Duyzer J,Hofschreuder P, van Leeuwen N, Römer F,Ruijgrok W, Wyers P (1995) Particle depo-sition to forests. In: Heij GJ, Erisman JW(eds.): Proceedings of the conference“Acid rain research: do we have enoughanswers?”, `s-Hertogenbosch, TheNetherlands, 10-12 Oct. 1994. Studies inEnvironmental Science 64: 115-126.

Erisman JW, Mennen MG, Fowler D,Flechard CR, Spindler G, Gr?ner A,Duyzer JH, Ruigrok W, Wyers GP (1998a)Deposition Monitoring in Europe. Environ.Monit. Assess. 53: 279-295.

Erisman JW, Bleeker A, van Jaarsveld H(1998b) Atmospheric deposition ofammonia to semi-natural vegetation inthe Netherlands-methods for mappingand evaluation. Atmos. Environ. 32: 481-489.

Ferm M (2001) Validation of a diffusive sam-pler for ozone in workplace atmospheresaccording to EN838. Proceedings fromInternational Conference Measuring AirPollutants by Diffusive Sampling,Montpellier, France 26-28 September2001: 298-303.

Ferm M, Svanberg PA (1998) Cost-efficienttechniques for urban and backgroundmeasurements of SO2 and NO2. Atmos.Environ. 32: 1377-1381.

Flechard CR, Fowler D (1998) Atmosphericammonia at a moorland site. II: Long-termsurface/atmosphere micrometeorological





flux measurements. Roy QJ Meteor. Soc.124: 733-757.

Flemming J (2003) Immissionsfelder ausBeobachtung, Modellierung und derenKombination, PhD thesis, Free UniversityBerlin, FB Geowissenschaften (inGerman).

Fowler D, Granat L, Koble R, Lovett G,Moldan F, Simmons C, Slanina SJ,Zapletal M (1993) Wet, cloud water andfog deposition. (Working Group report).In: Models and methods for the quantifi-cation of atmospheric input to ecosys-tems, edited by Lovblad G, Erisman JW,Fowler: 912. Copenhagen: Nordic Council

of Ministers.

Fowler D, Jenkinson DS, Monteith JL,Unsworth MH (eds.) (1995a) Theexchange of trace gases between landand atmosphere. Proceedings of a RoyalSociety discussion meeting. Philos. T.Roy. Soc. A 351 (1696), pp. 205-416.

Fowler D, Leith ID, Binnie J, Crossley A,Inglis DWF, Choularton TW, Gay M,Longhurst JWS, Conland DE (1995b)Orographic enhancement of wet deposi-tion in the United Kingdom: continuousemonitoring. Water Air Soil Pollut. 85:2107-2112.

Fowler D, Smith RI, Weston KJ (1995c)Quantifying the spatial distribution of sur-face ozone exposure at the 1kmx1kmscale. In: Fuhrer, Achermann (eds.) (1995):196-205.

Fowler D, Pitcairn CER, Erisman JW (eds.)(2001a) Air-Surface Exchange of Gasesand Priticles (2000). Proceedings of theSixth International Conference on Air-Surface Exchange of Gases and Particles,Edinburgh 3-7 July, 2000. Water Air andSoil Pollution Focus, Vol. 1 Nos. 5-6: 1-464.

Fowler D, Coyle M, Flechard C, HargreavesK, Storeton-West R, Sutton M, ErismanJW (2001b) Advances in micrometeoro-logical methods for the measurement andinterpretation of gas and particles nitro-gen fluxes. Plant Soil 228: 117-129.

Fowler D, Sutton MA, Flechard C, Cape JN,Storeton-West R, Coyle M, Smith RI

(2001c) The Control of SO2 dry deposition

on to natural surfaces and its effects onregional deposition. Water, Air and SoilPollution Focus, Vol. 1 Nos. 5-6: 39-48.

Fuhrer J (2002) Ozone impacts on vegeta-tion. Ozone-Sci. Eng. 24: 69-74.

Gallagher M, Fontan J, Wyers P, Ruijgrok W,Duyzer J, Hummelsh P, Fowler D (1997)Atmospheric Particles and theirInteractions with Natural Surfaces. In:Biosphere-Atmosphere Exchange ofPollutants and Trace Substances (ed.Slanina S.): 45-92. Springer-Verlag.

Gauger Th, Anshelm F, Schuster H, ErismanJW, Vermeulen AT, Draaijers GPJ, BleekerA, Nagel HD (2003) Mapping of ecosy-stem specific long-term trends in deposi-tion loads and concentrations of air pollu-tants in Germany and their comparisonwith Critical Loads and Critical Levels.Part 1: Deposition Loads 1990-1999. FinalReport of Project 29942210. Inst. forNavigation, Stuttgart University andUmweltbundesamt.

Grunhage L, Haenel HD (1997) Platin (plant-atmosphere interaction) In: A model ofplant-atmosphere interaction for estimat-ing absorbed doses of gaseous air pollu-tants. Environ. Pollut. 98: 37-50.

Hicks BB, Baldocchi DD, Meyers TP, MattDR, Hosker Jr RP (1987) A preliminarymultiple resistance routine for deriving drydeposition velocities from measuredquantities. Water Air Soil Pollut. 36: 311-330.

Hicks BB, McMillen R, Turner RS, Holdren JrGR, Strickland TC (1993) A national criti-cal loads framework for atmospheric dep-osition effects assessment: III. Depositioncharacterization. Environ. Manage. 17,No.3: 343-353.

Iversen T (editor) (1991) Comparison of threemodels for long term photochemical oxi-dants in Europe. EMEP/MSC-W Report3/91. The Norwegian MeteorologicalInstitute, Oslo, Norway.

Loibl W, Smidt S (1996) Ozone exposure -Areas of potential ozone risk for selectedtree species. Environ. Sci. Pollut. R. 3:213-217.

Lövblad G (1996) In: Knoflacher M,Schneider J, Soja G (eds.): Exceedance ofCritical Loads and Levels - Spatial andTemporal Interpretation of Elements inLandscape Sensitive to AtmosphericPollutants. Report from a UNECEWorkshop held in Vienna, Austria, 22-24November 1995. Federal Ministry forEnvironment, Youth and Family, Austria,Conference Papers 15: 236.

Lövblad G, Erisman JW (1992) Deposition ofNitrogen in Europe. Background docu-ment; In Grennfelt P, Thörnelöf E (eds.)(1992) Critical Loads for Nitrogen; reportfrom a workshop held at Lökeberg,Sweden, 6-10 April 1992. Nord 1992: 41.

Lövblad G, Erisman JW, Fowler D (eds.)(1993) Models and Methods for theQuantification of Atmospheric Input toEcosystems. Proceedings of an interna-tional workshop on the deposition ofacidifying substances, Göteborg 4-6November 1992. Nordiske Seminar- ogArbejdsrapporter 1993: 573, NordicCouncil of Ministers.

Lövblad G, Grennfelt P, Kärenlampi L, LaurilaT, Mortensen L, Ojanperä K, Pleijel H,Semb A, Simpson D, Skärby L, TuovinenJP, Tørseth K (1996) Ozone exposuremapping in the Nordic Countries, Reportfrom a Nordic cooperation projectfinanaced by the Nordic Council ofMiisters, Report Tema Nord 1996: 528.

Lyman J, Fleming RH (1940) Composition ofsea water. Journal of Marine Research 3:134-146.

Nemitz E, Milford C, Sutton MA (2001) A two-layer canopy compensation point modelfor describing bi-directional biosphere-atmosphere exchange of ammonia. RoyQJ Meteor. Soc. 127: 815-833.

PORG (1997) Ozone in the UK. Fourth reportof the Photochemical Oxidants ReviewGroup. 234pp. DETR (ITE Edinburgh).

RGAR (1997) Acid Deposition in the UnitedKingdom 1992-1994. Fourth Report of theReview Group on Acid Rain. 176 pp.HMSO, London.

Ruijgrok W, Tieben H, Eisinga P (1996) Thedry deposition of particles to a forest

canopy: a comparison of model andexperimental results. Atmos. Environ. 31:399-415.

Schaug, J., Iversen, T. and Pedersen, U.(1993): Comparison of measurementsand model results for airborne sulfur andnitrogen components with Kriging.Atmos. Environ., 27A, 831-844.

Schjoerring JK, Husted S, Mattsson M (1998)Physiological parameters controllingplant-atmosphere ammonia exchange.Atmos. Environ. 32: 491-498.

Singles R, Sutton MA, Weston KJ (1998) Amulti-layer model to describe the atmos-pheric transport and deposition of ammo-nia in Great Britain. Atmos. Environ. 32:393-399.

Sjöberg K, Lövblad G, Ferm M, Ulrich E,Cecchini S, Dalstein L (2001) Ozonemeasurements at forest plots using diffu-sive samplers. Proc. from InternationalConference Measuring Air Pollutants byDiffusive Sampling, Montpellier, France26-28 September 2001. p116-123.

Slanina S ed. (1997) Biosphere-AtmosphereExchange of Pollutants and TraceSubstances. Springer-Verlag.

Slinn WGN (1982) Predictions for particledeposition to vegetativ surfaces. Atmos.Environ. 16: 1785-1794.

Smith RI, Fowler D (2001) Uncertainty in esti-mation of wet deposition of sulphur.Water, Air and Soil Pollution Focus, Vol 1.Nos. 5-6: 341-354.

Smith RI, Fowler D, Bull KR (1995)Quantifying the scale dependence in esti-mates of wet and dry deposition and theimplication for critical load exceedances.In: Heij GJ, Erisman JW (eds.):Proceedings of the conference “Acid rainresearch: do we have enough answers?”,`s-Hertogenbosch, The Netherlands, 10-12 Oct. 1994. Studies in EnvironmentalScience 64: 175-186.

Smith RI, Fowler D, Sutton MA, Flechard C,Coyle M (2000) Regional estimation ofpollutant gas dry deposition in the UK:model description, sensitivity analysesand outputs. Atmos. Environ. 34: 3757-3777.





Spranger T (1992) Erfassung und ökosys-temare Bewertung der atmosphärischenDeposition und weiterer oberirdischerStoffflüsse im Bereich der BornhövederSeenkette. Ph.D. thesis, Christian-Albrechts Universität, Kiel, Germany (inGerman).

Spranger T, Kunze F, Gauger Th, Nagel HD,Bleeker A, Draaijers GPJ (2001) CriticalLoads exceedances in Germany and theirdependence on the scale of input data.Water, Air and Soil Pollution Focus, Vol. 1Nos. 1-2: 335-351.

Stedman JR, Vincent KJ, Campbell GW,Goodwin JWL, Downing CEH (1997) Newhigh resolution maps of estimated back-ground ambient Nox and NO2 concentra-tions in the UK. Atmos. Environ. 31: 3591-3602.

Sutton MA, Lee DS, Dollard GJ, Fowler D(eds.) (1998): International conference onatmospheric ammonia: emmison, deposi-tion and environmental impacts. Atmos.Environ. 32: 269-594.

Sutton MA, Nemitz E, Fowler D, Wyers GP,Otjes RP, Schjoerring JK, Husted S,Nielsen KH, San José, Moreno J,Gallagher MW, Gut A (2000) Fluxes ofammonia over oilseed rape Overview ofthe EXAMINE experiment. Agr. ForestMeteorol. 105: 327-349.

Sutton MA, Miners B, Tang YS, Milford C,Wyers GP, Duyzer JH, Fowler D (2001a)Comparison of low-cost measurementtechniques for long-term monitoring ofatmospheric ammonia. J. Environ.Monitor. 3: 446-453.

Sutton MA, Tang YS, Dragosits U, FournierN, Dore T, Smith RI, Weston KJ, Fowler D(2001b) A spat ial analysis of atmosphericammonia and ammonium in the UK. TheScientific World, 1 (S2): 275-286.

Sverdrup H, De Vries W, Hornung M, CresserM, Langan S, Reynolds B, Skeffington R,Robertson W (1995) Modification of thesimple mass-balance equation for calcu-lation of critical loads of acidity. In:Hornung M, Sutton MA, Wilson RB (eds.)Mapping and modelling of critical loadsfor nitrogen: a Workshop Report, Grange-

over-Sands, October 1994. Penicuik:Instsitute of Terrestrial Ecology: 87-92.

Sverdrup HU, Johnson MW, Fleming RH(1946) The Oceans - Their PhysicsChemistry and General Biology. Prentice-Hall, New York, 1087 pp.

UBA (1996) Manual on Methodologies andCriteria for Mapping Critical Levels/Loadsand geographical areas where they areexceeded. UBA-Texte 71/96.

Ulrich B (1983) Interaction of forest canopieswith atmospheric constituents: SO2 ,

alkali and earth alkali cations and chlo-ride. In: Ulrich B, Pankrath J (eds.): Effectsof accumulation af air pollutants in forestecosystems: 33-45.

UNECE ICP Forests (1999) Manual on meth-ods and criteria for harmonized sampling,assessment, monitoring and analysis ofthe effects of air pollution on forests. PartVI: Measurement of Deposition and AirPollution.

van Leeuwen EP, Erisman JW, Draaijers GPJ,Potma CJM, van Pul WAJ (1995)European wet deposition maps based onmeasurements. Report No. 722108006,National Institute of Public Health andEnvironmental Protection, Bilthoven, TheNetherlands.

van Pul WAJ, Potma C, van Leeuwen E,Draaijers G, Erisman JW (1995): EDACS:European deposition maps of acidifyingcomponents on a small scale. Modeldescription and preliminary results. RIVMReport no. 722401005, Bilthoven, TheNetherlands.

Walton S, Gallagher MW, Duyzer JH (1997)Use of a detailed model to study theexchange of Nox and O3 above andbelow a deciduous canopy. Atmos.Environ. 31: 2915-2932.

Who do you ask for further advice ?

For questions on ......... please contact:

EMEP long-range transport models:Erik Berge, The Norwegian MeteorologicalInstitute, P.O. Box 43 - Blindern, N - 0313Oslo, Norway, Tel. +47 2296 - 3000; Fax.+47 2296 - 3050

High resolution modelling of dry andcloud/fog deposition, combination withmaps of interpolated wet deposition:Jan-Willem Erisman, RIVM-LLO, P.O. Box 1,3720 BA Bilthoven, The Netherlands, Tel.+31-30-274-2824; Fax. +31-30-2287531David Fowler, Centre for Ecology andHydrology, Bush Estate, Penicuik,Midlothian, EH26 0QB, United Kingdom,Tel. +44-131-445-4343; Fax. +44-131-445-3943

Combination of high-resolution models withlong-range transport models: Jan-Willem Erisman, RIVM-LLO, P.O. Box 1,3720 BA Bilthoven, The Netherlands; Tel.+31-30-274-2824; Fax. +31-30-2287531Erik Berge, The Norwegian MeteorologicalInstitute, P.O. Box 43 - Blindern, N - 0313Oslo, Norway, Tel. +47 2296 3000; Fax. +472296 3050

Measurement and interpolation methodolo-gy (Ambient air concentrations, wet andbulk deposition): EMEP CCC, NILU, Postbox 100, N-2007Kjeller, Norway, Tel. +47-6389-8000; Fax.+47-6389-8050

Evaluating total deposition maps withthroughfall measurements:Gun Lövblad, Swedish EnvironmentalResearch Institute, Box 47086, 40258Göteborg, Sweden; Tel. +46-31-725 6240,Fax. +46-31- 725 6290, [email protected] Erisman, RIVM-LLO, P.O. Box 1,3720 BA Bilthoven, The Netherlands; Tel+31-30-274-2824; Fax +31-30-2287531

Diffusive samplers for air pollution monito-ring:Martin Fern, Swedish EnvironmentalResearch Institute, Box 47086, 40258Göteborg, Sweden; Tel. +46-31-725 6224,Fax. +46-31-725 6290 [email protected]

General information on mapping excercisescan also be obtained by contacting theCoordination Center for Effects, CCE,Netherlands: Jean-Paul Hettelingh, Tel. +31-30-74 30 48; Fax. +31-30-74 29 71

General information on modelling:Ron Smith, Centre for Ecology andHydrology, Bush Estate, Penicuik,Midlothian EH26 0QB, Tel. +44 131 4454343; Fax. +44 131 445 3943

General information on NOx, NO3:

Kim Pilegaard, Riso National Laboratory, POBox 49, DK-4000 Roskilde, Denmark Tel.+45 4677 4677, Fax. +45 4677 4160Jan Duyzer, TNO-MEP, Postbus 342, 7300AH, Apeldoorn, The Netherlands, Tel. +3155 549 3944; Fax. +31 55 549 3252



The purpose of this chapter is to provideinformation on the critical levels for sensitivevegetation and how to calculate critical levelexceedance. Methods for mapping pollutantconcentrations, deposition and exceedanceare provided elsewhere (Chapter 2).

Excessive exposure to atmospheric pollu-tants has harmful effects for a variety of veg-etation. Critical levels are described in differ-ent ways for different pollutants, includingmean concentrations, cumulative exposuresand fluxes through plant stomata. Theeffects considered significant here varybetween receptor and pollutant and includegrowth changes, yield losses, visible injuryand reduced seed production. The receptorsare generally divided into five major cate-gories: agricultural crops, horticulturalcrops, semi-natural vegetation, natural vege-tation and forest trees. However, for somepollutants, e.g. ozone, semi-natural vegeta-tion and natural vegetation have beengrouped together under the name (semi-)natural vegetation.

Critical levels were defined in the previousversion of this manual (UNECE, 1996) as “theatmospheric concentrations of pollutants inthe atmosphere above which adverse effectson receptors, such as human beings, plants,ecosystems or materials, may occur accord-ing to present knowledge”. For this revisedchapter, the critical levels for vegetation aredefined as the “concentration, cumulativeexposure or cumulative stomatal flux ofatmospheric pollutants above which directadverse effects on sensitive vegetation mayoccur according to present knowledge”.Critical level exceedance maps show the dif-ference between the critical level and themapped, monitored or modelled air pollutantconcentration, cumulative exposure orcumulative flux.

The critical level values have been set,reviewed and revised for O3, SO2, NO2 andNH3 at a series of UNECE Workshops: BadHarzburg (1988); Bad Harzburg (1989);

Egham (1992; Ashmore and Wilson, 1993);Bern (1993; Fuhrer and Achermann, 1994);Kuopio (1996; Kärenlampi and Skärby, 1996),Gerzensee (1999; Fuhrer and Achermann,1999) and Gothenburg (2002; Karlsson,Selldén and Pleijel, 2003).

For SO2, NOx and NH3, recommendations aremade for concentration-based critical levels.For ozone, separate cumulative concentra-tion-based (previously described as level I)and cumulative stomatal flux-based (previ-ously described as level II) critical levels aredescribed which use different scientificbases of risk assessment, as explained inSection 3.3. Since the previous version ofthis manual was published (UNECE, 1996),much progress has been made with the crit-ical levels for ozone and this chapter pro-vides an in-depth description of the criticallevels, their scientific bases and how to cal-culate exceedance. As part of this progress,it was agreed that the level I and level II ter-minology is no longer appropriate todescribe critical levels for ozone and theseterms are not used in this chapter.

3 Mapping Critical Levels for Vegetation

3.1 General remarks and objectives

Mapping Manual 2004 • Chapter III Mapping Critical Levels for Vegetation Page III - 1


Mapping Manual 2004 • Chapter III Mapping Critical Levels for VegetationPage III - 2

3.2.1 SO2

The critical levels for SO2 that were estab-lished in Egham in 1992 (Ashmore andWilson, 1993) are still valid (Table 3.1). Thereare critical levels for four categories ofreceptors – for sensitive groups of lichens,for forest ecosystems, (semi-) natural vege-tation and agricultural crops. These criticallevels have been adopted by WHO (2000).

Exceedance of the critical level for (semi-)natural vegetation, forests, and, whenappropriate, agricultural crops occurs wheneither the annual mean concentration or thewinter half-year mean concentration isgreater than the critical level; this is becauseof the greater impact of SO2 under winterconditions.

3.2.2 NOxThe critical levels for NOx are based on thesum of the NO and NO2 concentrationsbecause there is insufficient knowledge toestablish separate critical levels for the two

pollutants, although some evidence indi-cates that at low concentrations typical ofambient conditions, NO is more phytotoxicthan NO2. Since the type of response variesfrom a fertilizer effect to toxicity dependingon concentration, all effects were consid-ered to be adverse. Growth stimulationswere of greatest concern for (semi-) naturalvegetation because of the likelihood ofchanges in interspecific competition. Separate critical levels were not set forclasses of vegetation because of the lack ofavailable information. However, the followingranking of sensitivity was established:

(semi-) natural vegetation > forests > crops

Critical levels for NOx were first establishedin 1992 at the Egham workshop. The back-ground papers on NOx and NH3 presented atthe Egham workshop (Ashmore & Wilson,1993) were further developed as the basis ofthe Air Quality Guidelines for Europe,

published by WHO in 2000. This furtheranalysis incorporated a formal statisticalmodel to identify concentrations to protect95% of species at a 95% confidence level. Inthis re-analysis, growth stimulation was alsoconsidered as a potentially adverse

3.2 Critical levels for SO2, NOx, NH4and O3

Table 3.1: Critical levels for SO2 (µg m-3) by vegetation category

*The forest ecosystem includes the response of the understorey vegetation.

ecological effect. Furthermore, a critical levelbased on 24h mean concentrations wasconsidered to be more effective than onebased on 4h mean concentrations as includ-ed in the previous version of the MappingManual (UNECE, 1996). Since the WHOguidelines were largely based on analysisextending the background information pre-sented at the Egham workshop, the criticallevels in Table 3.2, which are identical tothose of WHO (2000), should now be used.For application for mapping critical levelsand their exceedance, it is strongly recom-mended that only the annual mean valuesare used, as mapped and modelled values ofthis parameter have much greater reliability,and the long-term effects of NOx are thoughtto be more significant than the short-termeffects.

Table 3.2: Critical levels for NOx (NO and NO2 added),expressed as NO2 (µg m-3)

Some biochemical changes may occur atconcentrations lower than the critical levels,but there is presently insufficient evidence tointerpret such effects in terms of critical lev-els.

3.2.3 NH3

The fertilisation effect of NH3 can in thelonger term lead to a variety of adverseeffects, including growth stimulation andincreased susceptibility to abiotic (drought,frost) and biotic stresses. In the short-termthere are also direct effects. As for NOx, forapplication for mapping critical levels andtheir exceedance, it is strongly recommend-ed that only the annual mean values of NH3are used, as mapped and modelled values of

this parameter have much greater reliability,and the long-term effects of NH3 are thoughtto be more significant than the short-termeffects. The critical levels in Table 3.3 refer to all veg-etation types, including the most sensitive.The aim of the critical levels defined is toprotect the functioning of plants and plantcommunities. The following distinction inreceptor sensitivity is proposed:

(semi-) natural vegetation > forests > crops

As for NOx, the Air Quality Guidelines forEurope for NH3 (WHO, 2000) were based onbackground papers developing informationpresented at the Egham workshop.Therefore, the critical levels for NH3 are nowthose proposed by WHO (2000), in which theprevious values for averaging times of 1 hourand 1 month have been deleted.

Table 3.3: Critical levels for NH3 (µg m-3)

3.2.4 O3

Three cumulative exposure approaches areused to define critical levels for ozone: stom-atal fluxes, ozone concentrations andvapour-pressure deficit-modified ozone con-centrations. Each approach uses the ozoneconcentration at the top of the canopy andincorporates the concept that the effects ofozone are cumulative and values aresummed over a defined time period. In eachcase, a specific threshold is used and onlyozone concentrations, vapour-pressuredeficit-modified ozone concentrations orinstantaneous stomatal fluxes above thatthreshold are summed. When the sum of thevalues above the threshold exceeds the





critical value defined in the relevant table foreach approach and vegetation type, then thecritical level for that approach and vegeta-tion type has been exceeded.

The previous version of this manual (UNECE,1996) included concentration-based criticallevels that used AOTX (ozone concentrationsaccumulated over a threshold of X ppb) asthe ozone parameter. However, severalimportant limitations and uncertainties havebeen recognised for using AOTX. In particu-lar, the real impacts of ozone depend on theamount of ozone reaching the sites of dam-age within the leaf, whereas AOTX-basedcritical levels only consider the ozone con-centration at the top of the canopy. TheGerzensee Workshop in 1999, recognisedthe importance of developing an alternativecritical level approach based on the flux ofozone from the exterior of the leaf throughthe stomatal pores to the sites of damage(stomatal flux). This approach required thedevelopment of mathematical models toestimate stomatal flux, primarily from knowl-edge of stomatal responses to environmen-tal factors. It was agreed at the GothenburgWorkshop in November 2002 that ozoneflux-effect models were sufficiently robustfor the derivation of flux-based critical levels,and such critical levels should be included inthis revised Manual for wheat, potato andprovisionally for beech. To accommodate these different approach-es, the terminology and symbols used indescribing critical levels of ozone have beenrevised and are described in Table 3.4 andthe critical levels are provided in Table 3.5.For guidance on the selection of critical lev-els and which methods to use, a flow chart isprovided in Figure 3.1. The scientific basesof these critical levels are provided inSection 3.3 and methods for calculatingexceedance are in Sections 3.4 and 3.5. Thethree approaches can be summarised as follows:

Stomatal flux-based critical levels(Clef) for ozone take into account thevarying influences of temperature,water vapour pressure deficit (VPD),light (irradiance), soil water potential

(SWP), ozone concentration and plantdevelopment (phenology) on thestomatal flux of ozone. They thereforeprovide an estimate of the criticalamount of ozone entering through thestomata and reaching the sites ofaction inside the plant. This is animportant new development in thederivation of critical levels because,for example, for a given ozone con-centration, the stomatal flux in warm,humid conditions with moist soil canbe much greater than that in hot, dryconditions with dry soil because thestomatal pores will be more widelyopen. Concentration-based criticallevels do not differentiate betweensuch climatic conditions and wouldnot indicate the increased risk ofdamage in warm, humid conditions.The hourly mean stomatal flux ofozone based on the projected leafarea (PLA), Fst (in nmol m-2 PLA s-1), isaccumulated over a stomatal fluxthreshold of Y nmol m-2 s-1. The accu-mulated stomatal flux of ozone abovea flux threshold of Y (AFstY), is calcu-lated for the appropriate time-windowas the sum over time of the differ-ences between hourly mean values ofFst and Y nmol m-2 PLA s-1 for the peri-ods when Fst exceeds Y. The stomatalflux-based critical level of ozone, CLef mmol m-2 PLA, is then the cumula-tive stomatal flux of ozone, AFstY,above which direct adverse effectsmay occur according to presentknowledge. Values of CLef have beenidentified for some crops (wheat andpotato) and provisionally for sensitiveforest trees (represented by birch andbeech). This approach cannot yet beapplied to (semi-) natural vegetation.

Concentration-based critical levels(CLec) for ozone use the concentra-tion at the top of the canopy accumu-lated over a threshold concentrationfor the appropriate time-window and

thus do not take account of the stomatal influence on the amount ofozone entering the plant. This value isexpressed in units of ppm h (µmol mol-1 h). The term AOTX (concen-tration accumulated over a thresholdozone concentration of X ppb) hasbeen adopted for this index; in thismanual “X” is either 30 or 40 ppb forAOT30 and AOT40 respectively. Valuesof CLec are defined for agriculturaland horticultural crops, forests and(semi-) natural vegetation.

VPD-modified concentration-basedcritical levels take into account themodifying influence of VPD on thestomatal flux of ozone by multiplyingthe hourly mean ozone concentrationat the top of the canopy by an fVPDfactor to get the VPD-modified ozoneconcentration ([O3]VPD). The [O3]VPDis accumulated over a threshold con-centration during daylight hours overthe appropriate time-window. Thisvalue is expressed in units of ppm h(µmol mol-1 h). The term AOT30VPD(VPD-modified concentration accu-mulated over a threshold ozone con-centration of 30 ppb) has been adopt-ed for this index; this index is onlyused to define the short-term criticallevel for the development of visibleinjury on crops.

It should be noted that consideration wasgiven at the Gothenburg Workshop(November 2002) to a fourth approach, theMaximum Permissible Ozone Concentration(MPOC) (Krause et al., 2003). The MPOCapproach could possibly be adopted as anadditional concentration-based index formapping potential risk in national evalua-tions, but should not be used for mappingpurposes at the European scale. Thisapproach was considered as potentially use-ful in the context of forest trees, but not for(semi-) natural vegetation and crops.Economic losses attributable to ozone pollu-tion could not be estimated on the basis of

this approach. Further validation and devel-opment of the methodology may be possibleunder the framework of ICP Forests. A briefdescription of the MPOC approach is provid-ed in Chapter 3, Annex 1.





Term Abbreviation Units Explanation

Terms for concentration-based critical levels

Concentration

accumulated over a

threshold ozone

concentration of X

ppb

AOTX ppm h The sum of the differences between the hourly

mean ozone concentration (in ppb) and X ppb when

the concentration exceeds X ppb during daylight

hours, accumulated over a stated time period. Units

of ppb and ppm are parts per billion (nmol mol-1

)

and parts per million (µmol mol-1

) respectively,

calculated on a volume/volume basis.

Concentration-based

critical level of ozone

CLec ppm h AOTX over a stated time period, above which

direct adverse effects on sensitive vegetation may

occur according to present knowledge.

Concentration

accumulated over a

threshold ozone

concentration of X

ppb modified by

vapour pressure deficit

(VPD)

AOTXVPD ppm h The sum of the differences between the hourly

mean ozone concentration (in ppb) modified by a

vapour pressure deficit factor ([O3]VPD), and X ppb

when the concentration exceeds X ppb during

daylight hours, accumulated over a stated time

period.

Terms for flux-based critical levels

Projected leaf area PLA m2 The projected leaf area is the total area of the sides

of the leaves that are projected towards the sun.

PLA is in contrast to the total leaf area, which

considers both sides of the leaves. For flat leaves

the total leaf area is simply 2*PLA.

Stomatal flux of ozone Fst nmol m-2

PLA s-1

Instantaneous flux of ozone through the stomatal

pores per unit projected leaf area (PLA). Fst can be

defined for any part of the plant, or the whole leaf

area of the plant, but for this manual, Fst refers

specifically to the sunlit leaves at the top of the

canopy. Fst is normally calculated from hourly

mean values and is regarded here as the hourly

mean flux of ozone through the stomata.

Stomatal flux of ozone

above a flux threshold

of Y nmol m-2

PLA s-1

Fst Y

nmol m-2

PLA s-1

Instantaneous flux of ozone above a flux threshold

of Y nmol m-2

s-1

, through the stomatal pores per

unit projected leaf area. FstY can be defined for any

part of the plant, or the whole leaf area of the plant,

but for this manual FstY refers specifically to the

sunlit leaves at the top of the canopy. FstY is

normally calculated from hourly mean values and is

regarded here as the hourly mean flux of ozone

through the stomata.

Accumulated stomatal

flux of ozone above a

flux threshold of Y

nmol m-2

PLA s-1

AFstY

mmol m-2

PLA Accumulated flux above a flux threshold of Y nmol

m-2

s-1

, accumulated over a stated time period

during daylight hours. Similar in concept to AOTX.

Flux-based critical

level of ozone

CLef mmol m-2

PLA Accumulated flux above a flux threshold of Y nmol

m-2

s-1

(AFstY), over a stated time period during

daylight hours, above which direct adverse effects

may occur on sensitive vegetation according to

present knowledge.

Table 3.4: Terminology for critical levels of ozone

Note: The recommendations of theGothenburg Workshop (2002), the 16th TaskForce Meeting of the ICP Vegetation and the17th Task Force Meeting of the ICPModelling and Mapping were:

For agricultural crops: use stomatal flux-based critical levels based on AFst6 if thenecessary inputs are available for a quantita-tive assessment of impacts on wheat andpotato, and the AOT40-based critical level toassess the risk of yield reduction if onlyozone concentration is available and a riskassessment for all crops is needed. TheAOT30VPD critical level should be used toassess the risk of visible ozone injury and

cannot be used to indicate the risk of yieldreduction.

For horticultural crops: use the AOT40-basedcritical level to assess the risk of effects onyield. The AOT30VPD critical level should beused to assess the risk of visible ozone injuryand cannot be used to indicate the risk ofyield reduction.For (semi-) natural vegetation: use theAOT40-based critical level.

For forest trees: use the AOT40-based criticallevel to assess the risk of growth reduction.The provisional stomatal flux-based criticallevel is provided for guidance only.



Approach Crops (Semi-) natural

vegetation

Forest trees

CLef

Wheat: An AFst6 of 1

mmol m-2

PLA

Potato: An AFst6 of 5

mmol m-2

PLA

Birch and beech:

Provisionally AFst1.6

of 4 mmol m-2

PLA

Time period Wheat: Either 970˚C

days, starting 270˚C

days before mid-

anthesis (flowering) or

55 days starting 15 days

before mid-anthesis

Potato: Either 1130˚C

days starting at plant

emergence or 70 days

starting at plant

emergence

One growing season

Stomatal flux-

based critical

level

Effect Yield reduction

Not available

Growth reduction

CLec

Agricultural crops: An

AOT40 of 3 ppm h

Horticultural crops: An

AOT40 of 6 ppm h

An AOT40 of 3

ppm h

An AOT40 of 5

ppm h

Time period Agricultural crops: 3

months

Horticultural crops: 3.5

months

3 months (or growing

season, if shorter)

Growing season

Concentration-

based critical

level

Effect Yield reduction for both

agricultural and

horticultural crops

Growth reduction in

perennial species and

growth reduction

and/or seed

production in annual

species

Growth reduction

CLec An AOT30VPD of 0.16

ppm h

Time period Preceding 8 days

VPD-modified

concentration-

based critical

level Effect Visible injury to leaves

Not available

Not available

Table 3.5: Critical levels for ozone. The methods for calculating each critical level are described in Sections 3.4 and 3.5.



Figure 3.1: A schematic diagram illustrating the steps involved in calculating exceedance of the flux-based andconcentration-based critical levels of ozone for agricultural and horticultural crops, (semi-) natural vegetation andforest trees.

3.3.1 Crops

3.3.1.1 Crop sensitivity to ozone

Table 3.6 provides an indication of the rela-tive sensitivity to ozone of a wide range ofagricultural and horticultural crops. Thistable was derived from a comprehensivereview of over 700 published papers on cropresponses to ozone leading to the derivationof response–functions for 19 crops (Mills etal., 2003). Data were included only whereozone conditions were recorded as 7h, 8h or24h means or AOT40, and exposure to ozoneoccurred for a whole growing season usingfield-based exposure systems. The yielddata presented in the published papersranged from “% of control treatment” to “tha-1” and were all converted to the yield rel-ative to that in the charcoal-filtered air treat-ment. Where the data were published as 7hor 8h means, data points were omitted fromthe analysis if the O3 concentration exceed-ed 100 ppb (considered to be outside thenormal range for Europe). Data were con-verted to AOT40 using a function derivedfrom the ICP Vegetation ambient ozonedatabase (Mills et al., 2003). Dose-responsefunctions were derived for each crop usinglinear regression.

The crops were ranked in sensitivity toozone by determining the AOT40 associatedwith a 5% reduction in yield. Wheat, pulses,cotton and soybean were the most sensitiveof the agricultural crops, with several horti-cultural crops such as tomato and lettuce

being of comparable sensitivity. Crops suchas potato and sugar beet that have greenfoliage throughout the summer months wereclassified as moderately sensitive to ozone.In contrast, important cereal crops such asmaize and barley can be considered to bemoderately resistant and insensitive toozone respectively.

3.3.1.2 Stomatal flux-based criticallevels for yield reduction inwheat and potato

At the workshop in Gothenburg, November2002, it was concluded that for the timebeing, it is only possible to derive flux-basedozone critical levels for the crops of wheatand potato. Ozone fluxes through the stom-ata of leaves found at the top of the canopyare calculated using a multiplicative algo-rithm based on the methodology describedby Emberson et al. (2000b). The stomatal fluxalgorithm used within the EMEP ozone dep-osition module is described in Section 3.4.4.

The index AFstY is used to quantify the flux ofozone through the stomata of the uppermostleaf level that is directly exposed to solarradiation and thus no calculation of lightexclusion, caused by the filtering of lightthrough the leaves of the canopy, is required.The sunlit leaf level has the largest gasexchange in terms of net photosynthesis (i.e.contributes most strongly to yield) andozone flux, both because it receives mostsolar radiation and because it is least senes-cent. Thus, the ozone flux is expressed asthe cumulative stomatal flux per unit sunlitleaf area in order to reflect the influence ofozone on the fraction of the leaf area whichis most important for yield.



3.3 Scientific bases of the critical levels for ozone

Sensitive Moderately sensitive Moderately resistant Insensitive

Cotton, Lettuce,

Pulses, Soybean,

Salad Onion,

Tomato, Turnip,

Watermelon, Wheat

Potato, Rapeseed,

Sugarbeet, Tobacco

Broccoli, Grape,

Maize, Rice

Barley, Fruit (plum

& strawberry)

Table 3.6: The range of sensitivity of agricultural and horticultural crops to ozone (see Mills et al., 2003 for response functions and definition of sensitivities)



In deriving relationships between the relativeyield and the stomatal flux of ozone in wheatand potato, it has been observed that thebest correlations between effect and accu-mulated stomatal flux are obtained whenusing a stomatal flux threshold (Y)(Danielsson et al., 2003; Pleijel et al., 2002).The strongest relationships between yieldeffects and AFstY were obtained using Y = 6nmol m-2 s-1 for both wheat and potato. Ineffect, this means that ozone exposure start-ed to contribute to AFstY at an ozone con-centration at the top of the crop canopy ofapproximately 22 ppb for wheat and atapproximately 14 ppb for potato if the stom-atal conductance was at its maximum. In thecase of lower conductance, which prevails inmost situations, a higher ozone concentra-tion than 22 ppb and 14 ppb is required tocontribute to AFst6 for wheat and potato,respectively. The threshold of 6 provides thehighest r² value for the relationship betweenyield reduction and AFstY for all of the thresh-olds tested for both crops. For example, forwheat the r² values were 0.51, 0.74, 0.83 and0.77 for Y= 0, 3, 6 and 9 respectively, whilstfor potato, the r² values were 0.6, 0.72, 0.76and 0.75 for Y= 0, 3, 6 and 9 respectively.

Based on the combination of data from anumber of open-top chamber experimentswith field-grown crops performed in severalEuropean countries, relationships betweenrelative yield (RY) and stomatal ozone flux(Fst) have been derived using the principlesintroduced by Fuhrer (1994) to calculate rel-ative yield. A relative yield of 1 representsthe absence of ozone effects. In the case ofwheat, thirteen experiments with field-grownwheat exposed to different ozone levels inopen-top chambers from four different coun-tries (Belgium, Finland, Italy and Sweden),representing five cultivars (Minaret, Dragon,Drabant, Satu and Duilio) were used, and forpotato seven experiments from four differentcountries (Belgium, Denmark, Finland, andSweden) were included, representing onecommonly grown cultivar (Bintje) (Pleijel etal., 2003).

In order to derive response relationshipswhich are consistent with the ozone deposi-

tion module of the EMEP model (Embersonet al., 2000b), the parameterisations of theconductance model presented in Pleijel et al.(2002, 2003) and Danielsson et al. (2003)were revised to achieve a full compatibilitywith the EMEP model calibration. The result-ing algorithms and their parameterisation arepresented in Sections 3.4.4 and 3.4.5. Thisrevision resulted in stronger correlationsbetween relative yield and AFstY for bothwheat (a change in r² from 0.77 to 0.83) andpotato (a change in r² from 0.64 to 0.76),compared to the relationships presented inPleijel et al. (2003).

The response relationship for wheat:

(3.1)

is presented in Figure 3.2.

The response relationship for potato:

(3.2)

is presented in Figure 3.3.

In line with earlier concepts used for cropcritical levels, 5% yield reduction was usedas the loss criterion for the identification ofstomatal flux-based critical levels (UNECE,1996). For wheat the suggested stomatalflux-based critical level for 5% yield loss ofan AFst6 of 1 mmol m-2 was statistically signif-icant according to the confidence limits ofthe yield response regressions, which is animportant criterion when using a yield losslevel such as 5% (Pleijel, 1996). For potato,the value was rounded upwards from 4.6mmol O3 m-2 (representing 5% yield loss) to 5mmol O3 m-2.

As such, the flux-based critical level, CLef ofozone for wheat is an AFst6 of 1 mmol O3 perunit projected flag leaf area, accumulatedduring an effective temperature sum period

RYwheat = 1.00 – (0.048*AFst6)

RYpotato = 1.01 – (0.013*AFst6)



Figure 3.2: The relationship between the relative yield of wheat and AFst6 for the wheat flag leaf based on fivewheat cultivars from four European countries using effective temperature sum to describe phenology.

Figure 3.3: The relationship between the relative yield of potato and the AFst6 for sunlit leaves based on data fromfour European countries and using effective temperature sum to describe phenology.



starting 270°C days before anthesis (flower-ing) and ending 700°C days after anthesis.The total period is 970°C days (base temper-ature 0°C). On average, the 970°C days cor-responded to 54 days in the experimentsused to calibrate the function.

The flux-based critical level, CLef, of ozonefor potato is an AFst6 of 5 mmol O3 per m² pro-jected, sunlit leaf area accumulated over1130°C days starting at plant emergence(base temperature 0°C). On average, the1130°C days corresponded to 66 days in theexperiments used to calibrate the function;this value has been rounded up to 70 daysfor Table 3.5.

3.3.1.3 AOTX-based critical levels for crop yield reduction

Agricultural crops

The concentration-based critical level foragricultural crops has been derived from alinear relationship between AOT40 and rela-tive yield for wheat, developed from theresults of open-top chamber experiments

conducted in Europe and the USA (Figure3.4). Newly published data (Gelang et al.2000) has been added to that derived by Fuhrer et al. (1997) and quoted in the previ-ous version of the Mapping Manual (UNECE,1996). Thus, the critical level for wheat isbased on a comprehensive dataset including9 cultivars. The AOT40 corresponding to a5% reduction in yield is 3.3 ppm h (95%Confidence Interval range 2.3-4.4 ppm h).This value has been rounded down to 3 ppm h for the critical level. The critical levelfor agricultural crops is only applicable whennutrient supply and soil moisture are notlimiting, the latter because of sufficient precipitation or irrigation (Fuhrer, 1995).

The time period over which the AOT40 is cal-culated should be three months and the tim-ing should reflect the period of active growthof wheat and be centred around the start ofanthesis (see Section 3.5.2.2 for guidance).

An optional additional AOT30-based criticallevel of ozone has also been derived for agri-cultural crops, based on a re-working of thewheat response-function data used by

Figure 3.4: Wheat yield-response function used to derive the concentration-based critical levels for agriculturalcrops (r² = 0.89) (Fuhrer et al., 1997, Gelang et al., 2000).

Fuhrer et al., 1997 using AOT30 as the doseparameter (r² = 0.90, data not presented).The value for this critical level is an AOT30 of4 ppm h applied to the same time-windowsas described for AOT40. Following discus-sions at the Gothenburg Workshop, the 16th Task Force Meeting of the ICP Vegetationand the 19th Task Force Meeting of the ICPModelling and Mapping, it was concludedthat AOT40 should continue to be used forthe concentration-based critical level foragricultural crops, but that AOT30 could beused in integrated assessment modelling onthe European scale if this considerablyreduces uncertainty in the overall integratedassessment model.

It is not recommended that exceedance ofthe concentration-based critical level foragricultural crops is converted into econom-ic loss; it should only be used as an indica-tion of ecological risk (Fuhrer, 1995).

Horticultural Crops

A concentration-based critical level hasbeen derived for horticultural crops that aregrowing with adequate nutrient and watersupply. An AOT40 of 6.02 ppm h is equivalentto a 5% reduction in fruit yield for tomato,and has been derived from a dose-responsefunction (Figure 3.5) developed from a com-prehensive dataset including 14 cultivars (r² = 0.48, p<0.001). This value has beenrounded down to 6 ppm h for the criticallevel. Although statistical analysis has indi-cated that water melon may be more sensi-tive to ozone than tomato, the dataset forwater melon is not sufficiently robust for usein the derivation of a critical level becausethe data is only for one cultivar (Mills et al.,2003). Tomato is considered suitable for thederivation of the critical level since it is classified as an ozone-sensitive crop (Table3.6). The data used in the derivation of thecritical level for horticultural crops is fromexperiments conducted in the USA(California and North Carolina), Germany andSpain. The time period for accumulation ofAOT40 is 3.5 months, starting from the emer-gence of the crop (see Section 3.5.3.2 forguidance).

It is not recommended that the exceedanceof the concentration-based critical level forhorticultural crops is converted into eco-nomic loss; it should only be used as an indi-cation of ecological risk during the mostsensitive environmental conditions (Fuhrer,1995).





3.3.1.4 VPD-modified AOT30 – based critical level for visible injury on crops

Note: This critical level was revised followingnew analysis conducted after the 17th TaskForce meeting of the ICP Vegetation(Kalamata, February 2004); participants wereasked by email if they objected to thechanges made and no objections wereraised.

Acute visible ozone injury, resulting fromshort-term ozone exposure, represents themost direct evidence of the harmfulness ofelevated ozone concentrations. The aim ofthis short-term critical level is to reflect therisk for this type of injury. For some horticul-tural crops, such as spinach, lettuce, saladonion and chicory sold for their foliage, visi-ble ozone injury can cause significant finan-cial loss to farmers. In addition, visible ozoneinjury can be easily demonstrated and thusused to highlight the problem of phytotoxicozone to a broad audience (Klumpp et al.,2002).

The short-term critical level is based onresults from experiments performed with

subterranean clover (Trifolium subterra-neum) which was used as the key bioindica-tor plant for a number of years within the ICPVegetation (Benton et al., 1995). Data fromthree participating countries (Sweden,Belgium and Austria) were used to derive acommon relationship between ozone expo-sure and the risk for visible injury (PihlKarlsson et al., 2004). Since Trifolium subter-raneum is well-documented as a very ozonesensitive plant in terms of having visiblesymptoms after ozone exposure, the criticallevel for visible injury on crops is expected toprotect other ozone sensitive plants fromvisible injury.

It has been shown that AOT30 is the bestAOTX exposure index to describe the risk forvisible ozone injury in subterranean cloverunder low VPD, i.e. relatively humid condi-tions (Pihl Karlsson et al., 2003). However, ithas also been demonstrated that in drier climates VPD is a very important modifier ofozone uptake through its limiting effect onstomatal conductance and thus of the riskthat a certain ozone concentration wouldcontribute to visible injury (Ribas &Peñuelas, 2003). A VPD modified AOT30

Figure 3.5: Tomato yield-response function used to derive the concentration-based critical levels for horticulturalcrops (r² = 0.48, p<0.001) (dotted line = 95% confidence limits). (Hassan et al., 1999; McLean & Schneider, 1976;Reinert et al., 1997; Temple et al., 1985; Temple, 1990; Bermejo, 2002; Calvo, 2003).

(AOT30VPD) approach adequately describesthe relationship between ozone exposureand risk for visible injury in subterraneanclover when grown in well-watered condi-tions (as in the ICP Vegetation experiments).

The critical level for visible injury is exceed-ed when the AOT30VPD during daylight hoursover eight days exceeds 0.16 ppm h. Thisrepresents a significant risk of having visibleozone injury on at least 10% ( ± 3.5%according to the 99% confidence limits ofthe regression shown in Figure 3.6) of theleaves on sensitive plants, such as subter-ranean clover. The 10% level for visibleozone injury was chosen based on the con-clusion from the studies by Pihl Karlsson etal. (2003).

The AOT30VPD index accumulated duringdaylight hours, using an exposure period ofeight days explained 60% of the variation ofthe observed extent of visible injury (% injured leaves) accumulated during daylight hours (p<0.001 for the slope andintercept). The relationship between visibleinjury and AOT30VPD during eight days

before obser-vation of visible injury is shownin Figure 3.6. It is not recommended that the exceedanceof the concentration-based critical level forvisible injury on agricultural and horticulturalcrops is converted into economic loss; itshould only be used as an indication of riskof injury during the most sensitive environ-mental conditions.



Figure 3.6: The extent of visible injury (percentage ozone injured leaves) versus AOT30VPD. The accumulation peri-od was eight days before observation of visible injury. The accumulation was made during daylight hours.Confidence limits for p = 0.99 are also presented (Pihl Karlsson et al., 2004).



3.3.2 (Semi-) natural vegetation

The critical level for (semi-) natural vegeta-tion is concentration based; flux-basedmethods are not sufficiently advanced forthis vegetation type for inclusion in this man-ual.

The criteria for adverse effects on (semi-)natural vegetation are an effect on growth forperennial species, and effects on growth andseed production for annual species. The crit-ical level is based on statistically significanteffects or growth reductions of greater than10% on sensitive taxa of grassland and fieldmargin communities and is applicable to allsensitive semi-natural vegetation and natu-ral vegetation, described here collectively as(semi-) natural vegetation (forest trees andwoodlands are not included here as (semi-)natural vegetation). The value of 3 ppm h issufficient to protect the most sensitive annu-al and short-lived perennial species whengrown in a competitive environment.

In contrast to crops and tree species, only

limited experimental data are available for asmall proportion of the vast range of speciesfound across Europe. This means that analy-sis of exposure-response data for individualspecies to derive a critical level value is moredifficult. Instead, the recommended criticallevel is based on data from a limited numberof sensitive species. The value of 3 ppm hwas originally proposed at the Kuopio work-shop (Kärenlampi & Skärby, 1996) and con-firmed at the Gerzensee workshop (Fuhrer &Achermann, 1999). At the time of the Kuopioworkshop, no exposure-response studieswere available for derivation of the criticallevel for (semi-) natural vegetation, based ona 10% response. Instead, data from field-based experiments with control and ozonetreatments were used to identify studiesshowing significant effects at relatively lowozone exposures. Table 3.7 summarises thekey field chamber and field fumigationexperiments which supported the originalproposal of this critical level for (semi-) natu-ral vegetation.

Species or

community

Most sensitive

species

Ozone

(AOT40, ppm h)

Response Reference

Individual plants Solanum nigrum

Malva sylvestris

4.2

3.9

-23%; shoot

mass

-54%; seed

mass

Bergmann et

al., 1996

Mesocosms of

four species

Trifolium repens 5.0 -13%; shoot

mass

Ashmore &

Ainsworth,

1995

Mesocosms of

seven species

Festuca ovina

Leontodon

hispidus

7.0

7.0

-32%; shoot

mass

-22%; shoot

mass

Ashmore et al.,

1996

Ryegrass-clover

sward

Trifolium repens 5.0 -20%; shoot

mass

Nussbaum et

al., 1995

Table 3.7: Summary of key experiments supporting the critical level of 3ppm h, as proposed at the Kuopio work-shop (Ashmore & Davison, 1996)

A number of studies have clearly demon-strated that the effects of ozone in speciesmixtures may be greater than those onspecies grown alone or only subject tointraspecific competition. Therefore, the crit-ical level needs to take into account the pos-sibility of effects of interspecific competitionin reducing the threshold for significanteffects; indeed three of the four experimentslisted in Table 3.7 include such competitiveeffects. The most comprehensive study ofozone effects on species mixtures involvingspecies which are representative of differentcommunities across Europe, is the EU-FP5BIOSTRESS (BIOdiversity in HerbaceousSemi-Natural Ecosystems Under STRESS byGlobal Change Components) programme.Results to date from the BIOSTRESS pro-gramme, including experiments with speciesfrom the Mediterranean dehesa community,indicate that exposures to ozone exceedingan AOT40 of around 3 ppm h may cause sig-nificant negative effects on annual and

perennial plant species (see Fuhrer et al.,2003). The BIOSTRESS mesocosm experi-ments with two-species mixtures indicatedthat exposures during only 4-6 weeks earlyin the growing season may cause shifts inspecies balance. The effect of this earlystress may last for the rest of the growingseason. This new evidence meant that theproposal of a six month growth period forperennial species adopted at the GerzenseeWorkshop was no longer supported at theGothenburg Workshop, and that the time-window for both annuals and perennials isnow set to three months.

The key experiments from the BIOSTRESSprogramme which support the proposedcritical level are summarised in Table 3.8.Taken together, these studies support a critical level in the range 2.5-4.5 ppm h, witha mean value of 3.3 ppm h.



Responsive

species

Competitor

species

Variable showing

significant response

Corresponding

AOT40

Reference

Trifolium

pratense

Poa pratensis Biomass (-10%) 4.4 ppm h* Gillespie &

Barnes,

unpublished data

Veronica

chamaedrys

Poa pratensis Species biomass

ratio

3.6 ppm h Bender et al.

(2002)

Trifolium

cherleri, T.

striatum

Briza maxima Flower production 2.2-2.7 ppm h Gimeno et al.

(2003a); Gimeno

et al. (2004).

Trifolium

cherleri

Briza maxima Seed output 2.4 ppm h Gimeno et al.

(2004)

Table 3.8: Summary of experiments from the BIOSTRESS programme which support the recommendedcritical levels (reviewed by Fuhrer et al., 2003)

*estimated from exposure-response functions



The value of the critical level is further sup-ported by more recent published data, forinstance, for wetland species by Power andAshmore (2002) and Franzaring et al. (2000).The latter authors observed a significantreduction in the shoot:root ratio in Cirsiumdissectum at 3.3 ppm h after 28 days ofexposure. In individual plants from wildstrawberry populations growing at high lati-tudes, Manninen et al. (2003) observed a sig-nificant biomass decline of >10% at 5 ppm hfrom June-August.

Guidance on the time-window for calculatingAOT40, for mapping (semi-) natural commu-nities at risk of ozone and determining theozone concentration at the canopy height isprovided in Section 3.5.5.

3.3.3 Forest trees

3.3.3.1 Provisional flux-based critical levels

At the UNECE workshop in Gothenburg inNovember 2002 (Karlsson et al., 2003a) itwas concluded that the effective ozonedose, based on the flux of ozone into theleaves through the stomatal pores, repre-sents the most appropriate approach for setting future ozone critical levels for foresttrees. However, uncertainties in the develop-ment and application of flux-basedapproaches to setting critical levels for forest trees are at present too large to justifytheir application as a standard risk assess-ment method at a European scale. AlthoughAFstY is much more physiologically relevantthan AOTX, more time and data are neededbefore AFstY-response relationships for treescould be considered sufficiently robust forestablishing a stomatal flux-based criticallevel of ozone for forest trees; calculation ofexceedance of a provisional flux-based critical level for beech and birch is describedin Section 3.4.6. Although model parameter-isation and flux-response relationships areavailable for other species (Karlsson et al.,2003b), parameterisation of the multiplica-tive model is given only for beech and birch,and it is intended that these species areused as indicators of risk to sensitive recep-

tors. The choice of beech and birch alsoensures continuity with the concentration-based critical level described below.

As for crop species, ozone flux is estimatedon a projected leaf area basis for sunlitupper canopy leaves. Ozone flux should beaccumulated for an exposure window asdefined in Section 3.4.6, either using thedefault values given, locally derived informa-tion or phenological models. Ozone concen-trations used for the calculation of stomatalozone flux are those at the top of the canopy,as discussed in Section 3.4.6.1. As is thecase for both wheat and potato, a fluxthreshold, Y, should be applied. For beechand birch, a value of 1.6 nmol m-2 PLA s-1 is proposed for Y, based on the statisticalanalysis of flux-response relationships for avariety of sensitive deciduous tree species(Karlsson et al., 2003b, 2004). This sameanalysis concluded that for Y=1.6 nmol m-2 s-1,a growing season cumulative stomatal flux(AFst1.6) of 4 mmol m-2 was statistically signif-icant (99% confidence) and represented a5% reduction in biomass under experimentalconditions. This indicative value representsthe provisional flux-based critical level,above which there is a risk of a negativeimpact of ozone on growth for the most sen-sitive tree species. The dataset on which thisanalysis is based, is identical to that fromwhich the AOT40-based critical level wasderived, and represents a combination ofdata for both birch and beech.

It should be emphasised that this provision-al critical level should be viewed as a firststep in the derivation of new critical levelswhich could be used to make quantitativeassessments of the ozone impacts onmature forest trees since the value currentlyselected is based on a number of assump-tions (these are discussed in more detail inKarlsson et al., 2003b, 2004). In the future itis envisaged that additional analysis andreworking of existing experimental data andimprovement of stomatal conductance mod-els for forest trees should provide betterestimates of both cumulative stomatal fluxand associated flux thresholds and criticallevels to provide more robust estimates ofdamage for forest trees.

3.3.3.2 AOTX-based critical levels for forest trees

The AOTX approach is retained as the recommended method for calculating criticallevels for forest trees until the stomatal fluxapproach becomes sufficiently refined.However, the methodology and critical levelshave been substantially revised since theprevious version of this Manual was published (UNECE, 1996).

The experimental database that was presented at the UNECE Workshop inGothenburg 2002 has been re-analysed andexpanded to include additional correlationswith AOT20, AOT30, and AOT50 (Karlsson et al., 2003b,). Furthermore, the tree speciesincluded in the analysis have been separated into four species categories(Table 3.9) based on the sensitivity of growthresponses to ozone. It should be empha-sised that this categorisation is based ongrowth as a measure of effect, and that therelative sensitivity of a given species maydiffer when an alternative measure such asvisible injury is used. As a result of this differ-entiation of species, linear regressionsbetween exposure and response have thehighest r² values and there are no significantintercepts (Table 3.10). The disadvantagewas that there are insufficient data availablefor the moderately sensitive coniferous cate-gory for adequate statistical analysis. Thismay change in the future, if more datasetsbecome available.

Using the sensitivity categories describedabove, AOT40 gave the highest r² values ofthe AOTX indices tested (Figure 3.6).However, the difference between the r² val-ues for AOT40 and AOT30 was small (0.62and 0.61 respectively for the combined birchand beech dataset, Table 3.10).



Ozone-sensitive species Moderately ozone-sensitive species

Deciduous Coniferous Deciduous Coniferous

Fagus sylvatica

Betula pendula

Picea abies

Pinus sylvestris

Quercus petrea,

Quercus robur

Pinus halepensis

Table 3.9: Sensitivity classes for the tree species based on effects of ozone on growth



Based on the analysis described above, theconcentration-based critical level of ozonefor forest trees, CLec, has been reduced froman AOT40 value of 10 ppm h (Kärenlampi &Skärby, 1996) to 5 ppm h (range 1-9 ppm h,determined by the 99% confidence inter-vals), accumulated over one growing season(Figure 3.7). This value of 5 ppm h is associ-ated with a 5% growth reduction per growing season for the deciduous sensitivetree species category (beech and birch,Figure 3.7). The 5% growth reduction wasclearly significant as judged by the 99% confidence intervals in Figure 3.7. This

increase in the robustness of the datasetand the critical level represents a substantialimprovement compared to the 10% growthreduction associated with the previousozone critical level of an AOT40 of 10 ppm h(Kärenlampi & Skärby, 1996). Furthermore, itrepresents a continued use of sensitive,deciduous tree species to represent themost sensitive species under most sensitiveconditions. As previously, it should bestrongly emphasized that these valuesshould not be used to quantify ozoneimpacts for forest trees under field condi-tions.

Ozone index/ plant category Linear regression

r2 p for the

slope

p for the

intercept

slope

AOT20

Birch, beech 0.52 <0.01 0.70 - 0.357

Oak 0.57 <0.01 0.73 - 0.142

Norway spruce, Scots pine 0.73 <0.01 0.31 - 0.086

AOT30

Birch, beech 0.61 <0.01 0.63 - 0.494

Oak 0.61 <0.01 0.79 - 0.170


AOT40

Birch, beech 0.62 <0.01 0.31 - 0.732

Oak 0.65 <0.01 0.73 - 0.216


AOT50

Birch, beech 0.53 <0.01 0.05 - 1.033

Oak 0.62 <0.01 0.82 - 0.248


Table 3.10: Statistical data for regression analysis of the relationship between AOTX ozone exposure indices (in ppm h) and percentage reduction of total and above-ground biomass for different tree species categories

Additional evidence to support the newozone critical levels for forest trees comesfrom the observation of visible injury inyoung trees in ambient air at Lattecaldo, insouthern Switzerland. It was concluded thata reduction of the ozone critical level to 5ppm h AOT40 was needed in order to protectthe most sensitive species from visible injury(Van der Hayden et al., 2001, Novak et al.,2003). Furthermore, Baumgarten et al. (2000)detected visible injury on the leaves ofmature beech trees in Bavaria well below 10ppm h AOT40.

An optional additional AOT30-based criticallevel of ozone has also been derived for for-est trees based on the response function forbirch and beech. The value for this criticallevel is an AOT30 of 9 ppm h applied to thesame time-windows as described for AOT40.Following discussions at the GothenburgWorkshop, the 16th Task Force Meeting ofthe ICP Vegetation and the 19th Task ForceMeeting of the ICP Modelling and Mapping,it was concluded that AOT40 should be usedfor the concentration-based critical level forforest trees, but that AOT30 could be used inintegrated assessment modelling on theEuropean scale if this considerably reducesuncertainty in the overall integrated assess-ment model.

3.4.1 Stages in calculating AFstY and CLef exceedance

To calculate the relevant AFstY to sunlitleaves and exceedance of the stomatal flux-based critical level, the following steps haveto be taken:

1. Hourly ozone concentrations at the top of the canopy are determined (see Section 3.4.2 for conversion of concen-trations considering ozone concentra-tion gradients above the canopy).

2. The time period to consider is deter-mined as described in Section 3.4.3.The hourly stomatal conductance (gsto) values for the relevant periods, for sunlit leaves of the receptor plant are identified using the algorithm pre-sented in Eq. 3.12.

3. Every hourly stomatal conductance thus identified is multiplied by the cor-responding hourly ozone concentra-



Figure 3.7: The relationship between percentage reduction in biomass and AOT40, on an annual basis, for thedeciduous, sensitive tree species category, represented by beech and birch. The relationship was analysed by linear regression with 99% confidence intervals. Explanations for the figure legends can be found in Karlsson et al. (2003b).

3.4 Calculating exceedance ofstomatal flux-based critical levelsfor ozone



tion at the top of the canopy, resultingin hourly stomatal fluxes of ozone, Fstexpressed in nmol m-2 PLA s-1 (Eq. 3.9).

4. The value Y is subtracted from each hourly Fst value, and then multiplied by 3600 to obtain hourly FstY values in nmol O3 m-2 PLA h-1.

5. The sum of all hourly FstY values is cal-culated for the specified accumulated period. The resulting value is AFstY in units nmol m-2 PLA.

6. If the AFstY value is larger than the flux-based critical level for ozone CLef, there is exceedance of the critical level.

3.4.2 Ozone concentrations at the canopy height

Note: Sources of ozone concentration esti-mates and their spatial interpolation are con-sidered in Chapter 2 of this manual.

All ozone indices described in this chapterare based on ozone concentrations at thetop of the canopy. For crops and other lowvegetation, canopy-top concentrations maybe significantly lower than those at conven-tional measurement heights of 2-5 m abovethe ground, and hence use of measured datadirectly or after spatial interpolation maylead to significant over-estimates of ozoneconcentrations and hence of the degree ofexceedance of CLec and CLef. In contrast, forforests, measured data may underestimateozone concentration at the top of thecanopy. The difference between measure-ment height and canopy height is a functionof several factors, including wind speed andother meteorological factors, canopy heightand the total flux of ozone, Ftot.

Conversion of ozone concentration at meas-urement height to that at canopy-top height(z1) can be best achieved with an appropri-ate deposition model. It should be noted,however, that the flux-gradient relationshipsthese models depend on are not strictly validwithin the roughness sub-layer (ground level

to 2-3 times canopy height), so even suchdetailed calculations can provide onlyapproximate answers. The model chosenwill depend upon the amount of meteorolog-ical data that is available. Two simple meth-ods are included here which can be used toachieve the necessary conversion if (a) nometeorological data are available, or (b)some basic measurements are available.

Method (a): Tabulated gradient

If no meteorological data are available at all,then a simple tabulation of O3 gradients canbe used. The relationship between O3 con-centrations at a number of different heightshas been estimated with the EMEP deposi-tion module (Emberson et al., 2000a), usingmeteorology from about 30 sites acrossEurope. Data were produced for an arbitrarycrop surface and for short grasslands. Forthe crop surface, the assumptions madehere are that we have a 1 m high crop withgmax = 450 mmol O3 m-2 PLA s-1. The total leafsurface area index (LAI) is set to 5 m² PLA /m²,and the green LAI is set at 3 m² PLA /m²,assumed to give a canopy-scale phenologyfactor (fphen) of 0.6. The soil moisture factor(fSWP) is set to 1.0. Constant values of theseparameters are used throughout the year inorder to avoid problems with trying to esti-mate growth-stage in different areas ofEurope. The concentration gradients thusderived are most appropriate to a fully devel-oped crop but will serve as a reasonableapproximation for the whole growing sea-son. Other stomatal conductance modifiersare allowed to vary according to the wheat-functions. For short grasslands, canopyheight was set to 0.1 m, gmax to 270 mmol O3 m-2 PLA s-1 and fSWP set to 1.0. Allother factors are as given for grasslands inEmberson et al. (2000b). For the micromete-orology, the displacement height (d) androughness length (z0) are set to 0.7 and 0.1 ofcanopy height (z1), respectively. Thus, theupper-boundary of the quasi-laminar layer(d+z0), is simply 0.8z1, where z1 is canopyheight.

Table 3.11 shows the average relationshipbetween O3 concentrations at selectedheights, derived from runs of the EMEPmodule over May-July, and selecting thenoontime factors as representative of day-time multipliers. O3 concentrations are nor-malised by setting the 20 m value to 1.0. Touse Table 3.11, measurements made abovecrops or grasslands may simply be extrapo-lated downwards to the canopy top for therespective vegetation. For example, with 30ppb measured at 3 m height (above groundlevel) in a crop field, the concentration at 1mwould be 30.0 * (0.88/0.95) = 27.8 ppb. Forshort grasslands we would obtain

30.0 * (0.74/0.96) = 23.1 ppb at canopyheight, 0.1 m. Experiments have shown thatthe vertical gradients found above for cropsalso apply well to tall (0.5 m) grasslands.Some judgement may then be required tochoose values appropriate to different vege-tation types.



O3 concentration gradient Height (m)

Crops (where z1=1 m,

gmax= 450 mmol O3 m-2

PLA s-1

)

Short Grasslands (where z1=0.1 m,

gmax=270 mmol O3 m-2

PLA s-1

)

20 1.0 1.0

10 0.99 0.99

5 0.97 0.97

4 0.96 0.97

3 0.95 0.96

2 0.93 0.95

1 0.88 0.92

0.5 0.81* 0.89

0.2 - 0.83

0.1 - 0.74

Table 3.11: Representative O3 gradients above artificial (1 m) crop, and short grasslands (0.1 m). O3 concentrationsare normalised by setting the 20 m value to 1.0. These gradients are derived from noontime factors and are intended for daytime use only.

*0.5 m is below the displacement height of crops, but may be used for taller grasslands, see text.



For forests, ozone concentrations mustoften be derived from measurements madeover grassy areas or other land-cover types.In principal, the O3 concentration measuredover land-use X (e.g. short grasslands) couldbe used to estimate the O3 concentration ata reference height, and then the gradientprofile appropriate for desired land use Ycould be applied. However, in order to keepthis simple methodology manageable, and inview of the uncertainties inherent in makinguse of any profile near the canopy itself, it issuggested that concentrations are estimatedby extrapolating the profiles given in Table3.11 upwards to the canopy height forforests. As an example, if we measure 30 ppb at 3 m above short grassland, theconcentration at 20 m is estimated to be30.0*(1.0/0.96) = 31.3 ppb.

It should be noted that the profiles shown inTable 3.11 are representative only, and thatsite-specific calculations would providesomewhat different numbers. However, with-out local meteorology and the use of a dep-osition model, the suggested procedureshould give an acceptable level of accuracyfor most purposes. Method (b): Use of neutral stability profiles

If we have wind speed, u (m s-1) at referenceheight zR, and an estimate of z0, then we findconcentration values appropriate to a heightz1 (e.g. 1 m) by making use of the constant-flux assumption and definition of aerody-namic resistance:

(3.3)

Where Vg(zR) is the deposition velocity (m s-1) at height zR, and Ra(zR,z1) is the aero-dynamic resistance between the twoheights (s m-1). Re-arranging the second twoterms, we get:

(3.4)

In neutral stability, friction velocity (u*) andRa are easily obtained:

(3.5)

(3.6)

where the von Kármán konstant, k = 0.4

The deposition velocity requires furtherinformation:

(3.7)

Note: Ra is here the aerodynamic resistancefrom zR to z0 (the level where Ra becomeszero), not to z1 (which is any height fairly nearthe ground, for example the height of the topof the canopy). For ozone, Rb = 6.85/u*. Thecanopy resistance, Rc, is a function of tem-perature, radiation, relative humidity and soilwater. If local meteorology allows an assess-ment of these, the formulation of Rc may bedirectly estimated using a canopy-scalestomatal flux algorithm (see Emberson et al.,2000b).

3.4.3 Accumulation period

It is important that the cumulative periodover which the AFstY value is calculated isconsistent with the period when the relevantcrop or forest species is actively growingand most sensitive to the absorbed ozonedose. Thus, receptor specific time periods

Total flux = Vg(zR). C(zR)

C(z1) = C(zR) . [1 – (Ra(zR,z1) . Vg(zR))]

−=

0

ln

)(*

z

dz

kzuu

−−=

dz

dz

kuR R

zza R

1

),( ln*

11

Vg(zR) = 1/( Ra(zR,z0) + Rb + Rc )

) = (C(zR)- C(z1)) / Ra(zR,z1)

Ra(zR,z1)

are defined in the appropriate sectionsbelow. For flux-based critical levels this isachieved by identifying, within the growingseason, the start (Astart) and end (Aend) of theaccumulation period by receptor type. Thefollowing sources of information can be con-sidered for determining the accumulationperiod:-

a) Information from local or nationalagricultural or forest experts

b) Phenological models. These pre-dict the start and end of the growingseason, typically defined as a func-tion of climatic factors such as cumu-lative temperature and water avail-ability.

3.4.4 The stomatal flux algorithm

The estimation of stomatal flux of ozone (Fst)is based on the assumption that the concen-tration of ozone at the top of the canopy represents a reasonable estimate of the concentration at the upper surface of thelaminar layer near the flag leaf (in the case ofwheat) and the sunlit upper canopy leaves(in the case of potato and trees). If c(z1) is the concentration of ozone atcanopy top (height z1, unit: m), in nmol m-3,then Fst (nmol m-2 PLA s-1), is given by:

(3.8)

The 1/(rb+rc) term represents the depositionrate to the leaf through resistances rb (quasi-laminar resistance) and rc (leaf surfaceresistance). The fraction of this ozone takenup by the stomata is given by gsto/(gsto+gext),where gsto is the stomatal conductance, andgext is the external leaf, or cuticular, resist-ance. As the leaf surface resistance, rc, isgiven by rc = 1/(gsto+gext), we can also writeEq. 3.8 as:

(3.9)

A value for gext has been chosen to keepconsistency with the EMEP deposition mod-ules “big-leaf” external resistance, Rext = 2500/SAI, where SAI is the surface areaindex (green + senescent LAI). Assumingthat SAI can be simply scaled:

(3.10)

In order to be used correctly in Eq. 3.8 and3.9, gsto from Eq. 3.12 has to be convertedfrom units mmol m-2 s-1 to units m s-1. At normaltemperatures and air pressure, the conver-sion is made by dividing the conductancevalue expressed in mmol m-2 s-1 by 41000 togive conductance in m s-1.

Consistency of the quasi-laminar boundarylayer is harder to achieve, so the use of aleaf-level rb term (McNaughton & van derHink, 1995) is suggested, making use of thecross-wind leaf dimension L (unit: m) and thewind speed at height z1, u(z1):

(3.11)

Where the factor 1.3 accounts for the differ-ences in diffusivity between heat and ozone.Potential values of L for wheat, potato andbeech/birch are 0.02 m, 0.04 m and 0.05 mrespectively.

The core of the leaf ozone flux model is thestomatal conductance (gsto) multiplicativealgorithm which has been developed overthe past few years as described in Embersonet al. (2000b) and incorporated within theEMEP ozone deposition module (Embersonet al. 2000a). The multiplicative algorithmhas the following formulation:



extsto

sto

cb

stgg

g

rrzcF

++= *

1*)( 1

cb

cstost

rr

rgzcF

+= **)( 1

gext = 1/2500 (m s-1

)

rb = 1.3 * 150 * )( 1zu

L (s m

-1)



(3.12)

Where gsto is the actual stomatal conduc-tance (mmol O3 m-2 PLA s-1) and gmax is thespecies-specific maximum stomatal con-ductance (mmol O3 m-2 PLA s-1). The parame-ters fphen, fO3

, flight, ftemp, fVPD and fSWP are allexpressed in relative terms (i.e. they takevalues between 0 and 1) as a proportionof gmax.

These parameters allow for the modifyinginfluence of phenology and ozone, and fourenvironmental variables (light (irradiance),temperature, atmospheric vapour pressuredeficit (VPD) and soil water potential (SWP))on stomatal conductance to be estimated.Figure 3.8 shows the derivation of thesefunctions and sources of data for parameter-isation are provided in Table 3.12 and 3.13(at the end of the chapter). The part of Eq. 3.12 related to fphen and fO3

is a most limiting factor approach. Either senescencedue to normal ageing is limiting or the premature senescence induced by ozone islimiting. Early in the growing season and atlow ozone exposure fphen is always limitingand fO3

then does not come into operation.The original parameterisations given inEmberson et al. (2000b) have been revisedrecently based on data collected from the lit-erature and from ozone exposure experi-ments conducted in Sweden for wheat(Danielsson et al., 2003) and from a numberof sites across Europe for potato (Pleijel et al., 2003).

To account for the effect by transpiration onleaf water potential, which may lead to a lim-itation of stomatal conductance in the after-noon hours, an additional SVPD algorithm isincluded (Eq. 3.18). This may result in astronger limitation of stomatal conductancethan that suggested by Eq. 3.12 which repre-sents the original formulation given inEmberson et al. (2000b). Receptor specificparameterisations are provided for wheatand potato in Table 3.15 (see Section 3.4.5.3)

and provisionally for birch/beech in Table3.16 (see Section 3.4.6.2).

The range of gmax values for different wheatand potato cultivars grown in Europe, andadditional data for potato cultuvars grown inthe USA. The horizontal line represents thechosen gmax values (450 mmol m-2 s-1 for wheat,(average of observations: 455 mmol m-2 s-1);750 mmol m-2 s-1 for potato (average of obser-vations 738 mmol m-2 s-1)). gmax values areexpressed on a projected leaf area basis.

gsto = gmax *[min(fphen, fO3)]* flight *

max{fmin, (ftemp * fVPD * fSWP)}



Figure 3.8: Parameterisation of wheat and potato stomatal conductance models



Figure 3.8 (cont):

gmax and fmin

Receptor-specific values are provided forgmax and fmin based on analysis of publisheddata (see Sections 3.4.5 and 3.4.6 fordetails).

fphen

The phenology function can be based oneither a fixed number of days or effectivetemperature sum accumulation and has thesame shape for both approaches. However,use of the effective temperature sum is gen-erally accepted to describe plant develop-ment more accurately than using a fixed timeperiod since it allows for the influence oftemperature on growth. fphen is calculatedaccording to Eq.s 3.13a, b and c (when usinga fixed number of days) and 3.14a, b and c(when using effective temperature sumaccumulation). Each pair of equations givesfphen in relation to the accumulation periodfor AFstY where Astart and Aend are the startand end of the accumulation period respec-tively.

Method (a): based on a fixed time interval

(3.13a)

(3.13b)

(3.13c)

where yd is the year day; Astart and Aend arethe year days for the start and end of theozone accumulation period respectively.

Method (b): based on thermal time accumu-lation

(3.14a)

(3.14b)

(3.14c)

where tt is the effective temperature sum in°C days using a base temperature of 0°C andAstart and Aend are the effective temperaturesums (above a base temperature of 0 °C) at the start and end of the ozone accu-mulation period respectively. As such Astartwill be equal to 0°C days

flight

The function used to describe flight is given inEq. 3.15

(3.15)

where PFD represents the photosyntheticphoton flux density in units of µmol m-2 s-1.



when Astart ≤ yd < (Astart + fphen_c)

fphen = (1-fphen_a) * ((yd-Astart)/fphen_c)+ fphen_a

when (Astart + fphen_c) yd (Aend – fphen_d)

fphen = 1

when (Aend-fphen_d) < yd Aend

fphen = (1-fphen_b) * ((Aend – yd)/fphen_d) + fphen_b

when Astart ≤ tt < (Astart + fphen_e)

( )tt)f(Af

f11f phen_estart

phen_e

phen_a

phen −+−

−=

when (Astart + fphen_e) tt (Aend – fphen_f)

fphen = 1

when (Aend-fphen_d) < tt Aend

( ))f(Attf

f11f phen_fend

phen_f

phen_b

phen −−−

−=

flight = 1-EXP((-light_a)*PFD)



ftemp

The function used to describe ftemp is given inEq. 3.16.

(3.16a)

(3.16b)

where T is the air temperature in °C, Tmin andTmax are the minimum and maximum temper-atures at which stomatal closure occurs tofmin, Topt is the optimum temperature and btis defined as follows:

(3.17)

fVPD and SSVPD routine

The VPD of the air surrounding the leaves isused in two different ways. First, there is amore or less instantaneous effect of highVPD levels on stomata resulting in stomatalclosure which reduces the high rate of tran-spiration water vapor flux rates out of theleaf under such conditions. Under dry andhot conditions such limitation of VPD mayoccur early during the day after the sunrise.This instantaneous response of the stomatato VPD is described by the fVPD function.

The data used to establish the fVPD relation-ship for both wheat and potato are shown inFigure 3.8. The functions used to describefVPD are given in Eq. 3.18.

(3.18)

Secondly, there is another effect on stomataby water relations which can be modelledusing VPD. During the afternoon, the air tem-perature typically decreases, which is nor-mally, but not always, followed (if theabsolute humidity of the air remains con-stant or increases) by declining VPD.According to the fVPD function this wouldallow the stomata to re-open if there hadbeen a limitation by fVPD earlier during theday. Most commonly this does not happen.This is related to the fact that during the daythe plant loses water through transpiration ata faster rate than it is replaced by rootuptake. This results in a reduction of theplant water potential during the course of theday and prevents stomata re-opening in theafternoon. The plant water potential thenrecovers during the following night when therate of transpiration is low. A simple way tomodel the extent of water loss by the plant isto use the sum of hourly VPD values duringthe daylight hours (as suggested by Uddlinget al., 2003). If there is a large sum it is likelyto be related to a larger amount of transpira-tion, and if the accumulated amount of tran-spiration during the course of the day (asrepresented by a VPD sum) exceeds a cer-tain value, then stomatal re-opening in theafternoon does not occur. This is represent-ed by the VPDsum function (SSVPD) which iscalculated in the following manner:

(3.19)

Where gsto_hour_n and gsto_hour_n+1 are the gstovalues for hour n and hour n+1 respectivelycalculated according to Eq. 3.12.

SSVPD (kPa) should be calculated for eachdaylight hour until the dawn of the next day.Thus, if SSVPD is larger than or equal toSSVPDcrit, the gsto value calculated using Eq. 3.12 is valid if it is smaller or equal to thegsto value of the preceding hour. If gstoaccording to Eq. 3.12 is larger than gsto of thepreceding hour, given that SSVPD is largerthan or equal to SSVPDcrit, it is replaced by thegsto of the preceding hour in the

when Tmin < T < Tmax

ftemp = max {fmin, [(T-Tmin) / (Topt-Tmin)] *

[(Tmax-T) / (Tmax-Topt)] bt}

when Tmin> T >Tmax

ftemp = fmin

bt = (Tmax-Topt) / (Topt-Tmin)

fVPD = min{1, max{fmin, ((1-fmin)*(VPDmin – V

VPD)/(VPDmin – VPDmax))+fmin}}

If ΣVPD ≥ ΣVPD_crit, then gsto_hour_n+1 gsto_hour_n

estimation of stomatal conductance.

The SSVPD routine acts as a more mechanis-tically oriented replacement of the time ofday function used in the Pleijel et al. (2002)and Danielsson et al. (2003) parameterisa-tions. The instantaneous effect of VPD repre-sented by fVPD is allowed to be in operationas a function to further reduce the stomatalconductance also after the SSVPD routine hasstarted to limit stomatal conductance.

fSWP

The function used to describe fSWP is given inEq. 3.20.

(3.20)

fO3The flux-effect models developed by Pleijelet al. (2002) and Danielsson et al. (2003)include a function to allow for the influenceof ozone concentrations on stomatal conductance (fO3

) on wheat and potato viathe onset of early senescence. As such thisfunction is used in association with the fphenfunction to estimate gsto. The fO3

functiontypically operates over a one-month periodand only comes into operation if it has astronger senescence-promoting effect thannormal senescence. The functions are givenin Eq.s 3.21 and 3.22.

The ozone function for spring wheat (basedon Danielsson et al. (2003) but recalculatedfor PLA):

(3.21)

where AFstO is accumulated from Astart

The ozone function for potato (based onPleijel et al. (2002)):

(3.22)

where AOT0 is accumulated from Astart



fSWP = min{1, max{fmin, ((1-fmin)*(SWPmin – SWP)

/(SWPmin – SWPmax))+fmin}}



3.4.5 Calculation of AFstY and excee-dance of CLef for agricultural crops

3.4.5.1 Ozone concentration at the top of the wheat and potato canopy

The ozone concentration at the top of thecanopy can be calculated using the methodsdescribed in Section 3.4.2. For wheat andpotato the default height of the top of thecanopy is 1.0 m.

3.4.5.2 Estimating the time period for ozone flux accumulation for wheat and potato

Section 3.4.3 described two methods to esti-mate the influence of phenology on stomatalconductance (i.e. fphen) based on a fixedtime interval and based on thermal timeaccumulation. This section describes, foreach of these methods, the procedures toestimate the AFstY accumulation period forboth wheat and potato. The start and end ofthe accumulation period are identified byAstart and Aend respectively. Application of thefixed time interval method requires that Astartand Aend be defined in terms of day of theyear whilst application of the thermal timemethod requires that Astart and Aend bedefined in terms of °C days above a basetemperature of 0°C; in this case Astart willalways equal 0°C days though it is necessaryto determine the year day for Astart.

The four methods that are suggested forestimating the timing of the AFstY accumu-

lation period listed in order of desirabilityare: i) the use of observational data describ-ing actual growth stages; ii) the use of localagricultural statistics/information describingthe timings of growth stages by region orcountry iii) the use of phenological growthmodels in conjunction with daily meteoro-logical data and iv) the use of fixed time peri-ods (which may be moderated by climaticregion or latitude).

Wheat

It is necessary to identify the timing of mid-anthesis (defined as growth stage 65according to Zadoks et al., 1974) for bothspring and winter wheat. In the absence ofobservational or statistical informationdescribing growth stages, mid-anthesis canbe defined using phenological models ifdaily mean temperature data for the entireyear are available.

For spring wheat, mid-anthesis can be esti-mated using a temperature sum value of1075°C days calculated from plant emer-gence. In the absence of local informationplant emergence can be estimated from typ-ical sowing dates given for spring wheat byclimatic region in Table 3.14. These can beused to estimate the timing of emergence byassuming that the temperature sum (above abase temperature of 0°C) required for emer-gence would be 70°C days (assuming anaverage sowing depth of 3 cm) (Hodges &Ritchie, 1991).

CLef

An AFst6 of 1 mmol m-2

PLA

Time period Either 970˚C days, starting 200˚C days before mid-

anthesis or 55 days starting 15 days before mid-anthesis

Wheat


CLef An AFst6 of 5 mmol m-2

PLA

Time period Either 1130˚C days, starting at plant emergence or 70

days starting at plant emergence

Potato


Ref

eren

ce

gm

ax

(mm

ol

O3

m-2

s-1

) P

LA

gm

ax d

eriv

ati

on

C

ou

ntr

y

Wh

eat

typ

e a

nd

cult

iva

r

Tim

e o

f d

ay

Tim

e o

f

yea

r

gst

o m

easu

rin

g

ap

pa

ratu

s

Ga

s /

lea

f

are

a

Gro

win

g

con

dit

ion

s

Lea

f

Ali

et

al.

(19

99

)

61

0 (

± 7

3)

Fro

m g

rap

h s

ho

win

g l

eaf

co

nd

ucta

nce p

lott

ed

ag

ain

st t

ime

in d

ay

s. M

ax

imu

m a

pp

rox

imate

ly

1 m

ol

H2O

m-2

s-1

; ±

0.1

2. ±

SE

of

4 t

o 6

rep

licate

s.

Den

mark

S

pri

ng

wh

eat,

Cad

en

sa

(Ass

um

ed

mid

-day

)

Au

gu

st

IRG

A L

I-6

20

0

H2O

/

(Ass

um

e

PL

A a

s u

se

LA

I)

Fie

ld

Ly

sim

ete

r

Fla

g

Ara

us

et a

l.

(19

89

)

44

4

Valu

e i

n T

ab

le. C

ult

ivar

an

d

sow

ing

tim

e (

av

era

ge o

f 3

) g

sto

use

d.

Mean

s o

f 5

to

7 r

ep

licate

s.

gst

o m

mo

l C

O2 m

-2 s

-1. A

dax

ial:

31

3, ab

ax

ial:

14

9.

Sp

ain

S

pri

ng

wh

eat,

Ko

lib

ri

9 t

o 1

3 h

rs

14

Marc

h t

o

21

May

LI

16

00

ste

ad

y

state

po

rom

ete

r

CO

2 /

PL

A

Fie

ld

Fla

g

Ara

us

et a

l.

(19

89

)

34

5

Valu

e i

n T

ab

le. M

ean

s an

d S

E ±

of

5 t

o 7

rep

licate

s. g

sto m

mo

l C

O2

m-2

s-1

. A

dax

ial:

26

7 ±

29

,

ab

ax

ial:

92

± 1

6.

Sp

ain

S

pri

ng

wh

eat,

Ast

ral

9 t

o 1

3 h

rs

14

Marc

h t

o

21

May

LI

16

00

ste

ad

y

state

po

rom

ete

r

CO

2 /

PL

A

Fie

ld

Fla

g

Ara

us

et a

l.

(19

89

)

33

6

Valu

e i

n T

ab

le. M

ean

s an

d S

E ±

of

5 t

o 7

rep

licate

s. g

s m

mo

l C

O2

m-2

s-1

. A

dax

ial:

25

1 ±

15

,

ab

ax

ial:

99

± 2

2.

Sp

ain

S

pri

ng

wh

eat,

Bo

ulm

ich

e

9 t

o 1

3 h

rs

14

Marc

h t

o

21

May

LI

16

00

ste

ad

y

state

po

rom

ete

r

CO

2 /

PL

A

Fie

ld

Fla

g

Gru

ters

et

al.

(1

99

5)

48

5

Valu

e i

n t

ex

t. M

ax

imu

m

measu

red

co

nd

ucta

nce (

0.9

7 c

m s

-

1 H

2O

to

tal

leaf

are

a a

fter

Jon

es

(19

83

)).

Germ

an

y

Sp

rin

g

wh

eat,

Tu

rbo

11

to

12

hrs

1

7 J

un

e t

o 7

Au

gu

st

LI

16

00

ste

ad

y

state

po

rom

ete

r

H2O

/ t

ota

l

leaf

are

a

Fie

ld

Fla

g

Kö

rner

et a

l.

(19

79

)

45

5

Valu

e g

iven

in

tab

le. 0

.91

cm

s-1

for

H2O

on

a t

ota

l le

af

surf

ace

are

a b

asi

s.

Au

stri

a

Du

rum

wh

eat,

Jan

us

- -

Ven

tila

ted

dif

fusi

on

po

rom

ete

r

H2O

/ t

ota

l

leaf

are

a

Fie

ld

Fla

g

Dan

iels

son

et a

l. (

20

03

)

50

7

Valu

e i

n t

ex

t. "

Th

e m

ax

imu

m

co

nd

ucta

nce v

alu

e,

41

4 m

mo

l

H2O

m-2

s-1

, w

as

tak

en

as

gm

ax f

or

the Ö

stad

mu

ltip

licati

ve m

od

el.

Th

e c

on

du

cta

nce v

alu

es

rep

rese

nt

the f

lag

leaf

an

d a

re g

iven

per

tota

l le

af

are

a".

Sw

ed

en

S

pri

ng

wh

eat,

Dra

go

n

13

hrs

1

3 A

ug

,

19

96

(A

A)

Li-

Co

r 6

20

0

H2O

/ t

ota

l

leaf

are

a

Fie

ld

OT

C a

nd

AA

Fla

g

Mean

Med

ian

45

5

45

5

Ran

ge:

33

6 t

o 6

10



Table

3.1

2:

Der

ivat

ion

of

whe

at (T

ritic

um a

estiv

um) g

max

par

amet

eris

atio

n. P

LA =

pro

ject

ed le

af a

rea



Ref

eren

ce

gm

ax

(mm

ol

O3

m-2

s-1

) P

LA

gm

ax d

eriv

ati

on

C

ou

ntr

y

Po

tato

cult

iva

r

Tim

e o

f d

ay

Tim

e o

f y

ear

gst

o m

easu

rin

g

ap

pa

ratu

s

Ga

s /

lea

f

are

a

Gro

win

g

con

dit

ion

s

Lea

f

Jeff

ries

(19

94

)

80

0

Valu

e g

iven

in

Fig

ure

.

Max

imu

m v

alu

e o

f 1

6 m

m s

-1.

Err

or

bar

rep

rese

nts

SE

of

the

dif

fere

nce b

etw

een

tw

o m

ean

s

(n=

48

).

Sco

tlan

d

Mari

s p

iper

8 t

o 1

6 h

rs

Jun

e

Dif

fusi

on

po

rom

ete

r

Ass

um

ed

H2O

/

ass

um

ed

PL

A

Fie

ld

Fu

lly

ex

pan

ded

in

up

per

can

op

y

Vo

s &

Gro

en

wald

(19

89

)

66

5

Valu

e g

iven

in

Fig

ure

.

Max

imu

m v

alu

e o

f 1

3.3

mm

s-

1. R

ep

licate

s ap

pro

x. 2

0, th

e

co

-eff

icie

nt

of

vari

ati

on

typ

icall

y r

an

ged

fro

m 1

5 t

o

25

%.

Neth

erl

an

ds

Bin

tje

- Ju

ne/J

uly

L

i-C

or

16

00

ste

ad

y

state

dif

fusi

on

po

rom

ete

r

H2O

/ P

LA

F

ield

Y

ou

ng

est

full

y g

row

n

leaf

Vo

s &

Gro

en

wald

(19

89

)

75

0

Valu

e g

iven

in

Fig

ure

.

Max

imu

m v

alu

e o

f 1

5 m

m s

-1.

Rep

licate

s ap

pro

x. 2

0, th

e c

o-

eff

icie

nt

of

vari

ati

on

ty

pic

all

y

ran

ged

fro

m 1

5 t

o 2

5%

.

Neth

erl

an

ds

Satu

rna

- Ju

ne

Li-

Co

r 1

60

0 s

tead

y

state

dif

fusi

on

po

rom

ete

r

H2O

/ P

LA

F

ield

Y

ou

ng

est

full

y g

row

n

leaf

Mars

hall

&

Vo

s (1

99

1)

64

3

Valu

e g

iven

in

Fig

ure

. g

max o

f

52

7 m

mo

l H

2O

m-2

s-1

at

inte

rmed

iate

N s

up

ply

. E

ach

po

int

rep

rese

nts

th

e m

ean

of

at

least

th

ree l

eav

es

(usu

all

y

fou

r).

Neth

erl

an

ds

Pro

min

en

t -

July

L

CA

2 p

ort

ab

le

infr

a-r

ed

gas

an

aly

ser

H2O

/

ass

um

ed

PL

A

Fie

ld

Mo

st

recen

tly

ex

pan

ded

measu

rab

le

leaf

Ple

ijel

et a

l.

(20

02

)

83

6

Valu

e g

iven

in

Tab

le. g

max o

f

13

71

mm

ol

m-2

s-1

for

H2O

per

pro

jecte

d l

eaf

are

a.

Germ

an

y

Bin

tje

12

Ju

ne

Li-

Co

r 6

20

0

H2O

/ P

LA

F

ield

F

ull

y

ex

pan

ded

in

up

per

can

op

y

Dan

iels

son

(20

03

)

73

7

Valu

e g

iven

in

tex

t. g

max o

f 6

04

mm

ol

H2O

m-2

s-1

per

tota

l le

af

are

a.

Sw

ed

en

K

ard

al

11

Ju

ly

Li-

Co

r 6

20

0

H2O

/ T

ota

l

leaf

are

a

Fie

ld

Fu

lly

ex

pan

ded

in

up

per

can

op

y

Mean

Med

ian

73

8

74

3

Ran

ge:

64

3 t

o 8

36

Table

3.1

3:

Der

ivat

ion

of

po

tato

(So

lanu

m t

uber

osu

m) g

max

par

amet

eris

atio

n. P

LA =

pro

ject

ed le

af a

rea

For winter wheat, growth can be assumed torestart after the winter when temperatureexceeds 0°C. Traditionally, the starting datefor the accumulation of the effective temper-ature sum to mid-anthesis for winter wheat isthe first date after 1 January when the tem-perature exceeds 0°C, or 1 January if thetemperature exceeds 0°C on that date.Using this start point, mid-anthesis can beestimated using a temperature sum of1075°C days after 1 January (it should benoted that these calculations ignore anyeffects of photoperiod).

In the absence of appropriate temperaturedata, the timing of mid-anthesis for bothspring and winter wheat could be approxi-mated as a function of latitude (degrees N)using Eq. 3.23. However, it should be recog-nised that this method is less preferable tothe use of the effective temperature summodels described above since latitude is notdirectly related to temperature and thismethod will not distinguish between springand winter wheat growth patterns.

(3.23)

Eq. 3.23 is based on data collected by theICP Vegetation (Mills et al., 1998) from tensites across Europe (ranging in latitude fromFinland to Slovenia) describing the date ofanthesis of commercial winter wheat.Applying Eq. 3.23 across the Europeanwheat growing region would give mid-anthe-sis dates ranging from the end of April tomid-August at latitudes of 35 to 65°Nrespectively. These anthesis dates fallappropriately within recognised springwheat growing seasons as described byPeterson (1965) and also from data for winter wheat supplied for Spain by Gimenoet al. (2003b).



Mid-anthesis = 2.57 * latitude + 40

Region Range Default

Northern Europe

Finland 1-30 May 30 May

Norway 1-20 May 20 May

Sweden 1-20 Apr 20 April

Denmark 1 Mar-20 Apr 20 March

Continental Central Europe

Poland 1-20 Apr 10 April

Czech Republic 10-30 Apr 20 April

Slovakia 10-30 Apr 20 April

Romania -

Hungary -

Germany 10 Mar-10 Apr 1 April

Atlantic Central Europe

UK 20 Feb-20 Mar 10 March

The Netherlands 1-30 Mar 15 March

France 1 Mar-10 Apr 20 March

Mediterranean Europe

Bulgaria -

Portugal 20 Jan-10 Mar 10 February

Spain 1-28 Feb 10 February

Italy (soft and durum wheat) - (not grown) 1

A di t B kh i (1969)

Table 3.14: Observed sowing dates for spring wheat in Europe1

1According to Broekhuizen (1969)



Potato

For potato, it is necessary to identify plantemergence, which normally takes place oneweek to ten days after sowing. Although thesowing date varies to a considerable extentacross Europe, information from the EU-funded research programme CHIP, whichinvestigated the effects of ozone and otherstresses on potato, found that plant emer-gence was obtained on average on day ofyear 146, with a variation from day 135 atsouthern and most western European sitesto day 162 in Finland. As such, in theabsence of local information describingsowing dates it is suggested that year day146 be used as a default to define Astart forpotato plant emergence. No phenologicalmodels are suggested for use to define Astartfor this species.

3.4.5.3 Parameterisation of stomatal flux models for wheat and potato

The original parameterisations given inEmberson et al. (2000b) have been revisedbased on data collected from more recentlypublished literature and from ozone expo-sure experiments conducted in Sweden forwheat (Danielsson et al., 2003) and from anumber of sites across Europe for potato(Pleijel et al., 2003). Those recommended foruse in calculating AFstY using the stomatalflux algorithm described in Section 3.4.4 areshown in Table 3.15.

Parameter Units Wheat

(Triticum aestivum)

Potato

(Solanum tuberosum)

gmax mmol O3 m-2

PLA s-1

450 750

fmin (fraction) 0.01 0.01

fphen_a (fraction) 0.8 0.4

fphen_b (fraction) 0.2 0.2

fphen_c days 15 20

fphen_d days 40 50

fphen_e °C days 270 330

fphen_f °C days 700 800

lighta (constant) 0.0105 0.005

Tmin °C 12 13

Topt °C 26 28

Tmax °C 40 39

VPDmax kPa 1.2 2.1

VPDmin kPa 3.2 3.5

VPDcrit kPa 8 10

SWPmax MPa -0.3 -0.5

SWPmin MPa -1.1 -1.1

Table 3.15: Summary of the parameterisation for the stomatal flux algorithms for wheat flag leaves and potatoupper-canopy sunlit leaves.

The gmax values for wheat and potato havebeen derived from published data conform-ing to a strict set of criteria so as to bedeemed acceptable for use in establishingthis key parameter of the flux algorithm. Onlydata obtained from gsto measurements madeon cultivars grown either under field condi-tions or using field-grown plants in open topchambers in Europe were considered.Measurements had to be made during thosetimes of the day and year when gmax wouldbe expected to occur and full details had tobe given of the gas for which conductancemeasurements were made (e.g. H2O, CO2,O3) and the leaf surface area basis on whichthe measurements were given (e.g. total orprojected). All gsto measurements were madeon the flag leaf for wheat and for sunlitleaves of the upper canopy for potato usingrecognized gsto measurement apparatus.Tables 3.12 and 3.13 give details of the published data used for gmax derivation onadherence to these rigorous criteria. Figure3.8 shows the mean, median and range ofgmax values for each of the six and four different cultivars that provide the approximated gmax values of 450 and 750mmol O3 m-2 PLA s-1 for wheat and potato,respectively.

It should be noted that the wheat gmax valuehas been parameterised from data collectedfor spring wheat cultivars. Comparisons withdata for winter wheat presented by Bunce(2000) would suggest that the gmax for springand winter wheat are likely to be similar.Bunce (2000) gives a gmax value of 464 mmol O3 m-2 s-1 on a projected leaf area basisfor winter wheat. These measurements wereconducted in the USA but nevertheless areuseful in indicating a similarity in the gmax ofboth wheat types. Similarly, for potato addi-tional gmax values from three USA grown cul-tivars are included in Figure 3.8 for compari-son (Stark, 1987) further substantiating thegmax value established for this crop type.

fmin

The data presented in Pleijel et al. (2003) andDanielsson et al. (2003) clearly show that for

both species, fmin under field conditions fre-quently reaches values as low as 1% of gmax.Hence an fmin of 1% of gmax is used to para-meterise the model for both species.

fphen

The data used to establish the fphen relation-ships for both wheat and potato are given inFigure 3.2 as °C days from gmax (where gmaxis assumed to occur at mid-anthesis forwheat and at the emergence of the first gen-eration of fully developed leaves in potato).Methods are also provided in Section 3.4.4to estimate fphen according to fixed time peri-ods using the parameterisation given inTable 3.15. These data not shown in Figure3.8.

flight

The data used to establish the flight relation-ship for both wheat and potato are shown inFigure 3.8.

ftemp

The data used to establish the ftemp relation-ship for both wheat and potato are shown inFigure 3.8.

fVPD and SSVPDcrit

The data used to establish the fVPD relation-ship for both wheat and potato are shown inFigure 3.8.Values of SVPDcrit for wheat and potato aregiven in Table 3.15.

fSWP

The data used to establish the fSWP relation-ship for both wheat and potato are given inFigure 3.8. It should be noted that the fSWPrelationship for potato is derived from datathat describe the response of potato gsto toleaf water potential rather than soil waterpotential. Vos and Oyarzun (1987) state thattheir results represent long-term effects ofdrought, caused by limiting supply of water





rather than by high evaporative demand, andhence can be assumed to apply to situationswhere pre-dawn leaf water potential is lessthan 0.1 to 0.2 MPa. As such, it may be nec-essary to revise this fSWP relationship so thatpotato gsto responds more sensitively toincreased soil water stress.

fO3

The functions described for wheat andpotato in Section 3.4.4 should be used.

Once parameterisation of the model is com-plete for the species being studied, AFstYand CLef exceedance can be calculated fol-lowing the stages described in Section 3.4.1.

3.4.6 Calculation of AFstY and excee-dance of CLef for forest trees

Note: The recommended method for foresttrees is AOTX-based critical levels; a provi-sional AFstY method is provided for guidanceonly.

As proposed at the Gothenburg workshop(Karlsson et al., 2003a), a default methodolo-gy for a flux-based risk assessment for for-est trees is presented, as an alternative toconcentration-based AOT40 exposureassessments. Concentration-based assess-ment methods do not take climatic limitation(soil moisture deficit, water vapour pressuredeficit) of exposure into account and AOTX-based approaches may, for example,overestimate environmental risk in southernEurope where summer moisture deficits arecommonplace. However, at the time of writ-ing this manual, it was considered that theuncertainties associated with using a flux-based assessment for forest trees for all ofEurope were too large for the method to befully recommended for use (these are dis-

cussed in detail in Karlsson et al., 2003b,2004). It should also be emphasised that themethodology presented here should not beused for economic assessments of damageor yield loss, but should be used solely formapping areas at potential risk from ozoneand as an alternative to the recommendedAOT40 approach.

3.4.6.1 Ozone concentration at the top of the forest canopy

The ozone concentration at the top of thecanopy can be calculated using the methodsdescribed in Section 3.4.2. For beech andbirch, the default height is 20 m.

3.4.6.2 Parameterisation of the stomatal flux model for beech and birch

The calculation of stomatal conductance(gsto) for sunlit leaves in the upper canopyuses the same multiplicative model asdescribed in Eq. 3.12 for wheat and potatoand described in Section 3.4.4, with theexceptions that there is no fO3

function andthe SVPD routine is not applied to estimatefVPD (see later). The parameterisation of thismodel is similar to that described inEmberson et al. (2000a), but has beenimproved as information has become available from recently published literature(as described in Karlsson et al., 2003b,2004). The most significant change has beento the ftemp parameterisation that now allowsgsto to occur at temperatures down to –5°C(previously the Tmin value had been 13°C).The model parameterisation described herewas used for the derivation of the forest treeflux-based critical level AFst1.6 of 4 mmol m-2 PLA as described in Karlsson et al.(2003b, 2004).

The accumulation period for AFstY for foresttrees currently lasts for the entire growingseason since no information is available todefine a particular phenologically sensitiveperiod. As such, Astart and Aend will be equalto the start (onset of bud-burst) and the end(end of leaf senescence period) of the grow-ing season respectively. The default value for

Clef

Provisionally an AFst1.6

of 4 mmol m-2

PLA

Time period One growing season

Beech

and

birch

Effect Growth reduction

Astart is year day 90 and for Aend is year day270. Currently, no data are available todefine fphen according to effective tempera-ture sum models so only the fixed time inter-val fphen methods described in Section 3.4.3are applied. Where local parameterisationsof the model are used, it is recommendedthat parallel assessments are also made andreported using the default values given inTable 3.16, thus maintaining compatibilitybetween individual national assessments.Insufficient information is available to definefO3

for forest trees and thus it is recommend-ed that fO3

is set to 1.0 in Eq. 3.12.

The SVPD routine described for wheat andpotato that accounts for lack of re-openingof stomata in the afternoon, cannot beapplied to forest trees with confidence asinsufficient data exists to complete the para-meterisation. It is thus recommended thatthis routine is omitted from the fVPD proce-dure described in Section 3.4.4.

AFstY and CLef exceedance should be calcu-lated following the stages described inSection 3.4.1.



Parameter Value

gmax 134 mmol O3 m-2

projected leaf area s-1

fmin 0.13 (fraction)

fphen_a 0.3 (fraction)

fphen_b 0.3 (fraction)

fphen_c 50 (days)

fphen_d 50 (days)

fphen_e no parameterisation available, use fixed time

period method

fphen_f no parameterisation available, use fixed time

period method

light_a 0.006

Tmin -5oC

Topt 22oC

Tmax 35oC

VPDmax 0.93 kPa

VPDmin 3.4 kPa

ΣVPDcrit No parameterisation available, omit ΣVPD

routine

SWPmax -0.05 MPa

SWPmin -1.5 MPa

Table 3.16: Default parameters for beech for use in estimating ozone flux for forest tree species.

Note: values have been modified since original publication of Emberson et al. (2000a).



3.5.1 Stages in calculating AOTX and CLec exceedance

The calculation of AOTX for comparison withthe CLec value for a given vegetation type orplant species is based on hourly mean val-ues of ozone at the top of the canopy (seeSection 3.4.2 for estimating the canopyheight ozone concentration from the meas-urement height ozone concentration). For alldaylight hours the difference between thehourly mean concentration and X ppb is cal-culated; then the sum of all hourly valueswith a positive value (i.e. with hourly meanozone concentrations above X ppb) is calcu-lated for the growth period of the receptor.This calculation is illustrated in Figure 3.9.

It is recommended that AOTX values forcomparison with CLec should be calculatedas the mean value over the most recent five

years for which appropriate quality assureddata are available. For local and national riskassessment, it may also be valuable tochoose the year with the highest AOT40 fromthe five years.

In summary, the following stages arerequired for calculation of AOTX andexceedance of CLec:

1. Identify the relevant growth periods for the receptor for accumulation of the exposure index.

2. Obtain high quality ozone data for this growth period for the five most recent years.

3. Adjust the ozone data from measure-ment height to canopy height using anappropriate model or the algorithm in this manual.

4. Calculate the AOTX index for each of the five years.

5. Obtain the mean of the five values of AOTX and compare with CLec for the receptor.

3.5 Calculating exceedance of concentration-based critical levels for ozone

0

20

40

60

80

100

120

140

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Time of day (h)

O3 c

oncentr

atio

n (

ppb)

Contributes to AOT40

Does not contribute to AOT40

Daylight hours

Figure 3.9: Calculation of ozone accumulated over a threshold of 40 ppb (AOT40) in ppb h for Balingen (6 May, 1992). The AOT40 for this day is 383 ppb h, calculated as 17 (exceedance of 40 ppb for 11th hour) + 35 (12th hour) + 30 (13th hour) + 47 (14th hour) + 51 (15th hour) + 55 (16th hour) + 52 (17th hour) + 51 (18th hour) + 45 (19th hour). Exceedance of 40 ppb in the 20th hour is not included because it occurred after daylight had ended.

3.5.2 Calculation of AOTX and CLecexceedance for agricultural crops

These critical levels are only applicablewhen nutrient supply and soil moisture arenot limiting, the latter because of sufficientprecipitation or irrigation.

3.5.2.1 Ozone concentrations at the canopy height for agricultural crops

In all AOTX calculations, the ozone concen-tration at the canopy height is used and calculated as described in Section 3.4.2. Thedefault canopy height is 1 m for agriculturalcrops.

3.5.2.2 Accumulation period for agricul-tural crops

The timing of the three month accumulationperiod for agricultural crops should reflectthe period of active growth of wheat and becentred on the timing of anthesis. A surveyof the development of winter wheat conducted at 13 sites in Europe by ICPVegetation participants in 1997 and 1998,revealed that anthesis can occur as early as2 May in Spain and as late as 3 July inFinland (Mills et al., 1998). Thus, a riskassessment for ozone impacts on cropswould benefit from the use of a moving timeinterval to reflect the later growing seasonsin northern Europe. For guidance, defaulttime periods have been provided for fivegeographical regions as indicated in Table3.17.

Once the time period has been established,AOT40 and CLec exceedance can be calculated following the stages described inSection 3.5.1.



CLec

An AOT40 of

3 ppm h

Time period 3 months

Agricultural

crops

Effect Yield

reduction

Region Three month time

period

Possible default countries

Eastern

Mediterranean

1 March to 31 May Albania, Bosnia and Herzogovina, Bulgaria, Croatia,

Cyprus, Greece, FYR Macedonia, Malta, Slovenia,

Turkey, Yugoslavia

Western

Mediterranean

1 April to 30 June Italy, Portugal, Spain

Continental Central

Europe

15 April to 15 July Armenia, Austria, Azerbaijan, Belarus, Czech Republic,

France1, Georgia, Germany, Hungary, Kazakhstan,

Krygyzstan, Liechtenstein, Moldova, Poland, Romania,

Russian Federation, Slovakia, Switzerland, Ukraine

Atlantic Central

Europe

1 May to 31 July Belgium, Ireland, Luxembourg, Netherlands, United

Kingdom

Northern Europe 1 June to 31 August

Denmark, Estonia, Faero Islands, Finland, Iceland,

Latvia, Lithuania, Norway, Sweden

Table 3.17: Regional classification of countries for default time periods for calculation of AOTX for agriculturalcrops

1as an average between Western Mediterranean and Atlantic Central Europe



3.5.3 Calculation of AOTX and CLecexceedance for horticultural crops

These critical levels are only applicablewhen nutrient supply and soil moisture arenot limiting, the latter because of sufficientprecipitation or irrigation.

3.5.3.1 Ozone concentrations at the canopy height for horticultural crops

In all AOTX calculations, the ozone concen-tration at the canopy height is used and cal-culated as described in Section 3.4.2. Thedefault canopy height is 1 m for horticulturalcrops.

3.5.3.2 Accumulation period for horticul-tural crops

Since horticultural crops are repeatedlysown over several months in manyMediterranean regions, it is recommendedthat locally-appropriate 3.5 month periodsare selected between March and August foreastern Mediterranean areas, and Marchand October for western Mediterraneanareas, although other time periods could beused depending on local crops and horticul-tural practices. Whereas it is most appropri-ate to recommend usage of this critical levelfor Mediterranean regions, it may also beappropriate to apply the critical level to otherparts of Europe since several of the cultivarsused to derive the critical level are grown inother regions of Europe (Figure 3.5). Forsuch countries, appropriate 3.5 month peri-ods should be selected within the periodApril to September.

Once the time period has been established,AOT40 and CLec exceedance can be calcu-lated following the stages described inSection 3.5.1.

3.5.4 Calculation of AOTXVPD and exceedance of the short-termcritical level for ozone injury

This critical level represents the risk ofdevelopment of visible injury on sensitiveagricultural and horticultural crop species. IfVPD values are not available, the AOT30value can be taken as an indication of poten-tial risk for visible injury without modifyingthe ozone concentrations. However, overlarge areas of Europe, VPD typically exerts aconsiderable restriction to stomatal conduc-tance during the growing season in situa-tions with elevated ozone concentrations.

The AOTXVPD index is calculated by usinghourly values of the ozone concentration,expressed in ppb at the height of the top ofthe canopy (see Section 3.4.2). A defaultheight of 1 m is applicable to agricultural andhorticultural crops. Each hourly ozone concentration [O3] is multiplied by an hourlyfVPD factor, reflecting the influence of VPD onstomatal conductance, to get hourly, modified ozone concentrations [O3]VPD:

(3.24)

where:

fVPD = 1 if VPD < 1.1 kPa

fVPD = - 1.1*VPD + 2.2 if 1.1 kPa £ VPD £

1.9 kPa

fVPD = 0.02 if VPD > 1.9 kPa

The calculation of VPD is described in textbooks (e.g. Jones, 1992).

In the case of the short-term critical level forozone injury X, in AOTXVPD, is 30 ppb. TheAOT30VPD is obtained by first subtracting

Clec

An AOT40 of

6 ppm h

Time period 3.5 months

Horticultural

crops

Effect Yield

reduction

CLec

A VPD-modified

AOT30 of 0.16 ppm h

Time period Previous 8 days

Agricultural

and

horticultural

crops Effect Visible injury to

leaves

[O3]VPD = fvpd * [O3]

30 ppb from each hourly [O3]VPD > 30 ppb,and then making a sum of the resulting val-ues. Thus, [O3]VPD values £ 30 ppb do notcontribute to AOTX30VPD. AOT30VPD is cal-culated over eight day periods to identify thepotential risk of ozone injury on sensitivecrops and expressed in units of ppm h. Thus,if the eight day AOT30VPD exceeds the CLecfor ozone injury, then injury is likely. It isimportant that the period over which theAOT30VPD value is calculated is consistentwith the period when the relevant receptor isactively growing and absorbing ozone. TheAOT30VPD value is calculated using runningeight day periods throughout the season.

3.5.5 Calculation of AOTX and CLecexceedance for (semi-) natural vegetation

The concentration-based critical level shownabove should be applied to all (semi-) natu-ral vegetation.

3.5.5.1 Ozone concentrations at canopy height

The AOT40 value should be calculated as theconcentration at canopy height, using theinformation provided in Section 3.4.2. Thetransfer functions to make this calculation,based on deposition models, depend on anumber of factors which may vary systemat-ically between EUNIS categories (describedin Section 3.5.5.3). These include canopyheight and leaf area index (both natural vari-ation and effects of management) and envi-ronmental variables such as vapour pressuredeficit and soil moisture deficit. If such infor-

mation is not available, it is recommendedthat the conversion factors described inSection 3.4.2 for short grasslands are usedas a default.

3.5.5.2 Time window for calculating AOT40 for ( semi-) natural vege-tation

Ideally, a variable time-window should beused in the mapping procedure to accountfor different growth periods of annuals andperennials in different regions of Europe. TheAOT40 is calculated over the first threemonths of the growing season. The start ofthe growing season can be identified using:

1. appropriate phenological models;

2. information from local or national experts; and

3. the default table below (Table 3.18).

For a small number of species, the growingseason may be less than three months induration. In such cases, values of AOT40should be calculated over the growing sea-son, identified using appropriate local infor-mation.


Table 3.18: Default timing for the start and end ofozone exposure windows for (semi-) natural vegeta-tion.

Note: regional classifications of countries are suggest-ed in Table 3.17.



Parameter

An AOT40 of 3 ppm h

Time period 3 months (or growing

season, if shorter)

(Semi-)

natural

vegetation

Effect Growth reduction in

perennial species and

growth reduction

and/or seed production

reduction in annual

species

Region Start End

Eastern Mediterranean* 1 March 31 May

Western Mediterranean* 1 March 31 May

Continental central Europe 1 April 30 June

Atlantic central Europe 1 April 30 June

Northern Europe 1 May 31 July

*Note: For mountainous areas where altitudeis >1500m use 1 April to 30 June.



3.5.5.3 Mapping ( semi-) natural vegeta-tion communities at risk from exceedance of the critical level

For more detailed mapping based on com-munities which are more likely to be sensi-tive to ozone, the classification of theEuropean Nature Information System, EUNIS(refer to http://mrw.wallonie.be/dgrne/sibw/Eunis/) may be used for the preliminary iden-tification of those grassland types acrossEurope for which the critical level is support-ed by experimental data. The EUNIS classesfor which evidence exists to support the crit-ical level are:

E1: Dry grasslands

E2: Mesic grasslands

E3: Seasonally-wet and wet

grasslands

E7.3: Dehesa

In addition, from reports of substantial visi-ble injury in the field, a further EUNIS catego-ry can be identified as probably sensitive:

E5: Woodland fringes and clearings and tall forb habitats

Within these categories, particular sub-divi-sions may be identified as sensitive fromtheir species composition, for example thepresence of high proportions of legumes.However, no specific mapping recommenda-tions for EUNIS sub-divisions can be pro-posed at this stage.

3.5.6 Calculation of AOTX and CLecexceedance for forest trees

3.5.6.1 Ozone concentrations at canopy height.

It is important that the calculation of AOT40is based on ozone concentrations at the topof the canopy as described in Section 3.4.2.The default canopy height for forest treesis 20 m.

3.5.6.2 Time-window for calculating ozone exposure for forest trees

For the purposes of this manual, the defaultexposure window for the accumulation ofAOT40 is suggested to be 1 April to 30 September for all deciduous and ever-green species in all regions throughoutEurope. This time period does not take alti-tudinal variation into account and should beviewed as indicative only. It should bestressed that it should only be used wherelocal information is not available. Whendeveloping local exposure windows, the following definitions should be used:

• Onset of growing season in deciduous species: the time at which flushinghas initiated throughout the entiredepth of crown.

• Cessation of growing season in deciduous species: the time at which the first indication of autumn colour change is apparent.

• Onset of growing season in evergreen species: when the night temperatures are above -4°C for 5 days: if they do not fall below –4°C, the exposurewindow is continuous.

• Cessation of growing season in ever-green species: when the nighttemperatures are below -4°C for5 days: if they do not fall below –4°C, the exposure window is continuous.


CLec

An AOT40 of 5 ppmh

Time period Growing season

Forest trees

Effect Growth reduction

Note: This method should not be used formapping at the European scale or foreconomic impact assessment.

A second approach is provisionally availablefor application in national-level evaluations,using an alternative concentration-basedindicator of potential risk. The MaximumPermissible Ozone Concentration approach(MPOC: see Grünhage et al., 2001; Krause etal., 2003 for detailed descriptions of themethodology) is a largely qualitative indica-tor, relating a range of significant responseeffects (including growth and physiology) tothe concentration of ozone above the canopy over a variety of timeframes. Thismethod should not be used for mapping atthe European scale. Using the data for lon-ger term effects identified with the MPOCapproach, it is possible to indicate that theupper and lower range for effects reported inthe literature are seasonal (April toSeptember) 24h mean ozone concentrationsin the range 25-74 ppb. As is the case withthe AOT40 approach to mapping areas inwhich sensitive receptors are at risk fromozone pollution, the MPOC approach doesnot account for environmental limitations(soil moisture deficit, leaf-air vapour pressure deficit) to physiologically effectiveozone exposure. The methodology is basedon a review of 22 peer-review publicationsand 50 individual datasets, including a rangeof different exposure facilities. Although thedatasets on which the concept are basedcover a range of species and locationsacross Europe, species common in southernEurope and the Mediterranean region areunder-represented. The methodology hasalso only been validated for sites inGermany, while datasets published after1999 have not been included in the analysisto date. It is thus suggested that the approach could possibly be adopted as anadditional concentration-based index formapping potential risk in national evalua-tions, but that economic losses attributable

to ozone pollution could not be estimated onthe basis of this approach. Further validationand development of the methodology maybe possible under the framework of ICPForests.



ANNEX1: MPOC approach to riskassessment for forest trees



The critical levels and methods described forozone in this chapter were prepared by leading European experts from availableknowledge on impacts of ozone on vegeta-tion, and thus represent the current "state ofknowledge". As described in the chapter, allindicators for ozone impacts are based onthe accumulation of ozone (either as concentration or stomatal flux) above a predetermined threshold over a specifiedtime period. The scientific support for thecritical levels of ozone is described inSection 3.3 and the different methods aresuitable for various purposes. For ease ofuse, this annex compiles the suggestionsincluded in the chapter for their applicationfor integrated modelling and where appropri-ate, for an economic impact assessment. Ingeneral, the flux-based approach is deemedsuitable for estimating both risk of damageand magnitude of damage, whereas the concentration-based approach is only considered suitable for indicating the risk ofdamage.A schematic illustrating the steps involved incalculating exceedance and including refe-rences to relevant sections is included in thechapter as Figure 3.1, Section 3.2.4.

Yield loss in agricultural crops: The stomatalflux-effect models for wheat and potato useAFst6 as the ozone parameter (Section 3.4.5),and provide the best available method forestimating impacts of ozone on crops. Theimpacts can be quantified in various ways,inter alia, by assessing the risk of damagemeasured as area of flux-based critical levelexceedance, or by estimating yield lossusing stomatal flux-based functions (Eq. 3.1and 3.2) which could also be used for econo-mic loss evaluation. The risk of damage canalso be determined for agricultural cropsusing AOT40-based critical levels (Section3.5.2) using a three month time interval forthe five climatic zones of Europe (Table 3.17,

Section 3.5.2.2) and a canopy height of 1m(Section 3.5.2.1). An AOT30-based criticallevel has been defined in Section 3.3.1.3 foragricultural crops that could also be used inintegrated modelling to determine the risk ofdamage. However, the AOT40 and AOT30methods do not take into account theinfluence of climatic conditions on ozoneflux and therefore have greater uncertaintyassociated with them. The methods pres-ented here would enable the quantificationof direct effects on yield loss and relatedeconomic value for two agricultural crops(wheat and potato). However, it is recogni-sed that an in depth economic assessmentof impacts on crops would include direct(e.g. yield, injury) as well as indirect effects(e.g. on pests and diseases) and the associ-ated uncertainty for a range of crops inclu-ding wheat and potato.

Yield loss in horticultural crops: The onlyoption currently available is to determine therisk of damage as exceedance of the AOT40-based critical level for horticulturalcrops (Section 3.5.3). A canopy height of1 m should be used (Section 3.5.3.1) as wellas a time interval of 3.5 months for the clima-tic zones suggested (Section 3.5.3.2).

Visible injury on agricultural and horticulturalcrops: Use the VPD-modified AOT30(AOT30VPD) critical level to assess the fre-quency of risk of ozone injury over running 8day periods (Section 3.5.4) using a canopyheight of 1 m. If hourly VPD data is not avai-lable, the AOT30-based critical level could beused as an alternative, but with less certainty.

(Semi-) natural vegetation: The only optionavailable at this stage is to determine the riskof damage (growth reduction or reducedseed production) as exceedance of theAOT40-based critical level for semi-naturalvegetation (Section 3.5.5) using a time inter-val of three months for five climatic zones asindicated in Table 3.18 and a canopy heightof 0.5 m.

ANNEX 2: Guidance for assessingimpacts of ozone on vegetation usingthe models described in this chapter

Forest trees: The risk of damage (growthreduction) to forest trees can be determinedas exceedance of the AOT40-based criticallevel for forest trees (Section 3.5.6) for thegrowing season (April to September) and acanopy height of 20 m. An AOT30-based critical level has been defined in Section3.3.3 for forest trees that could also be usedin integrated modelling to determine the riskof damage. A provisional flux-based criticallevel for forest trees has been presented inSection 3.4.6, however, it should be not beapplied to integrated modelling at this stage.The methods described in this chapter forforest trees should only be used to definerisk of damage as the area of exceedance ofthe critical level; the methods should not beused to quantify the magnitude of effects ongrowth.





Ali M, Jensen CR, Mogensen VO, AndersenMN, Hensen IE (1999) Root signalling andosmotic adjustment during intermittentsoil drying sustain grain yield of fieldgrown wheat. Field Crops Research 62:35-52.

Araus JL, Tapia L, Alegre L (1989) The effectof changing sowing date on leaf structureand gas exchange characteristics ofwheat flag leaves grown underMediterranean climate conditions.Journal of Experimental Botany 40 (215):639-646.

Ashmore MR, Ainsworth N (1995) The effectof ozone and cutting on the species com-position of artificial grassland communi-ties. Functional Ecology 9: 708-712.

Ashmore MR, Davison AW (1996) Towards acritical level of ozone for natural vegeta-tion. In: Kärenlampi L, Skärby L (eds)(1996). Op. cit.: 58-71.

Ashmore MR, Power SA, Cousins DA,Ainsworth N (1996) Effects of ozone onnative grass and forb species: a compari-son of responses of individual plants andartificial communities. In: Kärenlampi L,Skärby L (eds) (1996). Op. cit.: 193-197.

Ashmore MR, Wilson RB (eds) (1993) Criticallevels of Air Pollutants for Europe.Background Papers prepared for the ECEWorkshop on critical levels, Egham, UK:23-26 March 1992.

Basu PS, Sharma A, Garg ID, Sukumaran NP(1999) Tuber sink modifies photosyntheticresponse in potato under water stress.Environmental and Experimental Botany42: 25-39.

Baumgarten M, Werner H, Häberle KH,Emberson L, Fabian P, Matyssek R (2000)Seasonal ozone exposure of maturebeech trees (Fagus sylvatica) at high alti-tude in the Bavarian forest (Germany) incomparison with young beech grown inthe field and in phytotrons. EnvironmentalPollution 109: 431-442.

Bender J, Bergmann E, Dohrmann A, TebbeCC, Weigel HJ (2002) Impact of ozone onplant competition and structural diversityof rhizosphere microbial communities ingrassland mesocosms. Phyton 42: 7-12.

Benton J, Fuhrer J, Sanchez-Gimeno B,Skärby L, Sanders GE (1995) Results fromthe UN/ECE ICP-Crops indicate theextent of exceedance of the critical levelsof ozone in Europe. Water, Air and SoilPollution 85: 1473-1478.

Bergmann E, Bender J, Weigel HJ (1996)Ozone and natural vegetation: Nativespecies sensitivity to different ozoneexposure regimes. In: Fuhrer J,Achermann B (eds) (1999): Op. cit.: 205-209.

Bermejo V (2002) Efectos del ozono sobre laproducción y la calidad de frutos deLycopersicon esculentum. Modulaciónpor factores ambientales. Facultad deCiencias, Departamento de Biología,Universidad Autónoma de Madrid. PhDThesis.

Broekhuizen S (1969) Atlas of the cereal-growing areas in Europe. Pudoc, Centerfor Agricultural Publishing andDocumentation, Wageningen, TheNetherlands.

Bunce JA (2000) Responses of stomatalconductance to light, humidity and tem-perature in winter wheat and barley grownat three concentrations of carbon dioxidein the field. Global Change Biology 6: 371-382.

Calvo E (2003) Efectos del ozono sobre algu-nas hortalizas de interés en la cuencamediterránea occidental. Universitat deValencia. PhD Thesis.

Danielsson H (2003) Exposure, uptake andeffects of ozone. Department ofEnvironmental Science, GöteborgUniversity, Sweden. PhD Thesis.

Danielsson H, Pihl Karlsson G, Karlsson PE,Pleijel H (2003) Ozone uptake modellingand flux-response relationships – anassessment of ozone-induced yield lossin spring wheat. AtmosphericEnvironment 37: 475-485.

Dwelle RB, Hurley PJ, Pavek JJ (1983)Photosynthesis and stomatal conductac-ne of potato clones (Solanum tuberosumL.). Plant Physiology 72: 172-176.

Emberson LD (1997) Defining and mappingrelative potential sensitivity of Europeanvegetation to ozone. Imperial College,University of London. PhD Thesis.

References

Emberson LD, Ashmore MR, Cambridge HM,Simpson D, Tuovinen JP (2000a)Modelling stomatal ozone flux acrossEurope. Environmental Pollution 109: 403-413.

Emberson L, Simpson D, Tuovinen JP,Ashmore MR, Cambridge HM (2000b)Towards a model of ozone deposition andstomatal uptake over Europe. EMEPMSC-W Note 6/2000.

Franzaring J, Tonneijck AEG, Kooijman AWN,Dueck TA (2000) Growth responses toozone in plant species from wetlands.Environmental and Experimental Botany44: 39-48.

Fuhrer J, Ashmore MR, Mills G, Hayes F,Davison AW (2003) Critical levels for semi-natural vegetation. In: Karlsson PE,Selldén G, Pleijel H (eds) (2003a): Op. cit.

Fuhrer J (1995) Integration of the ozone crit-ical level approach in modelling activities.Background paper to the EMEPWorkshop on the Control ofPhotochemical Oxidants over Europe, 24-27 October 1995 in St. Gallen,Switzerland. Federal Office ofEnvironment, Forest and Landscape,Berne.

Fuhrer J (1994) The critical level for ozone toprotect agricultural crops – An assess-ment of data from European open-topchamber experiments. In: Fuhrer J,Achermann B (eds) (1994): Op. cit.: 42-57.

Fuhrer J, Achermann B (eds) (1999) CriticalLevels for Ozone – Level II. Swiss Agencyfor the Environment, Forests andLandscape, Berne. EnvironmentalDocumentation No. 115.

Fuhrer J, Achermann B (eds) (1994) CriticalLevels for Ozone. UNECE WorkshopReport, Schriftenreihe der FAC Berne-Liebefeld.

Fuhrer J, Skärby L, Ashmore MR (1997)Critical levels for ozone effects on vegeta-tion in Europe. Environmental Pollution97(1-2): 91-106.

Gelang J, Pleijel H, Sild E, Danielsson H,Younis S, Sellden G (2000) Rate and dura-tion of grain filling in relation to flag leafsenescence and grain yield in springwheat (Triticum aestivum) exposed to dif-ferent concentrations of ozone.Physiologia Plantarum 110: 366-375.

Gimeno BS, Bermejo V, Sanz J, De La TorreD, Gil JM (2003a) Ambient ozone levelsinduce adverse effects on the flower pro-duction of three clover species fromIberian Rangelands. In: Karlsson PE,Selldén G, Pleijel H (eds) (2003a): Op. cit.

Gimeno BS, De la Torre D, Gonzalez A,Lopez A, Serra J (2003b) Determination ofweighting factors related with soil wateravailability to assess ozone impact onMediterranean wheat crops (T. aestivumL.). In: Karlsson PE, Selldén G, Pleijel H(eds) (2003a): Op. cit.

Gimeno BS, Bermejo V, Sanz J, De la Torre D,Gil JM (2004) Assessment of the effects ofozone exposure and plant competition onthe reproductive ability of three thero-phytic clover species from Iberian pas-tures. Atmospheric Environment (inpress).

Gollan T, Passioura JB, Munns R (1986) Soilwater status effects the stomtal conduc-tance of fully turgid wheat and sunflowerleaves. Australian Journal of PlantPhysiology 13: 459-464.

Grünhage L, Krause GHM, Köllner B, BenderJ, Weigel HJ, Jäger HJ, Guderian R (2001)A new flux-orientated concept to deriveCritical Levels for ozone to protect vege-tation. Environmental Pollution 111: 355-362.

Gruters U, Fangmeier A, Jager HJ (1995)Modelling stomatal responses of springwheat (Triticum aestivum L. cv. Turbo) toozone at different levels of water supply.Environmental Pollution 87: 141-149.

Hassan IA, Bender J, Weigel HJ (1999)Effects of ozone and drought stress ongrowth, yield and physiology of tomatoes(Lycopersicon esculentum Mill. Cv.Baladey). Gartenbauwissenschaft 64:152-157.

Hodges T, Ritchie J (1991) The CERES-Wheat phenology model. In: Hodges T(ed.), Predicting crops phenology, CRCPress, Boca Raton, FL: 131-141.

Jeffries RA (1994) Drought and chlorophyllfluorescence in field-grown potato(Solanum tuberosum). PhysiologiaPlantarum 90: 93-97.

Jones HG (1992) Plants and microclimate: Aquantitative approach to environmentalplant physiology (2nd edition). CambridgeUniversity Press, Cambridge.





Kärenlampi L, Skärby L (eds) (1996) Criticallevels for ozone in Europe: testing andfinalising the concepts. UNECE WorkshopReport, University of Kuopio, Departmentof Ecology and Environmental Science.

Karlsson PE, Selldén G, Pleijel H (eds)(2003a) Establishing Ozone Critical LevelsII. UNECE Workshop Report, IVL report B1523, IVL Swedish EnvironmentalResearch Institute, Gothenburg, Sweden.http://www.ivl.se

Karlsson PE, Uddling J, Braun S,Broadmeadow M, Elvira S, Gimeno BS, LeThiec D, Oksanen E, Vandermeiren K,Wilkinson M, Emberson L (2003b) NewCritical Levels for Ozone Impact on TreesBased on AOT40 and Leaf CumulatedUptake of Ozone. In: Karlsson PE, SelldénG, Pleijel H (eds) (2003a): Op. cit.

Karlsson PE, Uddling J, Braun S,Broadmeadow M, Elvira S, Gimeno BS, LeThiec D, Oksanen E, Vandermeiren K,Wilkinson M, Emberson L (2004) New crit-ical levels for ozone effects on youngtrees based on AOT40 and simulatedcumulative leaf uptake of ozone.Atmospheric Environment (in press).

Klumpp A, Ansel A, Klumpp G, Belluzzo N,Calatayud V, Chaplin N, Garrec JP,Gutsche HJ, Hayes M, Hentze HW,Kambezidis H, Laurent O, Peñuelas J,Rasmussen S, Ribas A, Ro-Poulsen H,Rossi S, Sanz MJ, Shang H, Sifakis N,Vergne P (2002) EuroBionet: A pan-European biominotoring network forurban air quality assessment.Environmental Science and PollutionResearch 9: 199-203.

Korner C, Scheel JA, Bauer H (1979)Maximum leaf diffusive conductance invascular plants. Photosynthetica 13(1):45-82.

Krause GHM, Kollner B, Grünhage L (2003)Effects of ozone on European forest treespecies – a concept of local risk evalua-tion within ICP-Forests. In: Karlsson PE,Selldén G, Pleijel H (eds) (2003a): Op. cit.

Ku SB, Edwards GE, Tanner CB (1977)Effects of light, carbon dioxide and tem-perature on photosynthesis, oxygen inhi-bition of photosynthesis, and transpira-tion in Solanum tuberosum. PlantPhysiology 59: 868-872.

Machado EC, Lagôa AMMA (1994) Trocasgasosas e condutancia estomatica emtres especies de gramineas. BragantiaCampinas 53(2): 141-149.

Manninen S, Siivonen N, Timonen U,Huttunen S (2003) Differences in ozoneresponse between two Finnish wild straw-berry populations. Environmental andExperimental Botany 49: 29-39.

Marshall B, Vos J (1991) The relationshipbetween the nitrogen concentration andphotosynthetic capacity of potato(Solanum tuberosum L.) leaves. Annals ofBotany 68: 33-39.

McLean DC, Schneider RE (1976)Photochemical Oxidants in Yonkers, NewYork: Effects on Yield of Bean andTomato. Journal of Environmental Quality5: 75-78.

McNaughton KG, Van der Hurk BJJM (1995)‘Lagrangian’ revision of the resistors inthe two- layer model for cal-culating the energy budget of a plantcanopy. Boundary Layer Meteor 74: 261-288.

Mills GE, Ball GR (1998) Annual ProgressReport for the ICP-Crops (September1997-August 1998). ICP-CropsCoordination Centre, CEH Bangor, UK.

Mills G, Holland M, Buse A, Cinderby S,Hayes F, Emberson L, Cambridge H,Ashmore M, Terry A (2003) Introducingresponse modifying factors into a riskassessment for ozone effects on crops inEurope. In: Karlsson PE, Selldén G, PleijelH (eds) (2003a): Op. cit.

Morgan JA (1984) Interaction of water supplyand N in wheat. Plant physiology 76: 112-117.

Novak K, Skelly JM, Schaub M, Kräuchi N,Hug C, Landolt W, Bleuler P ( 2003) Ozoneair pollution and foliar injury developmenton native plants of Switzerland.Environmental Pollution (in press).

Nussbaum S, Geissmann M, Fuhrer J (1995)Ozone exposure-response relationshipsfor mixtures of perennial ryegrass andwhite clover depend on ozone exposurepatterns. Atmospheric Environment 29(9),989-995.

Oshima RJ, Taylor OC, Braegelmann PK,Baldwin DW (1975) Effect of ozone on theyield and plant biomass of a commercialvariety of tomato. Journal of

Environmental Quality 4: 463-464.

Peterson RF (1965) Wheat: Botany, cultiva-tion, and utilization. Leonard Hill Books,London.

Pihl Karlsson G, Karlsson PE, Danielsson H,Pleijel H (2003) Clover as a tool forbioindication of phytotoxic ozone – 5years of experience form SouthernSweden – consequences for the short-term Critical Level. Science of the TotalEnvironment 301: 1-3, 205-213.

Pihl Karlsson G, Soja G, Vandermeiren K,Karlsson PE, Pleijel H (2004) Test of theshort-term Critical Levels for acute ozoneinjury on plants – improvements by ozoneuptake modelling and the use of an effectthreshold. Atmospheric Environment, inpress.

Pleijel H (1996) Statistical aspects of CriticalLevels for ozone. In: Kärenlampi L, SkärbyL (eds) (1996): Op. cit.

Pleijel H, Danielsson H, Ojanperä K, DeTemmerman L, Högy P, Karlsson PE(2003) Relationships between ozoneexposure and yield loss in Europeanwheat and potato – A comparison of con-centration based and flux based expo-sure indices. In: Karlsson PE, Selldén G,Pleijel H (eds) (2003a): Op. cit.

Pleijel H, Danielsson H, Vandermeiren K,Blum C, Colls J, Ojanperä K (2002)Stomatal conductance and ozone expo-sure in relation to potato tuber yield –results from the European CHIP pro-gramme. European Journal of Agronomy17: 303-317.

Power SA, Ashmore MR (2002) Responses offen and fen meadow communities toozone. New Phytologist 156: 399-408.

Reinert RA, Eason G, Barton J (1997) Growthand fruiting of tomato as influenced byelevated carbon dioxide and ozone. NewPhytologist 137: 411-420.

Ribas A, Peñuelas J (2003) Biomonitoring oftropospheric ozone phytotoxicity in ruralCatalonia. Atmospheric Environment 37:63-71.

Skärby L, Karlsson PE (1996) Critical levelsfor ozone to protect forest trees - bestavailable knowledge from the Nordiccountries and the rest of Europe. In:Kärenlampi L, Skärby L (eds) (1996): Op.cit.

Stark JC (1987) Stomatal behaviour of pota-toes under non-limiting soil water condi-tions. American Potato Journal 64: 301-309.

Temple PJ (1990) Growth and yield respons-es of processing tomato (Lycopersiconesculentum) cultivars to ozone.Environmental and Experimental Botany30: 283-291.

Temple PJ, Surano KA, Mutters RG,Bingham GE, Shinn JH (1985) Air Pollutioncauses moderate damage to tomatoes.California Agriculture: 21-23.

Tuebner F (1985) Messung derPhotosynthese und Transpiration anWeizen, Kartoffel und Sonnenblume.Diplomarbeit University Bayreuth, WestGermany.

Uddling J, Pleijel H, Karlsson PE (2004)Modelling leaf diffusive conductance injuvenile silver birch, Betula pendula.Agricultural and Forest Meteorology (inpress).

UNECE (1988) Final Draft Report of theCritical Levels Workshop, Bad Harzburg,Germany, 14-18 March 1988. Available atFederal Environmental Agency Berlin, c/oDr. Heinz Gregor, Germany.

UNECE (1995) Effects of Nitrogen andOzone. Report of the InternationalCooperative Programmes and theMapping Program of the Working Groupon Effects. EB.AIR/WG.1/R.110.

UNECE (1996) Manual on methodologiesand criteria for mapping criticallevels/loads and geographical areaswhere they are exceeded. Texte 71/96,Umweltbundesamt, Berlin, Germany.

Van der Heyden D, Skelly J, Innes J, Hug C,Zhang J, Landolt W, Bleuler P (2001)Ozone exposure thresholds and foliarinjury on forest plants in Switzerland.Environmental Pollution 11: 321-331.

Vos J, Groenwald J (1989) Characteristics ofphotosynthesis and conductance of pota-to canopies and the effects of cultivar andtransient drought. Field Crops Research20: 237-250.

Vos J, Oyarzun PJ (1987) Photosynthesisand stomatal conductance of potatoleaves - effects of leaf age, irradiance,and leaf water potential. PhotosynthesisResearch 11: 253-264.





Weber P, Rennenberg H (1996) Dependencyof Nitrogen dioxide (NO2) fluxes to wheat

(Triticum aestivum L.) leaves from NO2concentration, light intensity, temperatureand relative humidity determined fromcontrolled dynamic chamber experi-ments. Atmospheric Environment 30(17):3001-3009.

WHO (2000) Air Quality Guidelines forEurope. WHO Regional Publications, No.91, World Health Organisation,Copenhagen. Available athttp://www.who.dk/

Zadoks JC, Chang TT, Konzak CF (1974) Adecimal code for the growth stages ofcereals. Weed Research 14: 415-421.

Zhang J, Xiangzhen S, Li B, Su B, Li J, ZhouD (1998) An improved water-use efficiencyfor winter wheat grown under reduced irri-gation. Field Crops Research 59: 91-98.

Atmospheric pollution is an important factorin material deterioration including degrada-tion of systems used for material protection.Due to pollution, the lifetime of technologicalproducts is shortened. Buildings, otherstructures, as well as objects of cultural her-itage exposed to the atmosphere deterioratemore rapidly. The resulting physico-chemicaland economic damage can be significant -not to mention the loss of unique parts of ourcultural heritage and hazards due to endan-gered reliability of complicated technologi-cal devices.

devices. Also, as the result of weatheringdue to especially acidifying pollutants, a sig-nificant part of the metals used in construc-tions and products are emitted to the bios-phere with a potential hazard to the environ-ment.

This part of the manual was revised in con-nection with the UNECE Workshop onMapping Air Pollution Effects on Materials,Including Stock at Risk, held in Stockholm,Sweden on 14-16 June 2000 (UNECE 2000).The present revision has been performed bythe UNECE ICP on Materials with supportfrom ICP Modelling and Mapping, based onresults obtained at evaluation of deteriora-tion of materials after 8-year field exposure(Tidblad, Kucera, Mikhailov 1998) his chapterconsiders the corrosive effects of gaseousSO2, NOx, ozone and acid rainfall in combina-tion with climatic parameters. It aims todefine procedures for mapping “acceptablevalues” of pollutants for buildings and mate-rials including cultural and historical monu-ments in an analogous way to the methodsdefined elsewhere for critical levels andloads for natural ecosystems.

Because atmospheric deterioration of mate-rials is a cumulative, irreversible process,which proceeds even in the absence of pol-lutants, “critical” values are not as easilydefined as for some natural ecosystems.Some rate of deterioration must be definedwhich may be considered “acceptable”

based on technical and economic consider-ations. This approach provides the basis formapping “acceptable areas” for corrosionand deriving areas where the acceptablepollution level/load is exceeded, in an analo-gous way to the maps produced for naturalecosystems.

Acceptable load/level. The “acceptablelevel or load” of pollutants for buildings andmaterials is the concentration or load whichdoes not lead to an unacceptable increase inthe rate of corrosion or deterioration.

Acceptable rate of corrosion or deterioration(Ka) may be defined as the corrosion, whichis considered “acceptable” based on techni-cal and economic considerations. In realitythis may not be considered practical.Therefore, it is recommended that theacceptable corrosion rate should beexpressed in terms of average corrosionrates in areas with “background” pollution.Within the UNECE ICP on Materials it wasdecided to recommend the background cor-rosion or deterioration rate (Kb) as the lower10 percentile of the observed corrosion ratesin the materials exposure programme whichstarted in 1987 and ended in 1995.Acceptable corrosion rates are then definedas a multiple (n) of the background corrosionrate

(4.1)

Ka = n · Kb

It is possible to relate such rates of corrosionto the “lifetimes” or economic values ofmaterials.

Dose-response function. The relationshipbetween the corrosion or deterioration rateand the levels or loads of pollutants in com-bination with climatic parameters.

Using the above definitions it is possible tocalculate the acceptable pollution level fromthe acceptable corrosion rate and a dose-response function which relates corrosionrate to pollutant and climate exposure.

4 Mapping of Effects on Materials

4.1 Introduction - objectives, defini-tions and general remarks

Mapping Manual 2004 • Chapter iV Mapping of Effects on Materials Page IV - 1


Mapping Manual 2004 • Chapter IV Mapping of Effects on MaterialsPage IV - 2

4.2.1 Introduction

Deterioration rates can be calculated usingdose-response functions. The functions rec-ommended have been derived from the fieldresearch programme undertaken as part ofthe UNECE ICP Materials ExposureProgramme. However, if other functions ofthe same or of other form are consideredmore suitable, then they can be used as analternative. The range of values over whichthe dose-response function is consideredvalid must be considered. At this time, scien-tific evidence from field exposure is limited,in contrary to laboratory experiments. Even ifNOx is not included in the dose-responsefunctions included in this manual it may beincluded for some materials in the future.

The following equations are based on 8-yearresults from the exposure within UNECE ICP

on Materials. They reflect the increased levelof the physico-chemical understanding of

the corrosion mechanisms including the syn-ergistic effects of SO2 and O3 in the case ofcopper. The calculation of an acceptablelevel for sulfur dioxide and ozone for materials where the synergistic effect withSO2 is included in the dose-response function is described in 4.2.3.The impact of wet deposition of acidicspecies on sensitive materials is in this revision of the Mapping Manual consideredas an effect of the total load of H+ deposition.The calculation and mapping of the acceptable load for materials where the H+

load is included in the dose-response func-tion is described in 4.2.3.

4.2.2 Dose-response functions

The equations for the following materials areat present available for mapping purposes.The equations are valid for unshelteredexposure of materials (Tidblad, Kucera,Mikhailov 1998, see also the ICP website: www.corrinstitute.se/ICPMaterials).

4.2 Mapping “acceptable corrosion

rates” and “acceptable

levels/loads” for pollutants

Structural metals r2 N eq.

Weathering steel (C<0.12%, Mn 0.3-0.8%, Si 0.25-0.7%, P 0.07-0.15%,

S<0.04%, Cr 0.5-1.2%, Ni 0.3-0.6%, Cu 0.3-0.55%, Al<0.01%)

ML = 34[SO2]0.33exp{0.020Rh34[SO2]

0.13exp{0.020Rh + f(T)}t0.33 0.68 148 (4.2)

f(T) = 0.059(T-10) when T≤10°C, otherwise -0.036(T-10)

Zinc

ML = 1.4[SO2]0.22exp{0.018Rh + f(T)}t0.85 + 0.029Rain[H+]t 0.84 98 (4.3)


Aluminium

ML = 0.0021[SO2]0.23Rh·exp{f(T)}t1.2 + 0.000023Rain[Cl-]t 0.74 106 (4.4)


Copper

ML = 0.0027[SO2]0.32[O3]

0.79Rh·exp{f(T)}t0.78 + 0.050Rain[H+]t0.89 0.73 95 (4.5)


where

ML = mass loss, g m-2

R = surface recession, µm

t = exposure time, years

L = maintenance interval (life time), years

Rh = relative humidity, % -

annual average

T = temperature, °C - annual average

[SO2] = concentration, µg m-3 -

annual average

[O3] = concentration, µg m-3 -

annual average

Rain = amount of precipitation, m year-1 - annual average

[H+] = concentration, mg l-1 -

annual average

[Cl-] = concentration, mg l-1 -

annual average


Mapping Manual 2004 • Chapter IV Mapping of Effects on Materials Page IV - 3

Bronze (Cu Sn6Pb7Zn5, ISO/R 1338 (Cu 81%, Sn 5.8%, Pb 6.7%,

Zn 4.5%, Ni 1.6% + trace elements)

ML = 0.026[SO2]0.44Rh·exp{f(T)}t0.86 + 0.029Rain[H+]t0.76

+ 0.00043Rain[Cl-]t0.76 0.81 144 (4.6)


Stone materials

Limestone

R = 2.7[SO2]0.48exp{-0.018T}t0.96 + 0.019Rain[H+]t0.96

0.88 100 (4.7)

Sandstone (White Mansfield dolomitic sandstone)

R = 2.0[SO2]0.52exp{f(T)}t0.91 + 0.028Rain[H+]t0.91

0.86 101 (4.8)

f(T) = 0 when T≤10°C, otherwise -0.013(T-10)

Paint coatings

Coil coated galv. steel with alkyd melamine

L = [5 / ( 0.084[SO2] + 0.015Rh + f(T) + 0.00082Rain )]1/0.43 0.73 138 (4.9)


Steel panels with alkyd

L = [5 / ( 0.033[SO2] + 0.013Rh + f(T) + 0.0013Rain)]1/0.41 0.68 139 (4.10)




The equations are valid for regions withoutstrong influence of sea salts with a chloridecontent in precipitation < 5 mg l-1 approx.

The functions for paint coatings areexpressed as lifetime equations. These life-times can be mapped but the functions cannot be used for calculating acceptable levels/loads using the concept of acceptablecorrosion rates.

For all these materials, however, also equations for exposure in sheltered posi-tions are available. Also equations are avail-able for glass materials representative ofmedieval stained glass windows, for paint

coatings and for the electric contact materials nickel and tin (Tidblad, Kucera,Mikhailov 1998).

The preferably recommended value of thebackground corrosion rate for individualmaterials is given in Table 4.1. In order tosimplify the procedure of obtaining accept-able levels/loads as described in section4.2.3 the rates in Table 4.1 are expressed fora 1-year exposure period. An analysis of thedose-response functions shows that usinglonger exposure periods than 1 year leads inprinciple to very similar acceptablelevels/loads.

Table 4.1: Background corrosion rates Kb, expressed for a 1-year exposure period (1997-98), for differentmaterials based on results from the UNECE ICP Materials exposure programme after 1, 2, 4 and 8 years ofexposure.

Material 1-year background corrosion rate, Kb

Weathering steel 72 g m-2

Zinc 3.3 g m-2

Aluminium 0.09 g m-2

Copper 3.0 g m-2

Bronze 2.1 g m-2

Limestone 3.2 µm

Sandstone 2.8 µm

4.2.3 Calculation and mapping ofacceptable levels/loads and theirexceedances

Mapping will produce a map which depictsthe mapping units that exceed the accept-able deterioration rate (Ka) for simple sam-ples of material used in particular locationsexpressed in a number of defined classes. Itmakes no correction for the form in whichthe material is used (e.g. the type of compo-nent). At present it is recommended that nvalues 1.5, 2.0 and 2.5 are used (see eq. 4.1).It should, however, be emphasised thatlower n values can be used for, e.g.

areas with valuable objects of cultural her-itage.

The data requirements are annual mean SO2and O3 concentrations in µg m-3, the temper-ature in °C, the relative humidity in %, thetotal rainfall in mm year-1, and the concentra-tion of H+ and Cl- in mg/l for each mappingunit (e.g. the EMEP grid size 50 x 50 km orsmaller, preferably as a fraction of the EMEPsize). In addition, the value for the background corrosion rate is required. Thevalues are then used to calculate the corro-sion rate and the corrosion rate as a ratio of

the background corrosion rate. All mappingunits that exceed the acceptable corrosionrate (n · Kb) are then identified.

In each mapping unit that exceeds theacceptable corrosion rate, the same datasets and dose-response functions are thenused to calculate the

• SO2 concentration• O3 concentration (for copper)• H+ load

that keeps this rate at an acceptable level.

The equations have the following form, illustrated on the example of copper. For theother materials the equations for SO2concentration and H+ load are created fromthe dose-response relations in an analogousway, except that O3 is not included in thefunctions. The basis for the calculationsshould be one year of exposure (t=1) sinceKb is expressed for this period.



(4.11)

(4.12)

(4.13)

where

[SO2]accCu = acceptable concentration for corrosion of copper at given values of other parameters, µg m-3

[O3]accCu = acceptable concentration for corrosion of copper at given values of other parameters, µg m-3

Rain[H+]accCu = acceptable value of H+ load for corrosion of copper at given values of other parameters (product of amount of precipitation, mm year-1, and concentra-tion of H+, mg l-1)

n = acceptable corrosion/deterioration rate (see eq. 4.1)

background corrosion/deterioration rate

and the other symbols as in point 4.2.2.

[SO2]accCu = [(3.0·n - 0.050Rain[H+]) / (0.0027[O3]

0.79Rh·exp{f(T)})]

1/0.32

[O2]accCu[O3]accCu = [(3.0·n - 0.050Rain[H+]) / (0.0027[SO2]

0.32Rh·exp{f(T)})]

1/0.79

(Rain[H+])accCu = (3.0·n - 0.0027[SO2]

0.32[O3]

0.79Rh·exp{f(T)}) / 0.050



The values of the climatic and pollution datafor each grid square can usually be obtainedfrom national or international meteorologicalcentres, international organisations e.g.(WHO), international research programmes(e.g. EMEP) or national organisations orauthorities responsible for environmentalprotection.

When data on ozone concentrations are notavailable, but concentrations of NO2 areknown, a crude approximation of ozone val-ues can be obtained using one of the follow-ing empirical equations, which are based ondata from the UNECE ICP MaterialsProgramme:

(4.14)

(4.15)

where

[O3] = average annual concentration, µg m-3

[NO2] = average annual concentration, µg m-3

Sun = sunshine duration, h year-1

Eq. 4.14 is based on values from the five firstyears of exposure (1987-1992) while eq. 4.15is based on values from all eight one-yearexposure periods (1987-1995). Eq. 4.15 ispreferred but 4.14 may be used if data onsunshine duration is not available. It shouldbe stressed that the equations are empiricaland obtained in the absence of a theory fromwhich a relation between ozone and nitrogendioxide concentrations could be derived.More importantly, the equations are basedon a set of NOx and ozone concentrationdata that include inner-city data. The latterare characterised by very high NOx and lowO3 , which is a situation not found inrural/remote areas that are of most interestfor the other parts of the mapping pro-

gramme and in the general framework of theConvention on Long-range TransboundaryAir Pollution. If only those values typical forrural/remote areas were used, a differentrelationship between O3 and NOx wouldemerge. Note that, according to eq. 4.14, along-term average ozone concentration larger than 57 µg m-3 would not be possible;however, this is a relatively low value inrural/remote areas.

4.2.4 Calculations and mapping of costs resulting from corrosion

The ultimate goal of the mapping activities inthe field of materials is the calculation ofcost of damage caused by air pollutants tomaterials. While it is not recommended toassess the absolute cost, it is possible toestimate the difference in cost between twoalternative scenarios using the equation

(4.16)

where DC is the cost difference, C is the costper surface area of material, for maintenance/replacement, S is the surfacearea of material, LpLs1 is the maintenanceinterval (life time) in polluted areas and Lc forscenario 1 and Ls2 is the maintenance interval for scenario 2. If eq. 4.16 is to beused for estimating the cost of corrosion dueto pollutants in the present situation it is inclean areas. recommended to use the pres-ent pollution situation as scenario 1 and thebackground corrosion as scenario 2. Whencalculating the lifetime from the backgroundscenario it is recommended to use the timedependence from the dry deposition termgiven in eqs. 4.2 - 4.8, for example t0.85 forzinc. For the paint coatings it is not recom-mended to use eqs. 4.9 - 4.10 for cost calcu-lations since the evaluation are based ondamage from an intentionally made scratch,which may be regarded as a form of acceler-ated testing.

[O3] = 57·exp{-0.012[NO2]}

[O3] = (38 + 0.013Sun) · exp{(-0.022+

+0.000005Sun)[NO2]}

r² n

0.70 103

0.78 132

∆C = C · S · (Ls1-1

– Ls2-1

)

Instead the practical functions described inKucera et al. (1993) are recommended.

A UNECE Workshop on economic evaluationof air pollution abatement and damage tobuildings including cultural heritage tookplace in Stockholm on 23-25 January 1996(UNECE 1997). The proceedings of the work-shop summarised the state of the art, and isgiven in the reference list together with aselection of other important publications(Kucera et al. 1993, ECOTEC 1986, Tolstoy etal. 1989, Cowell, Apsimon 1996). In theECOTEC study of 1986, building identikitsfor Birmingham (UK), Dortmund and Koeln(Germany) were compiled (ECOTEC 1986).For Stockholm (Sweden), Sarpsborg(Norway), and Prague (Czech Republic) sta-tistically based inventories of outdoor mate-rial surfaces were compiled by Kucera et al.(1993). These studies include percentagevalues of the most common constructionmaterials and their distribution, which inabsence of stock at risk data may provision-ally be used as default values.

Mapping of cost of damage will be similar tomapping of acceptable levels/loads but willinclude data on the stock of material in eachmapping unit and the economic costs asso-ciated with deterioration of these materials(e.g. replacement or repair costs). The class-es to be used are based on multiples of theBackground Deterioration Rate. The changein the rate of deterioration can be used toestimate the cost associated with the deteri-oration. This can be used to estimate thecosts resulting from the deterioration and toundertake a cost benefit analysis of pollutantemission reduction scenarios in the mappingarea.

In addition to the data required for mappingacceptable levels/loads estimates of thetotal area of the respective building materialin each mapping unit and the cost forreplacement and/or maintenance of thematerial is required.

4.2.5 Sources of uncertainty

The main sources of uncertainty will resultfrom • use of yearly mean values of pollution

parameters which do not take into

account the effect of fluctuations of pollu-tant levels,

• use of dose-response functions, whichmay be further developed i.a. after theevaluation of results of the ongoing “multi-pollutant” exposure in ICP Materials,

• assessment of environmental data andtheir extrapolation to each mappingsquare taking into account the importanceof the difference in local environment inpopulated and rural areas,

• translation of dose-response functionsobtained by exposure of test specimens todamage functions taking into accountthe exposure situation on a constructionand rational maintenance practice.

• assessment of the stock of materials atrisk,

• characterisation of reduced service life-time, and

• quantification of costs associated withreduced service lifetime.

The effect of ozone on corrosion of materialsis complex and there are today serious gapsof knowledge. Both ozone and NO2 have adirect effect on corrosion and degradation ofespecially some organic materials. In recentyears the synergistic effect of SO2 in combi-nation with O3 and NO2 has been shown tolead to severely increased corrosion on sev-eral inorganic materials in laboratory expo-sures. In field exposures so far only the syn-ergistic effect of SO2 and O3 has been shownin the UNECE exposure programme.

Recent studies of historical ozone measure-ments indicate that background tropospher-ic ozone concentrations over Europe haveapproximately doubled since the end of the19th century. This change is due to increasedemissions of precursor compounds fromhuman activities in form of NOx and VOC(volatile organic compounds). In urban areasthe ozone concentrations are suppresseddue to local emissions of NOx primarily from motor vehicle exhausts. Motor vehicle emission controls will serve to


4.3 Direct effects of ozone




reduce local emissions of NOx and as a con-sequence ozone concentrations in urbanareas will increase. Over the last decade thebackground ozone concentrations haveincreased by up to a few percent per yearover northern and central Europe and are atpresent approximately 30 ppb. For sensitiveorganic materials exposed to outdooratmospheres an acceptable level for ozoneis proposed to be 20 ppb as an annual mean.It may provisionally represent a acceptableabove which unacceptable shortening of thematerial lifetime will result. It should, howev-er, be emphasised that for protection of sen-sitive organic materials in e.g. museums,galleries and archives much lower levels forozone are recommended.

Any user who are considering to producemaps on effect of materials based on proce-dures specified in this chapter should alsoconsult the general chapters of this manual,Chapter 1 Introduction and, especially,Chapter 2 Guidance on MappingConcentration Levels and Deposition Loads.In chapter 2 the general methods of mapping, their underlying assumptions anddata requirements are given for many of theparameters needed for mapping effects onmaterials. Regarding the wet depositionparameters it should be noted that the termRain[Cl-] given in this chapter is identical tothe chloride wet deposition parameterdescribed in chapter 2. Deposition of protones (Rain[H+]) is not mentioned explicitly in chapter 2 but it should be notedthat the recommendation that maps of annual wet deposition should not be drawnby interpolating measured wet depositionbut instead by combining maps of interpolated solute concentrations and precipitation amounts separately is vialidalso for H+.

For materials damage the following types ofmaps are recommended. For recommended n values see 4.2.3.

Assisting maps

These maps illustrates the geographical distribution of some of the most importantenvironmental parameters and their combinations used in the dose-responsefunctions:

• [SO2]

• [O3]

• T

• Rh

• Rain

• pH

• Cl-

• Rain[H+]

Material damage maps (for each selectedmaterial)

Maps for present value of corrosion effects(based on dose-response functions):

• Exceedance of acceptable corrosion rate quantifying the degree of exceedance inclasses Ka/(n·Kb), preferably for n = 1.5,2.0 and 2.5.

• Contribution of wet deposition to the total corrosion effect or ratio between corro-sion effect of wet and dry deposition

Maps for different scenarios:

• Comparison between present corrosionrate with the rate at different levels of[SO2], [O3] and [H+]

Acceptable pollution levels/loads maps (foreach selected material)

• Acceptable SO2 levels for present level of[O3] and [H+]

• Acceptable O3 levels for copper for pre-sent level of [SO2] and [H+]

• Acceptable H+ loads for present level of[SO2] and [O3]

4.4 Data and mapping procedure

Maps for different scenarios:

• Acceptable SO2, O3 (for copper) levels or H+ loads for different levels/loads of thetwo other pollutant parameters.

4.4.1 Data and scale of mapping

It is recommended that countries make useof best available data and generate informa-tion at appropriate national scales. Thesedata when incorporated into European mapswill be mapped for EMEP grid areas (50 kmcurrently). It is thus advisable to use a gridnetwork which coincide with the EMEP network or if smaller (which is preferable) isa fraction of the EMEP size.

4.4.2 Urban and rural areas

It is recognised that urban areas differ fromrural especially in levels of pollutants. Alsorelevant climatic data as temperature andrelative humidity may show variationsbetween urban and rural areas. Having inmind that corrosion of materials is affectedboth by long-range transported pollutantsand by more local emissions it is advisableto produce maps with considerably smallergrids especially for mapping of effects inurban areas.





Cowell, Apsimon H (1996) Estimating thecost of damage to buildings by acidifyingatmospheric pollution in Europe, Atmos.Env., Vol. 30, pp. 2959-2968.

ECOTEC (1986) Identification andAssessment of Materials Damage toBuildings and Historic Monuments by AirPollution. Report to the UK Department ofthe Environment. ECOTEC ConsultingLtd., Birmingham.

Kucera V, Henriksen J, Knotkova D, SjöströmC (1993) Model for Calculations ofCorrosion Costs Caused by Air Pollutionand its Applications in three Cities, Proc.10th European Corros. Congr., Barcelona.

Tidblad J, Kucera V, Mikhailov AA (1998)UNECE International Co-operativeProgramme on Effects on MaterialsIncluding Historic and CulturalMonuments. Report No 30. Statisticalanalysis of 8-year materials exposure andacceptable deterioration and pollutionlevels. Swedish Corrosion Institute,Stockholm, Sweden.

Tolstoy N, Andersson G, Sjöström C, KuceraV (1989) External building materials –quantities and degradation. Statens insti-tut för byggnadsforskning, Gävle. ISBN91-540-9315-5.

UNECE (1997) Economic Evaluation of AirPollution Damage to Materials.Proceedings of the UNECE Workshop onEconomic Evaluation of Air PollutionAbatement and Damage to Buildingsincluding Cultural Heritage, Eds V.Kucera, D. Pearce and Y.-W. Brodin,Report 4761, Swedish EnvironmentalProtection Agency, Stockholm, Sweden,1997.

UNECE (2000) Mapping Air Pollution Effectson Materials, including Stock at Risk, EB.AIR/WG.1/2000/15, Workshop Reportpre pared by the Chairman of the TaskForce on the International CooperativeProgramme on Effects of Air Pollution onMaterials, Including Historic and CulturalMonuments and by the Chairman of the

Task Force on the International Co-opera-tive Programme on Modelling andMapping.

References

��

��

��

�� !"��!��#�$��%��&�� "��'�� (�� )��*�� +�,�!��%��&�� "-��.�� !"��#�,��/��"��0�� !��0!#�.��1��"��!��2��! ��0��0!��3�� !"��2��4��5��/��"-6��7��%&"-��2��"��+�#��0��/��"�1� ��0!��6��(��1��7��$��!��8��2��

The general definition of a critical load is“a quantitative estimate of an exposure toone or more pollutants below which signifi-cant harmful effects on specified sensitiveelements of the environment do not occuraccording to present knowledge”.

This definition applies to different receptors(e.g., terrestrial ecosystems, groundwater,aquatic ecosystems, and/or human health).‘Sensitive elements’ can be part or whole ofan ecosystem or of ecosystem developmentprocesses, such as their structure and func-tion. Critical loads have been defined forseveral pollutants and effects resulting fromtheir deposition.

Critical loads of sulphur and nitrogen acidityfor an ecosystem have been specificallydefined at the Skokloster Workshop as(Nilsson and Grennfelt 1988):“the highest deposition of acidifying compounds that will not cause chemicalchanges leading to long-term harmfuleffects on ecosystem structure and function.”Both sulphur and nitrogen compounds contribute to the total deposition of acidity.The acidity input has to be considered in thisbalance regardless whether it is due to S orN depositions. Thus, the ratio between sulphur and nitrogen may vary withoutchange in the acidity load (see section 5.3.2for details).

In addition to acidification, inputs of nitrogenmay influence the eutrophication and nutrient balances of ecosystems. The criticalload for nitrogen nutrition effects is definedas:“the highest deposition of nitrogen as NHxand/or NOy

1 below which harmful effects inecosystem structure and function do notoccur according to present knowledge.”Critical loads for heavy metals are definedaccordingly.

The basic idea of the critical load concept isto balance the depositions which an eco-

system is exposed to with the capacity ofthis ecosystem to buffer the input (e.g. theacidity input buffered by the weatheringrate), or to remove it from the system (e.g.nitrogen by harvest) without harmful effectswithin or outside the system.

In the context of a multi-pollutant multi-effects approach it is desirable to considerall effects simultaneously as far as possible.For eutrophication and acidification, this hasbeen done using so-called critical load func-tions. These are described in detail in section 5.3.

Critical loads can be determined either bysteady-state methods or by dynamic modelswith varying degree of complexity. Since critical loads are steady-state quantities, theuse of dynamic models for the sole purposeof deriving critical loads is somewhat inade-quate. However, if dynamic models are usedto simulate the transition to a steady statefor the comparison with critical loads, carehas to be taken that the steady-state versionof the dynamic model is compatible with thecritical load model. Dynamic models aredealt with in a separate chapter (Chapter 6).

5 Mapping Critical Loads

5.1 Introduction

Mapping Manual 2004 • Chapter V Mapping Critical Loads Page V - 1

1 NHx = NH3 + NH4+; NOy = NO + NO2 + NO2

- + NO3-


Mapping Manual 2004 • Chapter V Mapping Critical LoadsPage V - 2

5.2.1 Empirical critical loads of nutrient nitrogen

5.2.1.1 Introduction

The emissions of ammonia (NHx) and nitro-gen oxides (NOy) have strongly increased inEurope in the second half of the 20th century.Because of short- and long-range transportof these nitrogenous compounds, atmos-pheric nitrogen (N) deposition has clearlyincreased in many natural and semi-naturalecosystems. The availability of nutrients isone of the most important abiotic factors,which determine the plant species composi-tion in ecosystems. Nitrogen is the limitingnutrient for plant growth in many natural andsemi-natural ecosystems, especially of oligotrophic and mesotrophic habitats. Mostof the plant species from such conditions areadapted to nutrient-poor conditions, andcan only survive or compete successfully onsoils with low nitrogen availability. In addition, the N cycle in ecosystems is complex and strongly regulated by biologi-cal and microbiological processes, and thusmany changes may occur in plant growth,inter-specific relationships and soil-basedprocesses as a result of increased deposi-tion of air-borne N pollutants.

The series of events which occurs when Ninputs increase in an area with originally lowbackground deposition rates is highly complex. Many ecological processes interact and operate at different temporaland spatial scales. As a consequence, highvariations in sensitivity to atmospheric nitrogen deposition have been observedbetween different natural and semi-naturalecosystems. Despite this diverse sequenceof events, the following main effect “categories” can be recognised:

(a) Direct toxicity of nitrogen gases and aerosols to individual species (see Ncritical levels);

(b) Accumulation of nitrogen compounds, resulting in increased N availability and changes of species composition;

(c) Long-term negative effect of ammonium and ammonia;

(d) Soil-mediated effects of acidification;

(e) Increased susceptibility to secondary stress and disturbance factors such as drought, frost, pathogens or herbivores.

Recent experimental evidence, and practicalfield experience in ecosystem restoration,suggests that, once the process of alteredspecies composition and increased N miner-alisation occurred, spontaneous recovery ofthe vegetation may happen only over verylong time scales, or with very active man-agement intervention to decrease nitrogenstatus and cycling. This emphasises theneed for caution in setting critical loads atwhich these major changes in vegetationcomposition and nitrogen cycling do notoccur.

5.2.1.2 Data

Within the Convention on Long-rangeTransboundary Air Pollution (LRTAP), empiri-cal procedures have been developed to setcritical loads for atmospheric N deposition.Empirical critical loads of N for natural andsemi-natural terrestrial ecosystems and wet-land ecosystems were firstly presented in abackground document for the 1992 work-shop on critical loads held under the UNECELRTAP Convention at Lökeberg (Sweden)(Bobbink et al. 1992). After detailed discus-sion before and during the meeting, the pro-posed values were set at that meeting(Grennfelt and Thörnelöf 1992). Additionalinformation from the period 1992–1995 wasevaluated and summarised in an updatedbackground paper (Bobbink et al. 1996) andpublished as Annex III in the previous version of the Mapping Manual (UBA 1996).The updated N critical loads were discussedand accepted at an expert meeting held inDecember 1995 in Geneva (Switzerland).They were also used in the Air QualityGuidelines for Europe (2nd edition) of the

5.2 Empirical Critical Loads

World Health Organisation (WHO 2000). Itbecame clear that considerable new insightsinto, and data on, the impacts of N deposi-tion on natural and semi-natural ecosystemshave become available since the compilationof the last values in the mid-1990s.Therefore, new information from the period1996–2002 on the impacts of increasednitrogen deposition on the structure andfunction of natural and semi-natural ecosys-tems was evaluated and evaluated in a fullyadapted background paper (Bobbink et al. 2003). The updated N critical loadswere discussed and approved by full consensus at the November 2002 expertmeeting held under the LRTAP Convention inBerne (Switzerland, Achermann andBobbink 2003) and are given in Table 5.1.

Approach:Based on observed changes in the structureand function of ecosystems, reported in arange of publications, empirical N criticalloads were evaluated for specific receptorgroups of natural and semi-natural ecosys-tems in both 1992 and 1996. In the 2002updating procedure a similar ‘empiricalapproach’ was used as for the earlier back-ground documents. For this purpose,European publications on the effects of N innatural and semi-natural ecosystems fromthe period 1996 to mid 2002 were collectedas completely as possible. Peer-reviewedpublications, book chapters, nationally published papers and ‘grey’ reports of institutes or organisations, if available byrequest, were incorporated. Results fromfield addition experiments and mesocosmstudies, from correlative or retrospectivefield studies, and, in few cases, dynamicecosystem modelling was relevant in thisrespect.

Ranges and reliability:As in the 1992 and 1996, the empirical Ncritical loads were established within a rangefor each ecosystem class, because of: (i) realintra-ecosystem variation between differentregions where an ecosystem has been inves-tigated; (ii) the intervals between experi-mental additions of nitrogen; and (iii) uncer-tainties in presented total atmospheric deposition values, although the latter have

been checked by local specialists on atmos-pheric N deposition. Some additional infor-mation has been given on how to interpretthis range in specific situations for anecosystem. For every group of ecosystems,the empirical N critical loads are given withan indication of exceedance and of their reliability.

The reliability of the presented N critical loadfigures is indicated as before (Bobbink et al.1996):

- reliable ##: when a number of published papers of various studies show compara-ble results;

- quite reliable #: when the results of some studies are comparable;

- expert judgement (#): when no empirical data are available for this type of ecosys-tem. The N critical load is then based upon expert judgement and knowledge of ecosystems, which are likely to be more or less comparable with this ecosystem.

5.2.1.3 Ecosystem classification

To facilitate and harmonise the mapping pro-cedure, the receptor groups of natural andsemi-natural ecosystems were classifiedand ordered according to the EUNIS habitatclassification for Europe (Davies and Moss2002, http://eunis.eea.eu.int/index.jsp). Foran introduction of EUNIS classification withrespect to empirical N critical loads (see Hallet al. 2003). In general, the ecosystems usedin the 2002 updating procedure, were classi-fied down to level 2 or 3 of the EUNIS hierarchy. The following habitats groups(with EUNIS level 1 code between brackets)were treated:

- Woodland and forests habitats (G)- Heathland, scrub and tundra habitats (F)- Grassland and tall forb habitats (E)- Mire, bog and fen habitats (D)- Inland surface water habitats (C)- Coastal habitats (B)- Marine habitats (A)





Table 5.1: Empirical critical loads for nitrogen deposition (kgN/ha/yr) to natural and semi-natural groups of ecosys-tems classified according EUNIS (except for forests). Reliability: ## reliable, # quite reliable and (#) expert judgement.



a) use towards high end of range at P limitation, and towards lower end if P is not limiting;b) use towards high end of range when sod cutting has been practised, use towards lower end

of range with low intensity management;c) use towards high end of range with high precipitation and towards low end of range with low

precipitation;d) for D2.1 (quaking fens and transition mires): use lower end of range (#) and for D2.3 (valley

mires): use higher end of range (#);e) for high latitude or N-limited systems: use lower end of range.



In general, a good agreement was foundbetween the previously used classification ofecosystem groups (Bobbink et al. 1996) andthe EUNIS habitat classification now adopted. A main limitation for the use ofmany subcategories of the EUNIS classifica-tion was, unfortunately, a lack of researchand data on N impacts for those habitats.Finally, it was at this moment not possible touse the EUNIS classification with respect tothe setting of empirical N critical loads forforest ecosystems below level 1. It was onlypossible to set values of three broad EUNISclasses (G1, G3 & G4) for forest, with, how-ever, some separation for grouping of foresttypes, such as coniferous versus deciduousand boreal versus temperate. Even withinG1, G3 and G4 there are several types, e.g.wet-swamp forest and Mediterraneanforests for which no data were available andthus left out. As before, studies based onpure plantation stands were (if possible) notaccepted in the forest section, because theN critical loads of these intensively used systems are obtained via the steady-statemass balance method (see section 5.3). Anoverview of the old and new classification ispresented in Table 5.3 to assist the shift tothe EUNIS classification.

5.2.1.4 Use of empirical critical loads

Most of the Earth’s biodiversity is present insemi-natural and natural ecosystems. It isthus crucial to control the atmospheric Nloads, in order to prevent negative effects onthese semi-natural and natural systems. Theempirical N critical loads updated in 2002(Table 5.1) should be used to revise criticalload databases. High-resolution maps of thesensitive ecosystems of high conservationvalue are needed per country to map N criti-cal loads for these systems. It is advised touse both the mass balance and empiricallyderived N critical loads for forest ecosys-tems and other ecosystems for which dataneeded for the application of steady statemodels is available. If the two approachesyield different values, the one with the lowestvalues should be used until the backgroundfor this difference has been clarified.Furthermore, it is suggested to the differentcountries, where insufficient national data

for specific national ecosystems are avail-able, to use the lower, middle or upper partof the ranges of the N critical loads for (semi-)natural ecosystem groups accordingto the general relationships between abioticfactors and critical loads for N as given inTable 5.2.

Countries are advised to identify thosereceptor ecosystems of high sensitivity with-in the mentioned EUNIS classification relat-ing to their individual interest. Effort shouldbe directed to produce fine resolution mapsof sensitive ecosystems of high conserva-tion value. At this moment the empirical Ncritical loads have been set in values of totalatmospheric N (kgN/ha/yr). More informationis needed on the relative effects of oxidisedand reduced N deposition. It was empha-sised during the last two UNECE expertmeetings that there is increasing evidence ofNHx having greater evidence than NOy.

Particularly, bryophytes and lichens in anumber of ecosystems, and several, mostlyweakly buffered, ecosystems of EUNIS classF, E, C and B are (probably) more sensitive todeposition of reduced N. It is, however, atpresent not possible to set critical loads forboth forms of N, separately.

5.2.1.5 Recommendations

Serious gaps in knowledge exist on theeffects of enhanced N deposition (NOy andNHx) on semi-natural and natural ecosys-tems, although considerably progress hasbeen made in several habitat groups from1996 to 2002. The following gaps in knowl-edge have been recognised as most impor-tant:- research/data collection is required to

establish a critical load for the following ecosystems: steppe grasslands, all Mediterranean vegetation types, wet-swamp forests, many mire & fens, several coastal habitats and high altitude sys-tems;

- more research is needed in all distin-guished EUNIS items with expert judge-ment or few research;

- impacts of N enrichment in (sensitive) freshwater and shallow marine ecosys-

tems needs further research and are sometimes overlooked;

- additional effort is needed to allocated observed N effects to the appropriate EUNIS forest subtypes (division 2 & 3);

- the EUNIS classification needs clarifica-tion/adjustment with respect to some

grasslands groups, Nordic bogs and mires and surface water habitats;

- the possible differential effects of the deposited nitrogen species (NOy or NHx ) are insufficiently known to make a differ-entiation between these N species for critical load establishment;



Table 5.2: Suggested action for using lower, middle or upper part of the set critical loads of terrestrial ecosystems(excluding wetlands), if national data are insufficient.

Table 5.3: Cross-comparison between the ecosystem classification used in the 2002 empirical N critical load setting (according to the EUNIS system) and the classification previously used (Bobbink et al. 1996) (with n.d. = notdistinguished).



- in order to refine current critical loads, long-term (> 3–5 years) N addition experi-ments with a high resolution of treatments between 5 and 50 kgN/ha/yr at low back-ground regions or in mesocosms are use-ful. This would increase the certainty of deriving critical loads when the lowest treatment level considerably exceeds the critical load;

In conclusion, it is crucial to understand thelong-term effects of increased N depositionon ecosystem processes in a representativerange of ecosystems. It is thus very impor-tant to quantify the effects of nitrogen loadsby manipulation of N inputs in long-termecosystem studies in unaffected and affect-ed areas. These data are essential to vali-date the set critical loads and to developrobust dynamic ecosystem models and/ormultiple correlative species models, whichare reliable enough to calculate critical loadsfor nitrogen deposition in (semi-) naturalecosystems and to predict (natural) recoveryrates for N-affected systems.

5.2.2 Empirical Critical loads of acidity

Empirical approaches assign an acidity critical load to soils on the basis of soil mineralogy and/or chemistry. For example,at the Critical Loads Workshop at Skokloster(Nilsson and Grennfelt 1988) soil formingmaterials were divided into five classes onthe basis of the dominant weatherable minerals. A critical load range, rather than asingle value, was assigned to each of theseclasses according to the amount of aciditythat could be neutralised by the base cationsproduced by mineral weathering (Table 5.4).Other methods of estimating base cationweathering are discussed in Chapter 5.3.2.

In addition, a number of modifying factorswere identified that would enable the criticalload value to be adjusted within the ranges(Table 5.5, after Nilsson and Grennfelt 1988).For example, some factors could make thesoil more sensitive to acidification, requiringthe critical load to be set at the lower end ofthe range; while other factors could makethe soil less sensitive, setting the criticalload at the upper end of the range.

The classification of soil materials developedat Skokloster (Table 5.4) used a relatively

Table 5.4: Mineralogical classification of soil materials and soil critical loads.

Table 5.5: Modifying factors causing an increase or decrease in critical loads.

small range of primary silicate minerals andcarbonates. A larger range of minerals hasbeen classified by Sverdrup and Warfvinge(1988) and Sverdrup et al. (1990). The follow-ing mineral classes have been identified:

Very fast weathering minerals (carbon-ates) include minerals that have the poten-tial to dissolve very rapidly, in a geologicalperspective. The group includes calcite,dolomite, magnesite and brucite.Fast weathering minerals include the silicate minerals with the fastest weatheringrate. The group comprises minerals such asanorthite and nepheline, olivine, garnet,jadeite, diopside. A soil with a major contentof these minerals would be resistant to soilacidification.Intermediate weathering minerals includeenstatite, hypersthene, augite, hornblende,glaucophane, chlorite, biotite, epidote,zoisite. Slow weathering minerals include albite,oligoclase, labradorite, illite. Soils with amajority of such minerals will be sensitive tosoil acidification.Very slow weathering minerals include K-feldspar, muscovite, mica, montmoril-lonite, vermiculite. Soils with a majority ofthese minerals will be sensitive to soil acidi-fication.Inert minerals are those that dissolve soslowly or provide so little neutralising substance that they may be considered asinert for soil acidification purposes. Thisincludes minerals such as quartz, rutile,anatase, kaolinite, gibbsite.

For each of the above mineral classes,weathering rates for soils with different mineral contents have been proposed (Table 5.6, Sverdrup et al. 1990).

The information provided in Tables 5.4 to 5.6above provide the basis on which empiricalacidity critical loads can be assigned tosoils. If mineralogical data are available forthe units of a soil map, critical loads can beassigned to each unit and a critical loadsmap produced.

An example of the development of a criticalload map at the national scale using empiri-cal approaches is given by Hornung et al. (1995). In the UK this approach hasbeen used to define acidity critical loads for non-forest ecosystems, by setting a criticalload that will protect the soil upon which thehabitat depends (Hall et al. 1998, 2003). Thecritical load is effectively the base cationweathering rate, with the leaching of acidneutralising capacity (ANC) set to zero (seesection 5.3.2), and can be used in the calcu-lations of the maximum critical loads of sulphur and nitrogen (see section 5.3.3).



Table 5.6: Weathering rates (in eq/(ha·m)/yr) for four selected mineral classes of soil material based on a soil depthof one meter – to convert to critical load values multiply by soil thickness in meters.



The purpose of a model-based approach tocalculating critical loads is to link, via math-ematical equations, a chemical criterion(critical limit) with the maximumdeposition(s) ‘below which significant harm-ful effects on specified sensitive elements ofthe environment do not occur’, i.e. for whichthe criterion is not violated. In most casesthe ‘sensitive element of the environment’will be of a biological nature (e.g., the vitalityof a tree, the species composition of aheather ecosystem) and thus the criterionshould be a biological one. However, there isa dearth of simple yet reliable models thatadequately describe the whole chain fromdeposition to biological impact. Therefore,chemical criteria are used instead, and sim-ple chemical models are used to derive crit-ical loads. This simplifies the modellingprocess somewhat, but shifts the burden tofind, or derive, appropriate (soil) chemicalcriteria (and critical limits) with proven(empirical) relationships to biological effects.The choice of the critical limit is an importantstep in deriving a critical load, and much ofthe uncertainty in critical load calculationsstems from the uncertainty in the linkbetween (soil) chemistry and biologicalimpact.

In the following we consider only steady-state models, and concentrate on the so-called Simple Mass Balance (SMB) modelas the standard model for calculating criticalloads for terrestrial ecosystems under theLRTAP Convention (Sverdrup et al. 1990,Sverdrup and De Vries 1994). The SMBmodel is a single-layer model, i.e., the soil istreated as a single homogeneous compart-ment. Furthermore, it is assumed that thesoil depth is (at least) the depth of the rooting zone, which allows us to neglect thenutrient cycle and to deal with net growthuptake only. Additional simplifying assump-tions include:

• all evapotranspiration occurs on the topof the soil profile

• percolation is constant through the soil profile and occurs only vertically

• physico-chemical constants are assumed uniform throughout the whole soil profile

• internal fluxes (such as the weatheringrates, nitrogen immobilisation etc.) are independent of soil chemical conditions(such as pH)

Since the SMB model describes steady-state conditions, it requires long-term averages for input fluxes. Short-term variations – e.g., episodic, seasonal, inter-annual, due to harvest and as a resultof short-term natural perturbations – are notconsidered, but are assumed to be includedin the calculation of the long-term mean. Inthis context ‘long-term’ is defined as about100 years, i.e. at least one rotation period forforests. Ecosystem interactions andprocesses like competition, pests, herbivoreinfluences etc. are not considered in theSMB model. Although the SMB model is formulated for undisturbed (semi-natural)ecosystems, the effects of extensive management, such as grazing and the burning of moor, could be included.

Besides the single-layer SMB model, thereexist multi-layer steady-state models for calculating critical loads. Examples are theMACAL model (De Vries 1988) and the widely-used PROFILE model (Warfvinge andSverdrup 1992), which has at its core amodel for calculating weathering rates fromtotal mineral analyses. These models will notbe discussed here, and the interested readeris referred to the literature.

In the following sections we will derive theSMB model for critical loads of nutrientnitrogen (eutrophication) and critical loads of acidifying sulphur and nitrogen.

5.3 Modelling Critical Loads for Terrestrial Ecosystems

5.3.1 Critical loads of nutrient nitrogen (eutrophication)

5.3.1.1 Model derivation

The starting point for calculating criticalloads of nutrient N with the SMB model is themass balance of total nitrogen (N) for the soilcompartment under consideration (inputs = sinks + outputs):

(5.1)

where:Ndep = total N depositionNfix = N ‘input’ by biological fixationNad = N adsorptionNi = long-term net immobilisation of N in

soil organic matterNu = net removal of N in harvested vege-

tation and animalsNde = flux of N to the atmosphere due to

denitrificationNeros = N losses through erosionNfire = N losses in smoke due to (wild or

controlled) firesNvol = N losses to the atmosphere via NH3

volatilisationNle = leaching of N below the root zone

The units used are eq/ha/yr (or molc/ha/yr inproper SI nomenclature).

The following assumptions lead to a simplifi-cation of eq. 5.1:• Nitrogen adsorption, e.g., the adsorption

of NH4 by clay minerals, can temporarily lead to an accumulation of N in the soil, however it is stored/released only when the deposition changes, and can thus be neglected in steady state considerations.

• Nitrogen fixation is negligible in most (forest) ecosystems, except for N-fixing species.

• The loss of N due to fire, erosion and volatilisation is small for most ecosys-tems in Europe, and therefore neglected

in the following discussion. Alternatively, one could replace Ni by Ni+Neros+Nfire+Nvol–Nfix in the subsequent equations.

• The leaching of ammonium (NH4) can be neglected in all forest ecosystems due to (preferential) uptake and complete nitrifi-cation within the root zone (i.e. NH4,le=0,Nle=NO3,le).

Under these simplifying assumptions eq. 5.1becomes:

(5.2)

From this equation a critical load is obtainedby defining an acceptable limit to the leach-ing of N, Nle(acc), the choice of this limitdepending on the ‘sensitive element of theenvironment’ to be protected. If an accept-able leaching is inserted into eq. 5.2, thedeposition of N becomes the critical load ofnutrient nitrogen, CLnut(N):

(5.3)

In deriving the critical load of nutrient N aseq. 5.3, it is assumed that the sources andsinks do not depend on the deposition of N.This is unlikely to be the case and thus allquantities should be taken ‘at critical load’.However, to compute, e.g., ‘denitrification atcritical load’ one needs to know the criticalload, the very quantity one wants to com-pute. The only clean way to avoid this circu-lar reasoning is to establish a functional rela-tionship between deposition and the sink ofN, insert this function into eq. 5.2 and solvefor the deposition (to obtain the critical load).This has been done for denitrification: In thesimplest case denitrification is linearly relat-ed to the net input of N (De Vries et al. 1993,1994):

(5.4)





where fde (0£ f de<1) is the so-called denitrifi-cation fraction, a site-specific quantity. Thisformulation implicitly assumes that immobil-isation and uptake are faster processes thandenitrification. Inserting this expression forNde into eq. 5.2 and solving for the depositionleads to the following expression for the critical load of nutrient N:

(5.5)

An alternative, non-linear, equation for thedeposition-dependence of denitrificationhas been proposed by Sverdrup and Ineson(1993) based on the Michaelis-Menten reaction mechanism and includes a depend-ence on soil moisture, pH and temperature.Also in this case CLnut(N) can be calculatedexplicitly, and for details the reader isreferred to Posch et al. (1993).

More generally, it would be desirable to havedeposition-dependent equations (models)for all N fluxes in the critical load equation.However, these either do not exist or are soinvolved that no (simple) explicit expressionfor CLnut(N) can be found. Although this doesnot matter in principle, it would reduce theappeal and widespread use of the criticalload concept. Therefore, when calculatingcritical loads from eq. 5.3 or eq. 5.5, the Nfluxes should be estimated as long-termaverages derived from conditions not influ-enced by elevated anthropogenic N inputs.

5.3.1.2 The acceptable leaching of nitrogen

The value set for the acceptable N leachingdepends on the ‘harmful effects’ that shouldbe avoided. In general, it is not the N leach-ing flux itself that is ‘harmful’, but the con-centration of N in the leaching flux. Theacceptable N leaching (in eq/ha/yr) is calcu-lated as:

(5.6)

where [N]acc is the acceptable N concentra-tion (eq/m3) and Q is the precipitation surplus(in m3/ha/yr). Some values for acceptable Nconcentrations are shown in Table 5.7.

Although literature data indicate that nutrientimbalances may occur when N leachingincreases above natural background values(Van Dam 1990), no direct relationshipbetween N leaching and vegetation changeshas been substantiated. In general, the lowleaching values from the above table lead tocritical loads that are lower than empiricaldata on vegetation changes (e.g. Bobbink etal. 1998). It is the increase in N availabilitythrough enhanced N cycling that triggerschanges (Berendse et al. 1987).

An acceptable N leaching could also bederived with the objective to avoid N pollu-tion of groundwater using, e.g., the EC targetor limit value (25 and 50 mgN/L, resp.) asacceptable (but high!) concentration.

Table 5.7: Acceptable N leacheate concentrations to avoid nutrient imbalances or vegetation changes (quoted fromPosch et al. 1993).

5.3.1.3 Sources and derivation of input data

The obvious sources of input data for calcu-lating critical loads are measurements at thesite under consideration. However, in manycases these will not be available. A discus-sion on N sources and sinks can be found inHornung et al. (1995) and UNECE (1995).Some data sources and default values andprocedures to derive them are summarisedbelow.

Nitrogen immobilisation:Ni refers to the long-term net immobilisation(accumulation) of N in the root zone, i.e., thecontinuous build-up of stable C-N-com-pounds in (forest) soils. In other words, thisimmobilisation of N should not lead to signif-icant changes in the prevailing C/N ratio.This has to be distinguished from the highamounts of N accumulated in the soils over many years (decades) due to the increaseddeposition of N, leading to a decrease in theC/N ratio in the topsoil.

Using data from Swedish forest soil plots,Rosén et al. (1992) estimated the annual Nimmobilisation since the last glaciation at0.2–0.5 kgN/ha/yr (14.286–35714 eq/ha/yr).Considering that the immobilisation of N isprobably higher in warmer climates, valuesof up to 1 kgN/ha/yr (71.428 eq/ha/yr) could beused for Ni, without causing unsustainableaccumulation of N in the soil. It should bepointed out, however, that even higher values (closer to present-day immobilisationrates) have been used in critical load cal-culations. Although studies on the capacityof forests to absorb nitrogen have been carried out (see, e.g., Sogn et al. 1999), thereis no consensus yet on long-term sus-tainable immobilisation rates.

Nitrogen uptake:The uptake flux Nu equals the long-termaverage removal of N from the ecosystem.For unmanaged ecosystems (e.g., nationalparks) the long-term (steady-state) netuptake is basically zero whereas for managed forests it is the long-term netgrowth uptake. The harvesting practice is of

crucial importance, i.e., whether stems only,stems plus (parts of) branches or stems plusbranches plus leaves/needles (whole-treeharvesting) are removed. The uptake of N isthen calculated as:

(5.7)

The amount of N in the harvested biomass(stems and branches) can be calculated asfollowing:

(5.8)

where kgr is the average annual growth rate(m3/ha/yr), rst is the density of stem wood(kg/m3), ctN is the N content in stems (subscript st) and branches (subscript br)(eq/kg) and fbr,st is the branch-to-stem ratio(kg/kg). The contribution of branches shouldbe neglected in case of stem removal.

Values for the density of stem wood of mosttrees are in the range of 400–500 kg/m3 forconifers and 550–700 kg/m3 for deciduoustrees. The branch-to-stem ratio is about 0.15 kg/kg for conifers and 0.20 kg/kg fordeciduous trees (Kimmins et al. 1985, De Vries et al. 1990). According to Swedishdata (Rosén 1990; see also Reinds et al.2001) the contents of N in stems are 1 g/kgfor conifers and 1.5 g/kg in deciduous trees,whereas in branches of all tree species the Ncontent is 4 g/kg in the south and 2 g/kg inthe north. In a recent report Jacobsen et al.(2002) have summarised the results of alarge number of studies on that subject, andTable 5.8 shows the average element contents in 4 major tree species, both forstems and branches. For N, the values haveto be multiplied by 1/14=0.07143 to obtainthe N contents in eq/kg.

Growth rates used should be long-term aver-age values, typical for the site. It has to benoted that recent growth rates are higherdue to increased N input. Therefore it is





recommended to use older investigations(yield tables), preferably from before1960–70. An example of how to use nationalinventory information to compute forestgrowth (and critical loads) in Germany canbe found in Nagel and Gregor (1999).

Net uptake of N in non-forest natural andsemi-natural ecosystems is insignificant,unless they are used for extensive grazing.For example, in the United Kingdom netremoval of N in sheep (mutton/wool) due toextensive grazing is between 0.5 and 2.0kgN/ha/yr, depending on site fertility andgrazing density.

Denitrification:Dutch and Ineson (1990) reviewed data onrates of denitrification. Typical values of Ndefor boreal and temperate ecosystems are inthe range of 0.1–3.0 kgN/ha/yr (=7.14–214.3eq/ha/yr), where the higher values apply towet(ter) soils; rates for well drained soils aregenerally below 0.5 kgN/ha/yr.

With respect to deposition-dependent deni-trification, values for the denitrification fraction fde have been given by De Vries et al.(1993) based on data from Breeuwsma et al.(1991) and Steenvorden (1984): fde=0.8 for

peat soils, 0.7 for clay soils, 0.5 for sandysoils with gleyic features and fde=0–0.1 forsandy soils without gleyic features. Reindset al. (2001) related the denitrification fraction to the drainage status of the soilaccording to Table 5.9:

Precipitation surplus:The precipitation surplus Q is the amount ofwater percolating from the root zone. It isconveniently calculated as the differencebetween precipitation and actual evapotran-spiration and it should be the long-term cli-matic mean annual value. In many casesevapotranspiration will have to be calculatedby a model using basic meteorological inputdata (precipitation, temperature, radiationetc.). For the basics of modelling evapotran-spiration see Monteith and Unsworth (1990)and for an extensive collection of modelssee Burman and Pochop (1994). Historicaltime series of meteorological data can befound, e.g., on the website of the ClimateChange Research Unit of the University ofEast Anglia (http://www.cru.uea.ac.uk/cru/data/hrg.htm).

Table 5.9: Denitrification fraction fde as a function of the soil drainage (Reinds et al. 2001).

a)Note that for Ca data points from calcareous sites are included in the statistics.

Table 5.8: Mean (and standard deviation) of the element contents in stems and branches (both incl. bark) of fourtree species (Jacobsen et al. 2002; the number of data points ranges from 6 to 32).

Contents (g/kg) in stems (incl. bark) Contents (g/kg) in branches (incl. bark) Tree

species N Caa)

Mg K N Ca Mg K

2.10 2.47 0.18 1.05 6.19 4.41 0.44 2.00 Oak

quercus spp (0.46) (1.42) (0.07) (0.51) (1.02) (0.65) (0.14) (0.47)

Beech 1.54 1.80 0.26 1.04 4.27 4.02 0.36 1.50

fagus sylv. (0.25) (1.12) (0.09) (0.13) (1.36) (1.91) (0.13) (0.44)

Spruce 1.22 1.41 0.18 0.77 5.24 3.33 0.53 2.39

picea abies (0.49) (0.40) (0.06) (0.43) (1.66) (1.06) (0.27) (1.35)

Pine 1.09 1.08 0.24 0.65 3.61 2.07 0.43 1.67

pinus sylv. (0.30) (0.30) (0.09) (0.28) (1.28) (0.65) (0.11) (0.68)

5.3.2 Critical loads of acidity

5.3.2.1 Model derivation: the Simple Mass Balance (SMB) model

The starting point for deriving critical loadsof acidifying S and N for soils is the chargebalance of the ions in the soil leaching flux(De Vries 1991):

(5.9)

where the subscript le stands for leaching, Alstands for the sum of all positively chargedaluminium species, BC is the sum of basecations (BC=Ca+Mg+K+Na) and RCOO is thesum of organic anions. A leaching term isgiven by Xle=Q·[X], where [X] is the soil solu-tion concentration of ion X and Q is the precipitation surplus. All fluxes areexpressed in equivalents (moles of charge)per unit area and time (eq/ha/yr). The concen-trations of OH and CO3 are assumed zero,which is a reasonable assumption even forcalcareous soils. The leaching of AcidNeutralising Capacity (ANC) is defined as:

(5.10)

Combination with eq. 5.9 yields:

(5.11)

This shows the alternative definition of ANCas ‘sum of (base) cations minus strong acidanions’. For more detailed discussions onthe processes and concepts of (soil) chem-istry encountered in the context of acidifica-tion see, e.g., the books by Reuss andJohnson (1986) or Ulrich and Sumner (1991).

Chloride is assumed to be a tracer, i.e., thereare no sources or sinks of Cl within the soilcompartment, and chloride leaching is there-fore equal to the Cl deposition (subscriptdep):

(5.12)

In a steady-state situation the leaching ofbase cations has to be balanced by the netinput of base cations. Consequently the fol-lowing equation holds:

(5.13)

where the subscripts w and u stand forweathering and net growth uptake, i.e. thenet uptake by vegetation that is needed forlong-term average growth; Bc=Ca+Mg+K,reflecting the fact that Na is not taken up byvegetation. Base cation input by litterfall andBc removal by maintenance uptake (neededto re-supply base cations in leaves) is notconsidered here, assuming that both fluxesare equal (in a steady-state situation). Alsothe finite pool of base cations at theexchange sites (cation exchange capacity,CEC) is not considered. Although cationexchange might buffer incoming acidity fordecades, its influence is only a temporaryphenomenon, which cannot be taken intoaccount when considering long-termsteady-state conditions.

The leaching of sulphate and nitrate can belinked to the deposition of these compoundsby means of mass balances for S and N. ForS this reads (De Vries 1991):

(5.14)

where the subscripts ad, i, re and pr refer toadsorption, immobilisation, reduction andprecipitation, respectively. An overview ofsulphur cycling in forests by Johnson (1984)suggests that uptake, immobilisation andreduction of S have generally insignificant.Adsorption (and in some cases precipitationwith Al complexes) can temporarily lead to astrong accumulation of sulphate (Johnson et al. 1979, 1982). However, sulphate is onlystored or released at the adsorption complex when the input (deposition)changes, since the adsorbed S is assumed in





equilibrium with the soil solution S. Onlydynamic models can describe the time pattern of ad- and desorption of sulphate,but under steady-state conditions S ad- anddesorption and precipitation/mobilisationare not considered. Since sulphur is com-pletely oxidised in the soil profile, SO4,leequals Sle, and consequently:

(5.15)

For nitrogen, the mass balance in soil is (seeSection 5.3.1):

(5.16)

where the subscripts fix refers to fixation ofN, de to denitrification, and eros, fire and volto the loss of N due to erosion, forest firesand volatilisation, respectively. Ni is the long-term immobilisation of N in the root zone,and Nu the net growth uptake (see above).Furthermore, the leaching of NH4 can beneglected in almost all forest ecosystemsdue to (preferential) uptake and completenitrification within the root zone, i.e. NH4,le=0.Under these various assumptions eq. 5.16simplifies to:

(5.17)

Inserting eqs. 5.12, 5.13, 5.15 and 5.17 intoeq. 5.11 leads to the following simplifiedcharge balance for the soil compartment:

(5.18)

Strictly speaking, we should replace NO3,le inthe charge balance not by the right-handside of eq. 5.17, but bymax{Ndep–Ni–Nu–Nde,0}, since leaching cannot

become negative; and the same holds truefor base cations. However, this would lead tounwieldy critical load expressions; thereforewe go ahead with eq. 5.18, keeping this constraint in mind.

Since the aim of the LRTAP Convention is toreduce anthropogenic emissions of S and N,sea-salt derived sulphate should not be considered in the balance. To retain chargebalance, this is achieved by applying a sea-salt correction to sulphate, chloride andbase cations, using either Cl or Na as a tracer, whichever can be (safer) assumed tooriginale from sea-salts only. Denoting sea-salt corrected depositions with an asterisk, one has either Cl*

dep=0 or Na*dep=0

(and BC*dep=Bc*

dep), respectively. For procedures to compute sea-salt correcteddepositions, see Chapter 2.

For given values for the sources and sinks ofS, N and Bc, eq. 5.18 allows the calculation ofthe leaching of ANC, and thus assessment ofthe acidification status of the soil.Conversely, critical loads of S, CL(S), and N,CL(N), can be computed by defining a criticalANC leaching, ANCle,crit:

(5.19)

A so-called critical load of potential acidityhas earlier been defined (see Sverdrup et al.1990) as:

(5.20)

with Acpot = Sdep+Ndep–BC*dep+Cl*

dep. The term‘potential’ is used since NH3 is treated as(potential) acid due to the assumed complete nitrification. CL(Acpot) has beendefined to have no deposition terms in itsdefinition, since Bc and Cl deposition are notreally an ecosystem property and can (andoften do) change over time. However, sincethese depositions are partly of non-anthro-

pogenic origin (e.g., Saharan dust) and sincethey are not subject to emission reductionnegotiations, they are kept in the critical loaddefinition for convenience.

A further distinction has been made earlier(see, e.g., Sverdrup and De Vries 1994)between ‘land use acidity’ Bcu–Ni–Nu–Nde and‘soil acidity’ which is used to define a so-called critical load of (actual) acidity as:

(5.21)

The reason for making this distinction was toexclude all variables that may change in thelong term such as uptake of Bc and N, whichare influenced by forest management, and Nimmobilisation and denitrification, whichmay change due to changes in the hydrolog-ical regime. There are two problems with thisreasoning: (a) the remaining terms in eq. 5.21are also liable to change (e.g. ANC leachingdepends on precipitation surplus, see

below), and (b) uptake and other N process-es are a defining part of the ecosystem (veg-etation) itself. In other words, CL(A) may be acritical load of soil acidity, but it is rarely thesoil as such that is the ‘sensitive element’ tobe protected, but the vegetation growing onthat soil! Nevertheless, quantities such asCL(A) are computed and reported, and theycan have a role as useful short-hand notation for the variables involved.

Note that eq. 5.19 does not give a uniquecritical load for S or N. However, nitrogensinks cannot compensate incoming sulphuracidity, and therefore the maximum criticalload for sulphur is given by:

(5.22)

as long as N deposition is lower than all theN sinks, termed the minimum critical load ofN, i.e. as long as



Figure 5.1: Critical load function (CLF) of sulphur and acidifying nitrogen, defined by the three quantities CLmax(S),CLmin(N) and CLmax(N). (a) with constant denitrification Nde, and thus a 45o slope of the CLF; (b) with deposition-dependent denitrification, resulting in a smaller CLmin(N) and a flatter slope, depending on fde. The grey area belowthe CLF denotes deposition pairs resulting in an ANC leaching greater than ANCle,crit (non-exceedance of criticalloads; see Chapter 7).



(5.23)

Finally, the maximum critical load of nitrogen(in the case of zero S deposition) is given by:

(5.24)

The three quantities CLmax(S), CLmin(N) andCLmax(N) define the critical load function(CLF; depicted in Figure 5.1a). Every deposi-tion pair (Ndep,Sdep) lying on the CLF are critical loads of acidifying S and N.

Deriving critical loads as above assumesthat the sources and sinks of N do notdepend on the N deposition. This is unlikelyto be true; and as in Section 5.3.1 we con-sider also the case of denitrification beinglinearly related to the net input of N.Substituting eq. 5.4 for Nde into the equationsabove results in the following expressionsfor CLmin(N) and CLmax(N):

(5.25)

and

(5.26)

where fde (0£ f de<1) is the denitrification fraction; CLmax(S) remains the same (eq.5.22). An example of a critical load functionwith fde>0 is shown in Figure 5.1b.

5.3.2.2 Chemical criteria and thecritical leaching of Acid Neutralising Capacity

The leaching of Acid Neutralising Capacity(ANC) is defined in eq. 5.10. In the simplestcase bicarbonate (HCO3) and organic anions(RCOO) are neglected since in general theydo not contribute significantly at low pH values. In this case the ANC leaching is givenby:

(5.27)

where Q is the precipitation surplus inm3/ha/yr (see Section 5.3.1.3 for data).

It is within the calculation of ANCle that thecritical chemical criterion for effects on thereceptor is set. Selecting the most appro-priate method of calculating ANCle is impor-tant, since the different methods may resultin very different critical loads. If, for the sameecosystem, critical loads are calculatedusing different criteria, the final critical loadis the minimum of all those calculated. Themain decision in setting the criterion willdepend on whether the receptor consideredis more sensitive to unfavourable pH condi-tions or to the toxic effects of aluminium.ANCle can then be calculated by either set-ting a hydrogen ion criterion (i.e., a criticalsoil solution pH) and calculating the criticalaluminium concentration, or vice versa.

The relationship between [H] and [Al] isdescribed by an (apparent) gibbsite equilib-rium:

(5.28)

where Kgibb is the gibbsite equilibrium constant (see below). Eq. 5.28 is used to calculate the (critical) Al concentration froma given proton concentration, or vice versa.

Different critical chemical criteria are listedbelow together with the equations for calcu-

lating ANCle,crit. In this context the readercould also consult the minutes of an ExpertWorkshop on ‘Chemical Criteria and CriticalLimits’ (UNECE 2001, Hall et al. 2001).

Aluminium criteria:Aluminium criteria are generally consideredmost appropriate for mineral soils with a loworganic matter content. Three commonlyused criteria are listed below.

(a) Critical aluminium concentration:Critical limits for Al have been suggested forforest soils, e.g., [Al]crit=0.2 eq/m3. These areespecially useful for drinking water (groundwater) protection, e.g., the EC drinking waterstandard for [Al] of maximally 0.2 mg/L(about 0.02 eq/m3). ANCle,crit can then be calculated as:

(5.29)

(b) Critical base cation to aluminium ratio:Most widely used for soils is the connectionbetween soil chemical status and plantresponse (damage to fine root) via a criticalmolar ratio of the concentrations of basecations (Bc=Ca+Mg+K) and Al in soil solution, denoted as (Bc/Al)crit. Values for alarge variety of plant species can be found inSverdrup and Warfvinge (1993). The mostcommonly used value is (Bc/Al)crit=1, thevalue for coniferous forests.

The critical Al leaching is calculated from theleaching of Bc (compare eq. 5.13):

(5.30)

The factor 1.5 arises from the conversion ofmols to equivalents (assuming K as divalent).Using eqs. 5.27 and 5.28, this yields for thecritical ANC leaching:

(5.31)

Note that the expression Bcle=Bcdep+Bcw–Bcuhas to be non-negative. In fact, it has beensuggested that it should be above a minimum leaching or, more precisely, there isa minimum concentration of base cations inthe leacheate, below which they cannot betaken up by vegetation, i.e., Bcle is set equalto max{0,Bcdep+Bcw–Bcu–Q·[Bc]min}, with[Bc]min in the order of 0.01eq/m3.

Alternatively, if considered more appropriate,a critical molar ratio of calcium to aluminiumin soil solution can be used, by replacing allthe Bc-terms in eq. 5.31 with Ca-terms.

(c) Critical aluminium mobilisation rate:Critical ANC leaching can also be calculatedusing a criterion to prevent the depletion ofsecondary Al phases and complexes whichmay cause structural changes in soils and afurther pH decline. Aluminium depletionoccurs when the acid deposition leads to anAl leaching in excess of the Al produced bythe weathering of primary minerals. Thus thecritical leaching of Al is given by:

(5.32)

where Alw is the weathering of Al from primary minerals (eq/ha/yr). The weatheringof Al can be related to the Bc weathering via:

(5.33)

where p is the stoichiometric ratio of Al to BCweathering in primary minerals (eq/eq), witha default value of p=2 for typical mineralogyof Northern European soils (range: 1.5–3.0).The critical leaching of ANC becomes then:





(5.34)

Hydrogen ion criteria:A proton criterion is generally recommendedfor soils with a high organic matter content.Two such criteria are listed below.

(a) Critical pH:A critical pH limit is set at a pH below whichthe receptor is adversely affected. Criticallimits have been suggested for forest soils,for example, pHcrit=4.0 (corresponding to[H]crit=0.1 eq/m3). ANCle,crit can then be calcu-lated as:

(5.35)

(b) Critical base cation to proton ratio:For organic soils which do not contain Al-(hydr)oxides (such as peat lands), it issuggested to use a critical molar basecation to proton ratio (Bc/H)crit. The criticalANC leaching is then given by (no Al leach-ing!):

(5.36)

where the factor 0.5 comes from convertingmols to equivalents. For organic soils theweathering in eq. 5.36 will probably be negligible (Bcw=0). Values suggested for(Bc/H)crit are expressed as multiples of(Bc/Al)crit, these multiples ranging from 0.3for deciduous trees and ground vegetationto 1 for spruce and pine (Sverdrup and Warfvinge 1993).

Critical base saturationBase saturation, i.e., the fraction of basecations at the cation exchange complex, isan indicator of the acidity status of a soil,and one may want to keep this pool above acertain level to avoid nutrient deficiencies.Thus a critical (acceptable, minimum) basesaturation could be chosen as a criterion forcalculating critical loads of acidity (see Hallet al. 2001, UNECE 2001).

To relate base saturation to ANC requires thedescription of the exchange of cationsbetween the exchange complex and the soilsolution. Two descriptions are the mostcommonly used in dynamic soil models: theGapon and the Gaines-Thomas exchangemodel. For a comparison between differentexchange models and the implications forthe relationship between base saturationand soil solution concentrations see Reuss(1983).As an example, we consider the descriptionof the exchange between H, Al andBc=Ca+Mg+K as implemented in the VerySimple Dynamic (VSD) as well as the SAFEmodel (see Posch et al. 2003a or Chapter 6on dynamic modelling). For both models thecritical concentration [H]crit can be found asa solution of an equation of the type:

(5.37)

where the coefficients A, B and the exponentp are given in Table 5.10.

Note: The generalised relationship [Al]=KAlox [H]a has been used (see below).

Table 5.10: Coefficients in eq. 5.37 for the Gapon and Gaines-Thomas exchange model.

In general, eq. 5.37 is non-linear and willhave to be solved numerically. Only for theGapon model and the gibbsite equilibrium(a=3, KAlox=Kgibb) it becomes a linear equa-tion with the solution:

(5.38)

where kHBc and kAlBc are the two (site-spe-cific) selectivity coefficients describingcation exchange and [Bc]=Bcle/Q as above.[Al]crit is then computed from the gibbsiteequilibrium (eq. 5.28) and from that the critical ANC leaching can be obtained via eq. 5.29. Values of selectivity coefficients fora range of (Dutch) soil types and combina-tions of exchangeable ions are given by De Vries and Posch (2003).

In Figure 5.2 the critical ANC leaching isshown for a range of constants KGap. Thisrange encompasses a wide range of valuesfor the exchange constants. The figureshows that ANC leaching is very sensitive tolow values of the critical base saturation.

Base saturation is also used as criterion inthe New England Governors/EasternCanadian Premiers ‘Acid Rain Action Plan’for calculating sustainable S and N deposi-tions to upland forests with the SMB model(NEG/ECP 2001).

5.3.2.3 Sources and derivation of input data

The obvious sources of input data for calcu-lating acidity critical loads are measure-ments at the site under consideration.However, in many cases these will not beavailable. For data on the different N quanti-ties see Section 5.3.1. Some data sourcesand default values for the other variables,and procedures to derive them, are sum-marised below.

Gibbsite equilibrium constant (Kgibb):The equilibrium constant relating the Alconcentration to pH (eq. 5.28) depends onthe soil. Table 5.11 presents ranges of Kgibb(and pKgibb=–log10(Kgibb in (mol/L)-2) as a function of the soil organic matter content. Awidely used default value is Kgibb=108 (mol/L)-2=300 m6/eq2.



Figure 5.2: Critical ANC leaching (as defined by eq. 5.27, for Q=1 m/yr) as a function of the critical base saturation,EBc,crit, for [Bc]=0.02eq/m3, Kgibb=108 and KGap=0.005 (leftmost curve), 0.01, 0.03 and 0.05 (rightmost curve).(To obtainANCle,crit for arbitrary Q, multiply the values on the vertical axis by Q in m/yr; see also Figure 5.4 below.)



If sufficient empirical data are available toderive the relationship between [H] and [Al],these should be used in preference to thegibbsite equilibrium (see Sec. 5.3.2.4).

Base cation and chloride deposition:The base cation and chloride depositionsentering the critical load calculations shouldbe the deposition after all feasible abate-ment measures have been taken (ideally thenon-anthropogenic deposition), and theyshould be sea-salt corrected. Observationson a European scale are available from theEMEP Chemical Co-ordinating Centre(www.emep.int) or from national sources.See Chapter 2 for more details.

Base cation weathering:Weathering here refers to the release of basecations from minerals in the soil matrix dueto chemical dissolution, and the neutralisa-tion and production of alkalinity connectedto this process. This has to be distinguishedfrom the denudation of base cations from ionexchange complexes (cation exchange) andthe degradation of soil organic matter. Manymethods for determining weathering rateshave been suggested, and here we list thosewith the highest potential for regional appli-cations (in order of increasing complexity).

(a) The Skokloster assignment:This is a (semi-)empirical method devised atthe Critical Loads Workshop at Skokloster(Sweden) (Table 1, p.40 in Nilsson andGrennfelt 1988). Details can be found in thesection on empirical acidity critical loads(Section 5.2.2).

(b) The soil type – texture approximation:Since mineralogy controls weathering rates,weathering rate classes were assigned toEuropean (forest) soils by De Vries et al.(1993), based on texture class and parentmaterial class. Texture classes are defined inTable 5.12 as a function of their clay andsand content:

Using the FAO soil classification (FAO 1981),the parent material class has been definedfor each soil type in Table 5.13 (updated fromDe Vries et al. 1993).

From texture and parent material class theweathering rate class is obtained from Table5.14 (modified from De Vries et al. 1993).

Table 5.11: Ranges for Kgibb as a function of soil organic matter content.

Table 5.12: Soil texture classes as a function of their clay and sand content (Eurosoil 1999).

The actual weathering rate (in eq/ha/yr) for anon-calcareous soil of depth z (in m) is thencomputed as:

(5.39)

where WRc is the weathering rate class(Table 5.14), T (oC) is the average annual (soil)temperature and A=3600 K (Sverdrup 1990).For calcareous soil, for which critical loadsare not really of interest, one could set, e.g.,WRc=20 in eq. 5.39.

The above procedure provides weatheringrates for BC=Ca+Mg+K+Na. However, forcomputing the critical ANC leaching accord-ing to eq. 5.31, the weathering rate forBc=Ca+Mg+K is needed. Bcw can be approx-imated by multiplying Bcw with a factorbetween 0.70 for poor sandy soils and 0.85for rich (sandy) soils. Van der Salm et al.(1998) (for texture classes 2–5, see Table5.12) and De Vries (1994) (for texture class 1)provide regression equations for weatheringrates of Ca, Mg, K and Na as a function of the

sand (and silt) content of the soil, which canbe used to split Bcw into individual weather-ing rates.

(c) The total base cation content correlation:Using the ‘zirconium method’, Olsson et al.(1993) derived from 11 Swedish sites a correlation between the historical averageweathering rates of base cations and thetotal content of the respective element in theundisturbed bottom soil, with an additionaltemperature correction. For Ca, Mg and K theequations are (Olsson et al. 1993, convertedto eq/ha/yr):

(5.40)

where (X)tot is the total content of element X(in dry weight %) in the coarse fraction(<2mm) of the undisturbed C-horizon soiland ETS is the annual sum of daily tempera-tures above a threshold of +5oC. Care has tobe taken when applying these formulae,since they are based on Nordic geologicalhistory, they do not predict the weatherable



Table 5.13: Parent material classes for common FAO soil types (Posch et al. 2003b).

Acidic: Sand(stone), gravel, granite, quartzine, gneiss (schist, shale, greywacke, glacial till)Intermediate: Gronodiorite, loess, fluvial and marine sediments (schist, shale, greywacke, glacial till)Basic: Gabbro, basalt, dolomite, volcanic deposits.

Table 5.14: Weathering rate classes as a function of texture and parent material classes (Posch et al. 2003b).



soil depth, which was found to vary between20 and 200 cm in the field data, and theydon’t cover many soil types (mostly pod-zols).Using the part of the Swedish data (7-8 sitesdepending on the element, covering aweatherable depth of 20–100 cm), thismethod was adapted in Finland for estimat-ing weathering rates on a national scale(Johansson and Tarvainen 1997, Joki-Heiskala et al. 2003).

(d) The calculation of weathering rates withthe PROFILE model:Weathering rates can be computed with themulti-layer steady-state model PROFILE(Warfvinge and Sverdrup 1992 and 1995).Basic input data are the mineralogy of thesite or a total element analysis, from whichthe mineralogy is derived by a normativeprocedure. Generic weathering rates of eachmineral are modified by the concentration ofprotons, base cations, aluminium and organ-ic anions as well as the partial pressure ofCO2 and temperature. The total weatheringrate is proportional to soil depth and thewetted surface area of all minerals present.For the theoretical foundations of the weath-ering rate model see Sverdrup (1990). Forfurther information on the PROFILE modelsee www2.chemeng.lth.se.

(e) Other methods:Weathering rates can also be estimated frombudget studies of small catchments (see,e.g., Paces 1983). Be aware, however, thatbudget studies can easily overestimateweathering rates where there is significantcation release due to weathering of thebedrock. Other methods are listed anddescribed in Sverdrup et al. (1990).

Base cation uptake:The uptake flux of base cations, Bcu, enter-ing the critical load calculations is the long-term average removal of base cationsfrom the ecosystem. The uptake fluxesshould be calculated for the individual basecations (Ca, Mg and K) separately. The considerations and calculations are exactlythe same as for the uptake of N (see Section5.3.1). Average contents of Ca, Mg and K in

stems and branches can be found in Table5.8 (see also Jacobsen et al. 2002). Valueshave to be multiplied by 2/40.08, 2/24.31 and1/39.10 for Ca, Mg and K, respectively, toobtain contents in eq/kg.

The (long-term) net uptake of base cations islimited by their availability through deposi-tion and weathering (neglecting the deple-tion of exchangeable base cations).Furthermore, base cations will not be takenup below a certain concentration in soil solu-tion, or due to other limiting factors, such atemperature. Thus the values entering criti-cal load calculations should be constrainedby:

(5.41)

This is preferable to constraining the sumBcu=Cau+Mgu+Ku (see eq. 5.31). Suggestedvalues are 5 meq/m3 for [Ca]min and [Mg]min,and zero for [K]min (Warfvinge and Sverdrup1992). It should also be taken into accountthat vegetation takes up nutrients in fairlyconstant (vegetation-specific) ratios. Thus,when adjusting the uptake value for one ele-ment, the values for the other elements(including N) should be adjusted proportion-ally.

5.3.2.4 Possible extensions to the SMB model

In the following three suggestions are madefor generalising the SMB model, with theidea of improving the critical load calcula-tions but also with the aim to enhance thecompatibility with dynamic models. All threesuggestions are ‘backwards-compatible’,i.e. by setting key parameters to zero theoriginal SMB model is obtained. For an earli-er discussion of these extensions see alsoPosch (2000).

(a) Generalisation of the Al-H relationship:In the SMB model the relationship betweenAl concentration and pH is described as

gibbsite equilibrium (see eq. 5.21). However,Al concentrations, especially in the topsoil,can be influenced by the complexation of Alwith organic matter (Cronan et al. 1986,Mulder and Stein 1994). Therefore, the gibb-site equilibrium in the SMB model could begeneralised by:

(5.42)

with equilibrium constant KAlox and exponenta. Obviously, the gibbsite equilibrium is aspecial case of eq. 5.42 (setting a=3 andKAlox=Kgibb). The exponent a and KAloxdepend on the soil type and especially onthe soil horizon. As an example, in Table 5.15values for KAlox and a are presented for different soil groups and soil depths derivedfrom several hundred Dutch forest soil solution samples (see Van der Salm and DeVries 2001).

The data in Table 5.15 show that a standardgibbsite equilibrium constant and a=3 is reasonable for (Dutch) sandy soils. Very different values, however, are obtained forpeat soils and, to a lesser extent, also forloess and clay soils (especially for shallowparts of the soil, where the organic mattercontent is highest). Data from intensive for-est monitoring plots show that there is astrong correlation between a and log10KAlox(De Vries et al. 2003, p.118), which empha-sises that these two parameters cannot bechosen independently.

Figure 5.3 shows the relationship between[H] and [Al] as well as its logarithmic formfor different values of KAlox and a. DefiningpX=–log10[X], with [X] given in mol/L, onehas pH=3–log10([H]), if [H] is expressed ineq/m3; and for [Al] in eq/m3 the relationship ispAl=3–log10([Al]/3).

Note that, when using eq. 5.42 instead of eq. 5.28, the formulae for ANCle,crit have to beadapted as well (mostly replacing the exponent 3 by a and 1/3 by 1/a).

(b) Including bicarbonate leaching:The charge balance (eq. 5.9) and the definition of ANC leaching in eq. 5.10 alsoincludes the leaching of bicarbonate anions(HCO3,le=Q·[HCO3]). The concentration ofbicarbonates is a function of the pH:

(5.43)

where K1 is the first dissociation constant,KH is Henry’s constant and pCO2 is the partialpressure of CO2 in the soil solution (in atm).The two constants are weakly temperature-dependent, and the value for their product at8oC is K1·KH=10-1.7=0.02eq²/m6/atm. For systems open to the atmosphere, pCO2 isabout 370 ppm or 3.7·10–4 atm (in the year2000). However, in soils pCO2 is generallyhigher (ranging from 10–2 to 10–1 atm, Boltand Bruggenwert 1976), due to respiration



Table 5.15: Estimated values of KAlox and the exponent a based on regression between pAl and pH in the soil solution of Dutch forests (N = number of samples).



and oxidation of below-ground organic mat-ter. Respiratory production of CO2 is highlytemperature dependant (e.g. Witkamp 1966);based on soil temperature and mean grow-ing season soil pCO2, Gunn and Trudgill(1982) derived the following relationship:

(5.44)

where T is the (soil) temperature (oC). Brooket al. (1983) present a similar regressionequation based on data for 19 regions of theworld. In the absence of data or such relationships, the following default ranges

have been suggested (Bouten et al., 1987):5–10 times atmospheric pressure in theorganic layer, 5–15 times atmospheric pressure in the E-layer, 15–20 times atmos-pheric pressure in the B-layer and 15–30times atmospheric pressure in the upper C-layer.

For pCO2=0.0055 atm (about 15 times thepartial CO2 pressure in air) and Q=0.3 m/yr,eq. 5.43 yields a bicarbonate leaching ofalmost 100 eq/ha/yr at pH=5.5, not always anegligible quantity. Therefore, it would makesense to include the bicarbonate leachinginto the SMB model. Not only would thismake critical loads more compatible with

Figure 5.3: Relationships between H and Al concentration in eq/m3 (left) and in their logarithmic forms (right) forKAlox= 101, a=2 and KAlox=104.5, a=1.3 (solid lines) as well as three gibbsite equilibria (a=3) with Kgibb=107, 108 and 109

(dashed lines). Note: [H]=0.1 eq/m3 corresponds to pH=4.

steady-state solutions of dynamic models,but it is also the only way to allow the ANCleaching to obtain positive values! Eq. 5.27would than read:

(5.45)

All chemical criteria could be used, sincebicarbonate leaching could always be calcu-lated from Hle via eq. 5.43. We illustrate theinfluence of bicarbonates on the ANC leach-ing by re-drawing Figure 5.2, but now usingeq. 5.45 to calculate the ANC leaching.Comparing Figure 5.4 with Figure 5.2 illustrates that, depending on the parame-ters of the site, bicarbonate leaching canmake a significant contribution to the overallANC leaching.

[c) Including the dissociation of organic acids:

The charge balance (eq. 5.9) and the defini-tion of ANC leaching in eq. 5.10 also includethe leaching of organic anions (RCOOle). Thishas been neglected in the SMB model for (atleast) two reasons: (i) to keep the SMBmodel simple, and/or (ii) assuming that the

negatively charged organic anion concen-tration balances the positively chargedorganic Al-complexes. However, this doesnot hold for a wide range of pH values, andat sites with high concentrations of organicmatter the contribution of organic anions toANC leaching can be considerable.

Since it is difficult to characterise (let alonemodel) the heterogeneous mixture of natu-rally occurring organic solutes, so-called‘analogue models’ are used. The simplestassumes that only monovalent organicanions are produced by the dissociation ofdissolved organic carbon:

(5.46)

where DOC is the concentration of dissolvedorganic carbon (in molC/m3), m is the concen-tration of functional groups (the ‘charge den-sity’, in mol/molC) and K1 the dissociationconstant. Both DOC and m are site-specificquantities. While DOC estimates are oftenavailable, data for m are less easy to obtain.For example, Santore et al. (1995) report val-



Figure 5.4: Critical ANC leaching (for Q=1 m/yr) including bicarbonate leaching as a function of the critical basesaturation, EBc,crit, using the same parameters as in Figure 5.2.



ues of m between 0.014 for topsoil samplesand 0.044 mol/molC for a B-horizon in theHubbard Brook experimental forest in NewHampshire.

Since a single value of K1 does not alwaysmodel the dissociation of organic acids satisfactorily, Oliver et al. (1983) havederived an empirical relationship between K1and pH:

(5.47)

with a=0.96, b=0.90 and c=0.039 (andm=0.120 mol/molC). Note that eq. 5.47 givesK1 in mol/L. In Figure 5.5 the fraction ofm·DOC dissociated as a function of pH isshown for the Oliver model and a mono-protic acid with a ‘widely-used’ valueof pK1=4.5.

Figure 5.5 shows that, depending on theamount of DOC, the contribution of organicanions to the ANC leaching, even at fairly lowpH, can be considerable.

Other models for the dissociation of organicacids have been suggested and are in use indynamic models, such as di- and tri-proticanalogue models (see, e.g., Driscoll et al.1994), or more detailed models of the speci-ation of humic substances, such as the

WHAM model (Tipping 1994). Any modelcould be used for the calculation of criticalloads as long as the dissociation dependsonly on [H], so that a critical leaching oforganic anions can be derived from [H]crit (or[Al]crit).

Figure 5.5: Fraction of organic acids (m DOC) dissociated as a function of pH for the Oliver model (solid line) andthe mono-protic model (eq.5.46) with pK1=4.5 (dashed line).

The purpose of critical loads for aquaticecosystems is to estimate the maximumdeposition(s) below which ‘significant harm-ful effects’ on biological species do notoccur. Similar to terrestrial ecosystems, thelinks between water chemistry and biologi-cal impacts cannot be modelled adequatelyat present (see also Wright and Lie 2002) assuch, water quality criteria are generallyused to derive critical loads for aquaticecosystems.

In this Section we deal only with the model-ling of critical loads of acidity for aquaticecosystems. The models are restricted tofreshwater systems, since models for marineecosystems do not seem to exist. Empiricalcritical loads of nitrogen for fresh waters, aswell as coastal and marine habitats, can befound in Section 5.2.

The following description is largely based onthe review by Henriksen and Posch (2001),but amended with new or additional informa-tion where available. Three models for calcu-lating critical loads of acidifying N and S deposition are described. Models of criticalloads for surface waters also include theirterrestrial catchment to a greater or lesserextent. Therefore, it is advised to consultSection 5.3 for some of the terminology andvariables used in the context of critical loadsfor soils.

5.4.1 The Steady-State Water Chemistry (SSWC) model


The critical load of a lake or stream can bederived from present day water chemistryusing the SSWC model, if weighted annualmean values, or estimates thereof, are avail-able. It assumes that all sulphate (SO4

2– inrunoff originates from sea salt spray andanthropogenic deposition (no adsorption orretention). The model uses Acid NeutralisingCapacity (ANC) as the variable linking water

chemistry to sensitive indicator organisms infreshwaters.

In the SSWC model (Sverdrup et al. 1990,Henriksen et al. 1992, Henriksen and Posch2001) a critical load of acidity, CL(A), is calculated from the principle that the acidload should not exceed the non-marine,non-anthropogenic base cation input andsources and sinks in the catchment minus abuffer to protect selected biota from beingdamaged, i.e.:

(5.48)

where BC*dep (BC=Ca+Mg+K+Na) is the sea-salt corrected (with Cl as a tracer; seeChapter 2) non-anthropogenic deposition ofbase cations, BCw is the average weatheringflux, Bcu (Bc=Ca+Mg+K) is the net long-termaverage uptake of base cations in the biomass (i.e., the annual average removal ofbase cations due to harvesting), and ANClimitthe lowest ANC-flux that does not damagethe selected biota. Since the average flux ofbase cations weathered in a catchment andreaching the lake is difficult to measure or tocompute from available information, a critical load equation that uses water qualitydata alone has been derived.In pre-acidification times the non-marine fluxof base cations from the lake, BC*0, is givenby (all parameters are expressed as annualfluxes, e.g. in eq/m2/yr):

(5.49)

Thus we have for the critical load from eq. 5.48:

(5.50)

where the second identity expresses thecritical load in terms of the catchment runoffQ (in m/yr) and concentrations ([X]=X/Q).



5.4 Critical Loads for AquaticEcosystems



To estimate the pre-acidification flux of basecations we start with the present flux of basecations, BC*t, given by:

(5.51)

where BCexc is the release of base cationsdue to ion-exchange processes. Assumingthat deposition, weathering rate and netuptake have not changed over time, weobtain BCexc by subtracting eq. 5.49 from eq. 5.51:

(5.52)

This present-day excess production of basecations in the catchment is related to thelong-term changes in inputs of non-marineacid anions by the so-called F-factor (seebelow):

(5.53)

For the pre-acidification base cation flux wethus get from eq. 5.52 (DX = Xt–X0):

(5.54)

The pre-acidification nitrate concentration,NO3,0, is generally assumed zero.

5.4.1.2 The F-factor

According to eqs. 5.52 and 5.53, and usingconcentrations instead of fluxes, the F-factor is defined as the ratio of change innon-marine base cation concentrations dueto changes in strong acid anion concentra-tions (Henriksen 1984, Brakke et al. 1990):

(5.55)

where the subscripts t and 0 refer to presentand pre-acidification concentrations,respectively. If F=1, all incoming protons areneutralised in the catchment (only soil acidi-fication), at F=0 none of the incoming protons are neutralised in the catchment(only water acidification). The F-factor wasestimated empirically to be in the range0.2–0.4, based on the analysis of historicaldata from Norway, Sweden, U.S.A. andCanada (Henriksen 1984). Brakke et al.(1990) later suggested that the F-factorshould be a function of the base cation concentration:

(5.56)

where [S] is the base cation concentration atwhich F=1; and for [BC*]t>[S] F is set to 1.For Norway [S] has been set to 400 meq/m3

(ca. 8 mgCa/L) (Brakke et al. 1990).

In eq. 5.56 the present base cation concen-tration is used for practical reasons. To render the F-factor independent from thepresent base cation concentration (and tosimplify the functional form), Posch et al.(1993) suggested the following relationshipbetween F and the pre-acidification basecation concentration [BC*]0:

(5.57)

where [B] is a scaling concentration estimat-ed to be 131 meq/m3 from paleolimnologicaldata from Finland (Posch et al. 1993).Inserting this expression into eq. 5.55 gives anon-linear equation for [BC*]0 which has tobe solved by an iterative procedure. The twoexpressions for the F-factor give similarresults when used to calculate critical loadsfor surface waters in Norway (see Henriksenand Posch 2001).

The use of the F-factor, defined as a functionof the base cation concentration (Henriksen1984) was originally derived from Norwegianlake data. In Norway the range of runoff iswide (0.3–5 m/yr), with an average of about

1 m/yr. In other Nordic countries, such asSweden and Finland, runoff is low comparedto most of Norway. The weathering rate of acatchment is largely dependent on thebedrock and overburden. Thus, catchmentswith similar bedrock and overburden charac-teristics should have similar weatheringrates. If one catchment has a high runoff,e.g., 2 m/yr, and another one has a lowrunoff, e.g., 0.3 m/yr, their base cation fluxeswill be similar, but their concentrations willdiffer considerably. Thus, in the F-factor (eq. 5.56) the BC-flux should be used insteadof the concentration (Henriksen and Posch2001):

(5.58)

where S is the base cation flux at which F=1.For Norway, S has been estimated at 400 meq/m2/yr. Again, if Q·[BC*]t>S, F is setto 1. Similarly, fluxes could be introduced forthe formulation in eq. 5.57.

5.4.1.3 The non-anthropogenic sul-phate concentration

The pre-acidification sulphate concentrationin lakes, [SO4*]0, is assumed to consist of aconstant atmospheric contribution and ageologic contribution proportional to theconcentration of base cations (Brakke et al.1989):

(5.59)

The coefficients in this equation, estimated

for different areas and by different authors,are summarised in Table 5.16.

Details on the procedures and data sourcesfor estimating these coefficients can befound in the references given. In Henriksenand Posch (2001) it is shown that theexceeded area for Norwegian lakes (in 1994)is influenced very little by the choice of coef-ficients for calculating non-anthropogenicsulphate. Similar results have been reportedfor Irish lakes (Aherne and Curtis 2003).

Larssen and Høgåsen (2003) suggested thatthe atmospheric contribution in eq. 5.59 bederived from background S deposition, asestimated by atmospheric transport models:

(5.60)

For southern Norway, Sdep,0 is about 50 mgS/m2/yr from the EMEP long-rangetransport model, i.e., about 3 meq/m2/yr.With Q varying between 0.5 and 1 m/yr thisresults in an atmospheric contribution to[SO4*]0 of about 3–6 meq/m3.

The SSWC model has been developed forand is particularly applicable to dilute oligotrophic waters located on granitic andgneissic bedrock with thin overburden, suchas large parts of Fennoscandia, Scotland,Canada and Ireland. In such areas, surfacewaters are generally more sensitive to acidinputs than soils. The model assumes that allsulphate in runoff originates from depositionalone, except for a small geologic contri-bution. In areas where the geological condi-tions lead to more alkaline waters, the SSWCmodel has to be modified, since significant



Table 5.16: Constants to estimate the non-anthropogenic sulphate concentration with eq. 5.59, derived from empirical data (N is the number of samples and r is the correlation coefficient).



amounts of sulphate from geologicalsources can be present in the runoff water. Amodification for this kind of conditions hasbeen developed for Slovakia (Závodský et al.1995).

5.4.1.4 The ANC-limit

Lien et al. (1996) analysed the status of fishand invertebrate populations in the contextof surface water acidification and loss ofANC in Norwegian lakes and streams. Thedata for fish came from populations in 1095lakes, mostly from the regional lake surveycarried out in 1986 (Henriksen et al. 1988,1989). The critical level of ANC varied amongfish species, with Atlantic salmon being themost sensitive, followed by brown trout.They concluded that Atlantic salmonappeared to be a good indicator of acidi-fication of rivers, and trout seemed to be auseful indicator for acidification of lakes.Based on an evaluation of fish and inverte-brate populations, a critical lower limit of[ANC]=20 meq/m3 was suggested as the tolerance level for Norwegian surface waters(Lien et al. 1996; see Figure 5.6). This limithas been widely used (Kola, northern Russia:Moiseenko 1994; southern central Alps:Boggero et al. 1998; China: Duan et al. 2000);however, it has been set to zero in the UnitedKingdom (CLAG 1995) and to 40 meq/m3 insouth-central Ontario, Canada (Henriksen etal. 2002).

Lydersen et al. (2004) argued that the ANC-limit should be corrected with the amount of

organic acids present in the lake. And theyshowed that the fit between observed fishstatus and ANC can be (slightly) improved, ifan ‘organic acid adjusted’ ANC, [ANC]oaa, isused (instead of the ‘standard’ ANC). Theydefine this quantity as:

(5.61)

where m·TOC is the total organic carbonexpressed in meq/m3 (m being the chargedensity). Such a correction leads to a lowerANC-limit, i.e., higher critical loads.

Figure 5.6 indicates that in the ANC range0–50 meq/m3 there is a decreasing probabil-ity from about 50 to 0% of damage to fishpopulations. The lakes studied receive verylow to very high (for Norway) levels of depo-sition, thus including a wide range of affected lakes. This implies that for a givenANC-value lakes of varying sensitivity exist,receiving varying amounts of deposition.This could reflect that fish have respondedto the same ANC differently in different lakes,indicating that a catchment-dependent ANC-limit would be more appropriate than afixed value for all lakes. In other words, everylake has its own characteristic ANC-limit (inthe range shown in Figure 5.6). Less sensitive lakes, i.e., lakes with higher criticalloads, should have a higher ANC-limit, sinceless sensitive ecosystems will have a higherbiological variety/diversity and thus require ahigher ANC-limit to keep that diversity intact.

Figure 5.6: Relationship between the ANC concentration in lake water and the probability for damage and extinc-tion of fish (brown trout) populations in lakes, derived from Norwegian data (after Lien et al. 1996).

The simplest functional relationship with thisfeature is a linear relationship between[ANC]limit and the critical load CL:

(5.62)

This gives the following implicit equation forthe critical load (see eq. 5.50):

(5.63)

which yields after re-arranging for CL:

(5.64)

and thus from eq. 5.62:

(5.65)

This is a special case of a more generalexpression derived earlier using somewhatdifferent arguments (Henriksen et al. 1995).As for the constant [ANC]limit used earlier,the proportionality constant k should bederived from data. If we assume that forCL=0, the [ANC]limit=0; if we further assumethat for a critical load of 200 meq/m2/yr theANC-limit should not exceed 50 meq/m3, ashas been assumed in Sweden, we arrive at ak-value of 50/200 = 0.25 yr/m. In addition, forCL-values above 200 meq/m2/yr we set the[ANC]limit to the constant value of 50meq/m3. The value of k is derived from experience in the Nordic countries and, assuch, reflects the geology, deposition history and biological diversity (fish species)of that region. For different regions other k-values may be more appropriate.

5.4.2 The empirical diatom model

The empirical diatom model is an alternativeapproach to the SSWC model and is devel-

oped from paleolimnological data (Battarbeeet al. 1995). Diatom assemblages in coresfrom acidified lakes usually show that priorto acidification the diatom flora, and there-fore water chemistry, changed little overtime. The point of acidification is indicatedby a shift towards a more acidophilousdiatom flora. Diatoms are amongst the mostsensitive indicators of acidification in fresh-water ecosystems, hence it can be arguedthat the point of change in the diatom recordindicates the time at which the critical loadfor the site was exceeded.The acidification status (as defined bydiatom analyses) of 41 sites in the UnitedKingdom (UK) was compared to site sensitivity (defined by lake-water calciumconcentrations) and current deposition loading. The optimal separation of acidifiedand non-acidified sites is given by a [Ca]:Sdepratio of 94:1 (Battarbee et al. 1995), acidifiedsites having a ratio less than 94:1. This critical ratio, determined by logistic regression, can be used to define critical sulphur loads for any site, including streams.Critical load values are calculated from pre-acidification calcium concentrationsusing the F-factor (Brakke et al. 1990). Forexample, the critical sulphur load for a lakewith a [Ca]0-value of 40 meq/m3 is approxi-mately 0.43 keq/ha/yr.

The diatom model has been adapted to provide critical loads, and critical loadexceedances, for total acidity (sulphur andnitrogen). Exceedance values for total acidity require a measure of the fraction ofdeposited nitrogen leached to the surfacewaters. This is calculated from the differ-ences between the ratios of sulphate/nitratein the water and in the deposition at the site.In this way the fraction of the nitrogen depo-sition contributing to acidification, fN, isadded to the value of sulphur deposition toprovide a total ‘effective’ acid deposition:

(5.66)





This model assumes equilibrium betweensulphur deposition and sulphate in water,and only applies to sites with no additionalcatchment nitrogen inputs. The diatommodel has been re-calibrated for total acidity loads by substituting total effectiveacid deposition for sulphur deposition. Theresulting critical ratio is 89:1, slightly lowerthan when considering sulphur alone. Thebasic equation for the critical load of totalacidity in the empirical diatom model istherefore as follows:

(5.67)

where CL(A) is in keq/ha/yr and [Ca*]0 inmeq/m3. The pre-acidification Ca-concentra-tion is calculated as:

(5.68)

with

(5.69)

and [SCa] is the Ca-concentration at whichFCa=1. It can vary between 200 and 400meq/m3, depending on location. In the UKcritical loads mapping exercise a value of[SCa]=400 meq/m3 has been used, and inwaters with [Ca*]t>[SCa], FCa was set to 1.The pre-acidification nitrate concentration,[NO3]0, is assumed zero. The pre-acidifica-tion sea-salt corrected sulphate concentra-tion, [SO4*]0, is estimated according to eq. 5.59 (Brakke et al.1989).

The diatom model has been calibrated usingsites and data from the UK. However, amajor advantage of the approach is that predictions for any lake can be validated byanalysing diatoms in a sediment core. In thisway the applicability of the model to sitesoutside the UK can be tested.

5.4.3 The First-order Acidity Balance (FAB) model

The First-order Acidity Balance (FAB) modelfor calculating critical loads of sulphur (S)and nitrogen (N) for a lake takes into accountsources and sinks within the lake and its terrestrial catchment. The original version ofthe FAB model has been developed andapplied to Finland, Norway and Sweden inHenriksen et al. (1993) and further describedin Posch et al. (1997). A modified versionwas first reported in Hindar et al. (2000) andis described in Henriksen and Posch (2001).The FAB model is designed to be equivalentto the Simple Mass Balance model for acatchment, and it largely follows its deriva-tion (see Section 5.3), the main differencebeing that the leaching of ANC is modelledaccording to the SSWC model (see section5.4.1).


The lake and its catchment are assumedsmall enough to be properly characterisedby average soil and lake water properties.With A we denote the total catchment area(lake + terrestrial catchment), Al is the lakearea, Af the forested area and Ag the areacovered with grass/heath land. We have Al + Af + Ag £ A, and a non-zero differencerepresents a land area on which no transfor-mations of the deposited ions take place(‘bare rock’).

Starting point for the derivation of the FABmodel is the charge balance (‘acidity balance’) in the lake water running off thecatchment:

(5.70)

where BC* stands for the sum of (non-marine) base cations and ANC is theacid neutralising capacity. In the aboveequation we assume that the quantities aretotal amounts per time (e.g. eq/yr). In orderto derive critical loads we have to link theions in the lake water to their depositions,taking into account their sources and sinks

in the terrestrial catchment and in the lake.

For X = S, N and BC the mass balance in thelake is given by:

(5.71)

where Xin is the total amount of ion Xentering the lake and Xret the amount of Xretained in the lake. The in-lake retention ofS and N is assumed to be proportional to theinput of the respective ion into the lake:

(5.72)

where 0 £ rX £ 1 is a dimensionless retentionfactor. Thus the mass balances for the lakebecome:

(5.73)

The total amount of sulphur entering the lakeis given by:

(5.74)

where Sdep is the total deposition of S per unitarea. Immobilisation, reduction and uptakeof sulphate in the terrestrial catchment areassumed negligible, and sulphate ad/des-orption is not considered since we modelsteady-state processes only. eq. 5.74 statesthat all sulphur deposited onto the catch-ment enters the lake, and no sources orsinks are considered in the terrestrial catch-ment.

In the case of nitrogen we assume thatimmobilisation and denitrification occur bothin forest and grass/heath land soils, whereasnet uptake occurs in forests only (equallingthe annual average amount of N removed byharvesting); the deposition onto the remain-ing area (lake + ‘bare rocks’) enters the lakeunchanged. Thus the amount of N entering

the lake is:

(5.75)

where Ndep is the total N deposition, Ni is thelong-term net immobilisation of N (whichmay include other long-term steady-statesources and sinks; see Chapter 5.3), Nde is Nlost by denitrification, and Nu the net growthuptake of N, all per unit area. The symbol (x)+or x+ is a short-hand notation for max{x, 0},i.e., x+= x for x > 0 and x+ = 0 for x £ 0. Theeffects of nutrient cycling are ignored andthe leaching of ammonium is considerednegligible, implying its complete uptakeand/or nitrification in the terrestrial catch-ment.

While immobilisation and net growth uptakeare assumed independent of the N deposi-tion, denitrification is modelled as fraction ofthe available N:

(5.76)

where 0 £ fde < 1 is the (soil-dependent) denitrification fraction. The above equationis based on the assumption that denitrifica-tion is a slower process than immobilisationand growth uptake. Inserting eq. 5.76 intoeq. 5.75 one obtains:

(5.77)

If sufficient data for quantifying the sourcesand sinks of base cations in the catchment,such as deposition, weathering and uptake,





are available, the runoff of base cations(BC*runoff) could be described in the sameway as S and N. This would be in analogy tothe derivation of the SMB model for (forest)soils. Alternatively, water quality data can beused to quantify the runoff of base cationsand ANC, as is done in the SSWC model (seesection 5.4.1).

To arrive at an equation for critical loads, alink has to be established between a chemi-cal variable and effects on aquatic biota. Themost commonly used criterion is the so-called ANC-limit (see above), i.e. a minimumconcentration of ANC derived to avoid‘harmful effects’ on fish: ANCrunoff,crit =A·Q·[ANC]limit.

Defining Lcrit = (BC*runoff - ANCrunoff,crit)/A,inserting eq. 5.74 and 5.77 into eq. 5.73 andeq. 5.70 and dividing by A yields the follow-ing equation to be fulfilled by critical deposi-tions (loads) of S and N:

(5.78)

where we have defined:

(5.79)

Eq. 5.78 defines a function in the (Ndep, Sdep)-plane, the so-called critical load function(see Figure 5.7), and in the following we willlook at this function in more detail. The general form of the critical load function is:

(5.80)

with

(5.81)

The quantity MN and the dimensionless coefficient bN depend on Ndep:

(a) Ndep £ Ni: In this case (Ndep-Ni)+ = 0 and(Ndep-Ni-Nu)+ = 0, which means that all Nfalling onto forests and grassland is immo-bilised and only the N deposition fallingdirectly onto the lake and ‘bare rocks’ contributes to the leaching of N:

(5.82)

(b) Ni < Ndep £ Ni + Nu: In this case (Ndep-Ni)+= Ndep–Ni, but (Ndep-Ni-Nu)+ = 0, meaning thatall N deposition falling onto forests is immo-bilised or taken up, but N falling onto theother areas is (partially) leached:

(5.83)

(c) Ndep > Ni + Nu: Some N deposition isleached from all areas:

(5.84)

The maximum critical load of sulphur isobtained by setting Ndep = 0 in eq. 5.78:

(5.85)

Setting Sdep = 0 and considering the three dif-ferent cases for Ndep, gives the followingexpression for the maximum critical load fornitrogen:

(5.86)

?

5.4.3.2 Systems of lakes

The above derivation of the FAB model is for(small) headwater lakes only. Critical loadsare generally calculated for such lakes, sincelakes with (many) upstream lakes tend tohave larger catchments, and several (implicit) assumptions of the FAB model, e.g.uniform depositions, will be violated.Nevertheless, in some areas systems oflakes can be found on a small scale, andtherefore a model for such systems is desirable.

When computing the critical load of aciditywith the SSWC model (which uses annualaverage lake water chemistry) for a lakereceiving runoff from upstream lakes, oneimplicitly computes the critical load for thatlake including all its upstream lakes, sincewater samples taken from (the outlet of) thelowest lake is a mixture of the water of thatlake and all its upstream lakes.Consequently, when applying the FAB modelto such a lake, one has to be aware that onealso computes the critical load for the wholesystem of lakes and thus must take intoaccount the catchment and lake characteris-tics of all lakes in the system. To do this in amore explicit way, two methods for com-puting the critical load of a system of lakeshave been developed (Hindar et al. 2000).Both require the same input data, but theydiffer in the complexity of the calculations

involved. The formulae will not be derivedhere, and the interested reader is referred tothe literature (see also Hindar et al. 2001),where also the differences between themethods are demonstrated, using data fromlake systems in the Killarney Provincial Parkin Ontario, Canada. An application to lakes inthe Muskoka river catchment (Ontario,Canada) can be found in Aherne et al. (2004).

5.4.4 Input data

In addition to the data required for the SSWCand diatom model (runoff and concentra-tions of major ions in the lake runoff water),the FAB model needs also information on (a)the area of lake, catchment and differentland cover classes, (b) terrestrial nitrogensinks, and (c) parameters for in-lake reten-tion of N and S.

Runoff:The runoff Q is the amount of water leavingthe catchment at the lake outlet, expressedin m/yr. It is derived from measurements orcan be calculated as the difference betweenprecipitation and actual evapotranspiration,averaged over the catchment area. A long-term climatic mean annual value should betaken. Sources for data and models forevapotranspiration can be found in Section5.3.



Figure 5.7: Piece-wise linear critical load function of S and acidifying N for a lake as defined by catchment proper-ties. Note the difference with the critical load function for soils (see Figure 5.1). The grey area below the CL functiondenotes deposition pairs resulting in an ANC leaching greater than Q·[ANC]limitt (non-exceedance of critical loads; seeChapter 7).



Ion concentrations:In addition to runoff, the concentrations ofmajor ions in the runoff water, i.e. sulphate,nitrate and base cations, are needed to calculate SSWC critical loads, and thesecome from the analysis of representativewater samples.

The critical load for a site should be calcu-lated with yearly flow-weighted averagechemistry and yearly average runoff. Sincesuch values are not available for a largenumber of lakes, critical loads are mostlycalculated on the basis of a single sampleconsidered representative of yearly flow-weighted averages. A sample collectedshortly after the fall circulation of a lake isgenerally assumed to fulfil this purpose. Tocheck this claim, Henriksen and Posch(2001) compared critical load values calcu-lated from yearly flow-weighted averageconcentrations with critical loads calculatedfrom a single fall value for sites for whichlong-term data series are available. Resultsfor seven Norwegian catchments show thatthe single fall value is fairly representative forthe annual average chemistry. Similarly,results from eight Canadian catchmentsshow that a single spring sample is fairlyrepresentative of the annual average chemistry (Henriksen and Dillon 2001).

Lake and catchment characteristics:The area parameters A, Al, Af and Ag, whichare needed in the FAB model, can generallybe derived from (digital or paper) maps.

Terrestrial N sinks:The uptake of N can be computed from theannual average amount of N in the harvestedbiomass. If there is no removal of trees fromthe catchment, Nu=0.Ni is the long-term annual immobilisation(accumulation) rate of N in the catchmentsoil. Note that at present immobilisation maybe substantially higher due to elevated Ndeposition.

The denitrification fraction fde depends onthe soil type and its moisture status. In earlier FAB applications it has been esti-mated from the fraction of peat-lands, fpeat,in the catchment by fde=0.1+0.7·fpeat

(Posch et al. 1997).

For more details on these parameters seeSection 5.3.

In-lake retention of N and S:Concerning in-lake processes, the retentionfactor for nitrogen rN (see eq. 5.72) is modelled by a kinetic equation (Kelly et al. 1987):

(5.87)

where z is the mean lake depth, t is the lake’sresidence time, r is the lake:catchment ratio(=Al/A) and sN is the net mass transfer coefficient. There is a lack of observationaldata for the mass transfer coefficients, espe-cially from European catchments, but Dillonand Molot (1990) give a range of 2–8 m/yr forsN. Values for Canadian and Norwegiancatchments are given in Kaste and Dillon(2003).

An equation analogous eq. 5.87 for rS with amass transfer coefficient sS is used to modelthe in-lake retention of sulphur. Baker andBrezonik (1988) give a range of 0.2–0.8 m/yrfor sS.

5.5.1 General methodological aspects of mapping critical loads of heavy metals

5.5.1.1 Calculation of different types of critical loads

The method to calculate critical loads ofheavy metals is based on the balance of allrelevant metal fluxes in and out of a consid-ered ecosystem in a steady state situation.In order to keep the approach compatiblewith the simple mass balance approachused for nitrogen and acidity, the internalmetal cycling within an ecosystem isignored, such that calculations can be keptas simple as possible. In consequence thecritical load of a metal can be calculatedfrom the sum of tolerable outputs from theconsidered system in terms of net metaluptake and metal leaching.

The assumption of steady state signifies thatthe concentration in the system does notchange in time because the amount of heavymetal entering the system is equal to theamount that leaves the system. The validityof this assumption depends on the magni-tude of the time scales of the various inputand output processes. If e.g. a metal sorbsvery strongly to the soil, it may take a longtime (up to hundreds of years), before asteady state is reached. This has to be keptin mind when comparing a present load withthe critical load (De Vries and Bakker 1998).Critical loads of cadmium (Cd), lead (Pb) andmercury (Hg) can be calculated in depend-ence on the receptors and the metal of con-cern. Critical limits of these heavy metalsaddressing either ecotoxicological ecosys-tem effects or human health effects arederived with specific approaches. Criticalloads on the basis of such limits should becalculated separately for aquatic and terres-trial ecosystems. In consequence four typesof critical loads can be derived for eachmetal, an overview is provided in Table 5.17,which is however not a complete review of

possible effects of these metals. Indicators of effects on ecosystems listed inTable 5.17 are mainly ecotoxicologicaleffects. Secondary poisoning through thefood chain has also been studied (De Vries etal. 2003). These effects give partly morestringent critical limits, however their modelling includes more uncertainties and istherefore not considered in this manual.

Critical loads for terrestrial ecosystemsaddressing human health effects can be cal-culated, either in view of not violating foodquality criteria in crops or in view of groundwater protection (keeping quality criteria fordrinking water of WHO 2004). An appropriateindicator for critical load calculationsaddressing human health effects via foodintake is the Cd content in wheat. Keeping aconservative food quality criterion forwheat, as described in Section 5.5.2.2.1,protects at the same time against effects onhuman health via other food and foddercrops (including also the quality of animalproducts, since the pathway of Cd to wheatleads to the lowest critical Cd content in soilsaccording to De Vries et al. (2003). Such critical load calculations are in principle alsopossible for lead, and for other food and fodder crops, if the soil-plant transfer can bedescribed with sufficient accuracy and canbe done in addition on a voluntary basis.

Among terrestrial ecosystems, critical loadsof Cd and Pb are to be calculated from theviewpoint of ecotoxicology for areas covered by non-agricultural land (forests,semi-natural vegetation) or agricultural land(arable land and grassland). Organic forest(top)soils are considered as the only criticalreceptor with respect to atmospheric Hgpollution, based on knowledge on effects onmicrobial processes and invertebrates (Meiliet al. 2003a). The critical exposure of terrestrial ecosystems to atmospheric Hgpollution can be calculated in much thesame way as for Pb and Cd by a simple massbalance, as discussed in Section 5.5.3.2.



5.5 Critical Loads of cadmium, lead and mercury



For aquatic ecosystems the critical limits ofPb and Cd are related to ecotoxicologicaleffects, while human health effects by thispathway are less relevant and therefore notconsidered here. Critical limits of Hg refer toboth human health effects (Hg concentrationin fish and other animals that serve as a foodsource to humans) and ecotoxicologicaleffects, since microbiota and higher wildlifeitself may also be affected.

Although it might be useful to calculate andmap each of the different types of criticalloads and the critical Hg level in precipitationseparately for comparison purposes, the aimis ultimately to provide maps for at most fourcritical loads per metal (or Hg level, respec-tively) related to:

-Ecotoxicological effects for all terrestrial ecosystems.

-Human health effects for all terrestrial ecosystems.

-Ecotoxicological effects for all aquatic ecosystems.

-Human health effects for all aquatic ecosystems.

If different indicators within each category(map) have been considered (e.g. Cd inwheat and Cd in soil drainage water in viewof ground water protection for humanhealth), the final map should indicate theminimum critical Cd load for both effects tohuman health. The reason for providing dif-ferent critical loads for different types ofecosystems is because the critical load forterrestrial ecosystems does not automati-cally protect aquatic ecosystems, receivingmuch or most of their metal load by drainagefrom the surrounding soils, and vice versa.

Table 5.17: Four types of critical loads of Pb, Cd, Hg, related receptors and indicators

*) In italics: these calculations can be done in addition on a voluntary basis. To perform such calculations, moreinformation on the derivation of critical limits based on critical metal contents in food/fodder crops and in animal

products is given in Annex 2 and 3, respectively, of the background document (De Vries et al. 2004b).

A critical load indicates only the sensitivity ofan ecosystem against the anthropogenicinput of the metal of interest. It implies apotential risk at sites where the critical loadis exceeded. In agricultural ecosystems, theexceedance of critical loads of heavy metalsis not only determined by atmosphericinputs (being generally the only source innon-agricultural ecosystems), but by totalinputs, including fertilizer and animal manureinputs.

5.5.1.2 Limitations in sites that allow critical load calculations

Critical load calculations can not be carriedout for sites with:

-Negative water balances, since there is no leaching but a seepage influx of water, leading to accumulation of salts and very high pH; such regions do, how-ever, hardly occur in Europe.

-Soils with reducing conditions (e.g. wet-lands), because the transfer functions do not apply for such soils. In the topsoil, to which the critical load calculations apply, such situations do, however, hardly occur apart from water logged soils where the simplified critical load calcula-tion can not be applied anyhow because of a deviating hydrology.

Weathering inputs of metals are neglecteddue to i) low relevance of such inputs and ii)high uncertainties of respective calculationmethods. It is, however, recommended touse estimates of weathering rate to identifysites with a high geogenic metal input, wherenatural weathering may already exceed thecritical load. This should be considered,when critical limits and loads exceedancesare to be interpreted. For methods to calcu-late weathering rates, see De Vries andBakker (1998) and Hettelingh et al. (2002).More information on how sites with highgeogenic contents of metals can be identi-fied are described in Farret et al. (2003). Themost important information is summarised inAnnex 6 of the background document (DeVries et al. 2004b).

5.5.1.3 Definitions and symbols/ abbreviations used in critical load calculations

General definitions of critical loads, criticallevels and exceedances, and others can befound in the related chapters of theModelling and Mapping Manual. The follow-ing definitions refer specifically to the appli-cation in the context of critical loads ofheavy metals.

Definitions

The receptor is a living element of the envi-ronment that is subject to an adverse effect.It can be a species of interest includinghuman beings, or several species consid-ered representative of a larger group (e.g.plants, soil invertebrates, fish, algae, etc), orthe whole ecosystem (typically the subject ofinterest in the critical load approach).

The critical limit is a concentration thresholdwithin the ecosystem, based on adverseeffects, i.e. it is a short expression of “effect-based critical limit”. Below this critical limitsignificant harmful effects on human healthor specified sensitive elements of the envi-ronment do not occur, according to presentknowledge. To avoid confusion, limits thatare not based on effects should not be called“critical limits”.

The critical load is the highest total metalinput rate (deposition, fertilisers, otheranthropogenic sources) below which harm-ful effects on human health as well as onecosystem structure and function will notoccur at the site of interest in a long-termperspective, according to present know-ledge. The critical load is derived from thecritical limit through a biogeochemical fluxmodel, assuming steady-state for the fluxesas well as chemical equilibrium (which is atheoretical situation in an undeterminedfuture, consistent with concepts of sustain-ability). For this purpose the critical limit hasto be transformed to a critical total concen-tration of the metal in the output fluxes bywater (leaching from the soil or outflow froman aquatic ecosystem).





An overview of used symbols and abbrevia-tions is given below (Table 5.18).

Some general abbreviations:

M = a flux of a metal M[M] = a content (in soil, plants, other

biota) or a concentration (in a liquid) of a metal M

[M]...(crit) = a critical content (in soil, plants, other biota) or a critical concen-

tration (in a liquid) of a metal M,not explicitly explained in table5.18 for all the individual contents or concentrations

f = a fractionc = a factor for conversion of units,

not explained in the table

sdw = in soil drainage water

sw = in surface water

Table 5.18: Symbols and abbreviations used in the calculation of critical loads of heavy metals

5.5.1.4 Stand-still approach versus calculation of critical limit exceedance

The harmonised methodological basis for afirst preliminary calculation and mapping ofcritical loads of Cd and Pb related to ecotox-icological effects (Hettelingh et al. 2002),was based on a guidance document (DeVries et al. 2002). In this document a stand-still approach, which aims at avoiding any(further) accumulation of heavy metals in thesoil, was also included as an alternative tothe effect-based approach. This method is,however, not included in this manual since itimplies the continued addition of metals onhistorically polluted soils with high leachingrates. The current leaching may then alreadyimply significant effects, both on terrestrialas well as aquatic ecosystems receiving thedrainage water from the surrounding soils,and is thus not per se acceptable in the longterm. Furthermore, it does lead to criticalload exceedance at soils which stronglyadsorb heavy metals, whereas the effectdoes occur through the soil solution.

Instead, it is suggested to calculate criticalconcentrations of metals in the soil, the soildrainage water or the surface water basedon the critical limits and compare these tothe present soil or water metal concentra-tions to assess the critical limit exceedancein the present situation. This implies that onehas to map the present metal concentrationsin the country (expressed as total or reactivesoil contents, total dissolved concentrationsor even free ion concentrations). Such acomparison can be seen as an intermediatestep for dynamic models for heavy metals. Ifthe present soil metal content exceeds thecritical concentration (limit), the metal inputhas to be less than the critical load to reachthe critical concentration at a defined timeperiod. In the reverse case, the metal inputcan be larger than the critical load for adefined time period not exceeding duringthat period the critical concentration.However, only keeping the critical load willnot lead to exceedance of the critical limit inthe long run. More information on how to cal-culate the critical concentration is given inthe background document.





5.5.2 Terrestrial ecosystems

5.5.2.1 Simple steady-state mass balance model and related input data

5.5.2.1.1 Steady-state mass balance model

The method to calculate critical loads ofheavy metals for terrestrial ecosystems isfocusing in particular on the upper soil layer.The critical load of a metal can be calculatedas the sum of tolerable outputs from thisconsidered soil layer by harvest and leach-ing minus the natural inputs by weatheringrelease (De Vries and Bakker, 1998). Becauseweathering causes only a minor flux ofmetals in topsoils, while uncertainties ofsuch calculations are very high, the modelwas further simplified by assuming thatweathering is negligible within the topsoiloutside ore-rich areas. As mentioned in theintroduction of this chapter, the calculationof weathering rates is recommended to iden-tify areas, where the natural input exceedstolerable outputs; and such sites can beexcluded from the database, subject to deci-sion by the National Focal Centres.

The described approach implies that thecritical load equals the net uptake by forestgrowth or agricultural products plus anacceptable metal leaching rate:

(5.88)

where:CL(M) = critical load of a heavy metal

M (g ha-1 a-1)Mu = metal net uptake in harvestable

parts of plants under critical load conditions (g ha1 a-1)

Mle(crit) = critical leaching flux of heavy metal M from the considered soil layer (g ha-1 a-1), whereby only the vertical drainage flux is consid-ered

The notation has been related to the critical load equations for acidity and nutrient nitro-gen: M stands for flux of a heavy metal and

can be substituted by the chemical symbolof the individual metal (Cd, Pb, Hg) underconsideration. The critical metal leachingMle(crit) refers to the total vertical leachingrate, including dissolved, colloidal and par-ticulate (metal) species in the drainagewater. For a critical load, the critical metalleaching is based on a critical (toxic) metalconcentration in soil or the (free ion or total)metal concentration in soil water.

In mass balance models for Hg, re-emission(volatilization) of deposited Hg occurs as anadditional flux. This flux can, however, beignored when calculating critical loads of Hg,because this re-emission is treated as part ofthe atmospheric net deposition in the mod-elling by EMEP MSC-E (Ryaboshapko et al.1999, Ilyin et al. 2001). Therefore, in order toavoid double consideration in the calculationof critical load exceedances, it should beexcluded from the critical loads model.

Appropriate and consistent calculation ofcritical loads for terrestrial ecosystemsrequires a consistent definition of the topsoilcompartment and its boundaries. The depthcan be variable. Relevant boundaries havebeen derived considering on one hand theexpected probability of adverse impacts onthe main target groups of organisms (plants,soil invertebrates, soil microbiota), or groundwater quality, and on the other hand theoccurrence and location of relevant metalfluxes within the soil profile:

-For Pb and Cd it is assumed thatecotoxicological effects as well as the main proportion of uptake by plants occur in (from) the organic layer (O horizon) and the humus rich (top)soil horizons (Ah , Ap). Therefore the depth of the biological active topsoil (zb) should be considered for arable land, grass-land, and forests as far as the criticalload calculations are addressing ecotox-icological effects, or the protection offood/fodder quality, respectively. For forest soils covered by an organic layer, the critical loads for both the organic layer, and the upper mineral horizon should be calculated separately. In these

cases the most sensitive of both layers should be presented in the critical loads map. For all terrestrial ecosystems the maximum depth of the topsoil (zb) to be considered is the lower boundary of the uppermost mineral horizon (in most soil classification systems called the A-horizon).

Default values of zb are:

for forests: 0.1 m (O and/or Ah horizon)grassland: 0.1 m (Ah horizon)arable: 0.3 m (Ap horizon, plough

layer)

-Regarding Hg, the critical receptor interrestrial ecosystems is the organictopsoil (mor or humus layer) of forest soils (O-horizon excluding litter, which is sometimes divided into L, F and H horizons), where microbial processes are suspected to be affected. For calcu-lating the critical load of Hg in forests, the topsoil is therefore defined as the humus layer, excluding underlyingmineral soil layers.

Note, that for calculations of critical loadswith respect to protection of groundwaterquality the entire soil column has to beincluded. However, it is preliminarily notplanned within the critical loads work tomodel the whole pathway of the metal fluxwith drainage water, considering the bindingcapacity of layers between rooting zone andupper groundwater. Therefore, for simplifica-tion the critical leaching of metals from theviewpoint of ground water protection is cal-culated by multiplying the drainage waterflux below the rooting zone (soil depth = z)with the critical limit for drinking water (see5.5.2.2.2).

5.5.2.1.2 Heavy metal removal from the topsoil by net growth and harvest of plants

For critical load calculations, the removal ofheavy metals refers to a future steady-statelevel where critical limits in the ecosystemcompartments are just reached (criticalloads conditions). The calculation of a critical removal of metals on the basis of a

critical concentration for soil solution ishardly practicable since for many metalsthere are no clear relationships betweenconcentrations in soil solution (or even freemetal ions) and the content of the metals inharvestable part of the plants. Reasons areamongst others the plant specific exclusionof metals from root uptake or accumulationin specific tissues (detoxification). An excep-tion is the transfer of Cd from soil to wheatgrains, used to calculate critical loads relat-ed to food quality criteria (see 5.5.2.2.1).

Therefore a simplified approach is proposedto describe the tolerable removal of heavymetals by biomass net uptake. The averageyield (or growth increment) of harvestablebiomass is multiplied with the heavy metalcontent in harvestable plant parts and with afactor to account for the fraction of metaluptake from the relevant soil layer relative tothe uptake from the total rooting zone (eq. 5.89):

(5.89)

where:Mu = metal net uptake in harvestable

parts of plants under critical load conditions (g ha-1 a-1) (see eq. 5.88),

fMu = fraction of metal net uptake within the considered soil depth (zb or z), accounting also for metal uptake due to deposition on vegetation surfaces (–); in calculations ofcritical loads to protect ground water, fMu = 1, otherwise fMu is a value between 0 and 1

Yha = yield of harvestable biomass (dry weight) (kg ha-1 a-1),

[M]ha = metal content of the harvestable parts of the plants (g kg-1 dw), including also metals deposited on vegetation surfaces (when the metal content is given in mg kg-1 dw, the value has to be divided by 1000).

As a default approximation, a root uptakefactor (fMu,zb) of 1 can be used for all ecosys-





tem types, assuming that most uptake ofnutrients and pollutants occurs in the topsoil. In forests values around 80 % havebeen reported for uptake from the humuslayer alone (based on lead isotopes in Scotspine, Bindler et al. 2003). Thus, for calcula-tions referring to the humus layer, fMu,zb maybe 0.8, but, if the top of the underlying mineral soil is included in the calculations,fMu,zb is likely to approach 1, also in forests.If fMu,zb is 1, the uptake from the upper horizon is equal to that of the entire rootingdepth (assumed to be limited to the depthwhere 90 % of the root biomass is distrib-uted). This implies that there is no differencein the uptake calculation of critical loadsrelated to ecotoxicological effects and inview of ground water protection. Moredetailed values of fMu,zb may be used, if information is available.

Data on yields for forests can in principle beobtained from the database of critical loadsof acidity and nutrient nitrogen. Data onyields in agro-ecosystems are available fromrelated statistics of the countries. The spatialpattern can be derived using information onland use as well as on soil quality and climate.

To get data on metal contents in harvestablebiomass, studies from relatively unpollutedareas should be used. Median values (oraverages) of metals contents in plants fromsuch databases do in general not exceedquality criteria for food and fodder crops orphyto-toxic contents, respectively. Relatedfluxes can therefore be considered as toler-able. As far as appropriate national data arenot available, the default values or ranges inTable 5.19 can be used for orientation, e.g.the average of a range.

If critical loads related to quality criteria offood or fodder are to be calculated, the crit-ical concentrations in the harvestable plantparts should be multiplied with the yields(net crop removal), considering for arableland the coverage by the crops of interest, inorder to calculate the tolerable output ofmetals by biomass harvest.

If contents are available for different harvest-ed parts of the plants (e.g. stem and bark), amass weighted mean should be used.Beware that only the net uptake is calculat-ed. For instance, for agricultural land theamount of metals in stalks or the leaves ofbeets remaining on the field should not beconsidered. The removal of heavy metals in

Table 5.19: Ranges of mean values (averages, medians) of contents of Pb, Cd, and Hg in biomass for various species(harvestable parts)

*) Hg in spruce stems » 10-20% of needle content (Schuetze and Nagel 1998)**) Northern Sweden (Alriksson et al. 2002 and unpublished), for spruce stems without/with bark

Other data sources: De Vries and Bakker (1998), Nagel et al. (2000), Jacobsen et al. (2002)

this case is the product of the yield ofgrains/beets and the mean contents in theseparts of the plants. For forest ecosystems,only the net increment should be consid-ered, but not the uptake into needles, leaves,etc., which also remain in the system.

In ecosystems with appreciable precipitationsurplus or with a very limited growth, theremoval of metals by harvest may often bevery low compared to metal losses by leach-ing at critical load. In these cases the uptakecalculation do not deserve high efforts.Instead, it is better to concentrate on sophis-ticated calculations for the critical leachingrate.

5.5.2.1.3 Critical leaching of heavy metalsfrom the soil

The critical leaching flux of a heavy metalfrom the regarded soil layer can be calculated according to the equation:

(5.90)

where:Mle(crit) = critical leaching flux of heavy

metal from the topsoil(g ha-1 a-1) (see eq. 5.88)

Qle = flux of drainage water leach-ing from the regarded soil layer defined as above (m a-1).

[M]tot.sdw(crit) = critical total concentration of heavy metal in the soildrainage water (mg m-3)(derived from critical limits, see 5.5.2.2)

cle = 10 g mg-1 m2 ha-1, factor for appro-priate conversion of flux units

Flux of drainage water

In order to calculate critical loads in view ofgroundwater protection the data on precipi-tation surplus from the database on criticalloads of acidity and nutrient nitrogen can beused. Deviating from this, the proportion oftranspiration removing water from the upper

horizons (O, and /or Ah, Ap) has to be account-ed for by using a scaling (root uptake) factorwhen critical loads with respect to ecotoxi-cological effects or to food/fodder qualityare addressed.

The drainage water flux leaching from thetopsoil at the bottom of the topsoil (Qle,zb) atsteady state can be calculated according to:

(5.91a)

where:P = precipitation (m a-1)Ei = interception evaporation (m a-1)Es = actual soil evaporation within the

topsoil defined as above (m a-1)Et = actual plant transpiration (m a-1)fEt,zb = scaling or root uptake factor,

fraction of water uptake within the topsoil (–)

This approach is based on the assumptionthat soil evaporation (Es) only takes placedown to the depth zb. Interception evapora-tion can be calculated as a function of theprecipitation (De Vries et al. 1991). For siteswithout detailed water balance data, theannual mean water percolation Qle can alsobe determined by the long-term mean annual temperature (mainly determining thepotential evapotranspiration, Epot) and precipitation (mainly influencing the actualevapotranspiration, Eact) according to:

(5.91b)

where:Pm = annual mean precipitation (m a-1,

data adjusted for common meas-urement bias)

Tm = annual mean air temperature (°C)Em,pot = annual mean potential evapotran-

spiration in humid areas at Tm = 0°C; Em,pot » 0.35 m a-1

in forests, possibly less in other terrestrial ecosystems.





fE,zb = Fraction of total annual mean evapotranspiration above zb (–); fE,zb » 0.8 for the organic top soil layer of forests.

For forested areas, this relationship is supported by data not only on river runoffbut also on soil percolation (e.g. based onMichalzik et al. 2001), which together sug-gest that about 80% or more of the totalevapotranspiration takes place above orwithin the organic top soil layer. Thus, themean water flux from the organic top layer(Q) can easily be estimated from annualmeans of precipitation (P) and air tempera-ture (T), which are two traditional climatenormals available in traditional climate maps(see Background document):

In European forest regions, Qle,zb is typically0.1-0.6 m a-1, but may reach >2 m a-1 in coastalmountain regions. The standard parameteruncertainty is on the order of ±0.1 m a-1 (i.e.about ±30%) at the landscape scale.Depending on climate, Qle can account for10 to 90% of P in temperate-boreal forests,but is usually close to half. In very dryregions the percentage of Qle in P canbecome very low. With eq. 5.91b, Qle almostnever drops below 0.1 m-1 in Europe (consid-ering EMEP-50 km grid square means). Foreq. 5.91a, a suggested minimum value is 5 %of the precipitation. This seems a reason-able lower value since there are always peri-ods during the year with downward percola-tion and a situation of no leaching hardly (ornever) occurs on a yearly basis. The use ofmonthly water balances is not advocated asthe effect of all seasonal variations is notincluded in the critical limits, since theserepresent annual or long-term means, in linewith the critical load approach for acidity.

Critical total dissolved or total concentra-tions of heavy metals in soil drainage water

Information on the derivation of critical totaldissolved concentrations of heavy metals insoil drainage water, [M]dis,sdw(crit), eitherdirectly, through transfer functions (plant -

soil solution) or through [M]free,sdw(crit) isgiven in the next section (5.5.2.2), with back-ground information on used approaches inthe Annexes 1-3. The critical total dissolvedmetal concentrations related to ecotoxico-logical effects in soils require some specificconsiderations. These critical total metalconcentrations in soil solution are deter-mined as the sum of the critical concentra-tion of the free metal ion M2+, [M]free,sdw(crit),and the metals bound to dissolved inorganiccomplexes [M]DIC,sdw such as MOH+, HCO3

+,MCl+, and to dissolved organic matter,[M]DOM,sdw, according to:

(5.92)

where:[M]dis,sdw(crit) = critical total dissolved metal

concentration in soil drainage water (mg m-3)

[M]free,sdw(crit) = critical free metal ion concentration in soil drainage water (mg m-3)

[M]DIC,sdw = concentration of metal bound to dissolved inorganic complexes in soil drainage water (mg m-3)

[M]DOM,sdw = concentration of metal bound to dissolved organic matter in soil drainage water (mg.kg-1)

[DOM]sdw = concentration of dissolved organic matter in soil drainage water (kg m-3)

Geochemical equilibrium partitioning of theheavy metal between the different fractionsis assumed. Further, the water draining fromthe soil also contains metals bound to suspended particulate matter, [M]SPM,sdw,according to:

(5.93)

where:[M]tot,sdw(crit) = critical total metal concentra-

tion in soil drainage water (mg m-3)

[SPM]sdw = concentration of suspended particulate matter in soil drainage water (kg m-3)

In the calculations, we suggest the particulate fraction to be neglected to getcomparable values of critical concentrationsfor the different effects pathways (seeSection 5.5.2.2.3). In this manual, thedescription of methods is adapted to the useof the critical total dissolved metal concen-trations, [M]dis,sdw(crit), beeing equal to totalmetal concentrations in soil solution, implicitly assuming that the concentration ofmetals bound to suspended particulate matter is negligible ([SPM]sdw = 0), i.e.[M]dis,sdw(crit) equals [M]tot,sdw(crit).

5.5.2.2 Critical dissolved metal concentrations derived from critical limits in terrestrial ecosystems

Critical total concentrations of the heavymetals Cd, Pb and Hg in the soil solution,[M]dis,sdw(crit), depend on the target to be protected. These values have to be derivedfrom critical limits (see Table 5.17):

-Critical metal contents in plants (Cd, Pb, Hg) in view of human health or animal health effects through intake of plant products.

-Critical metal concentrations in ground water (Cd, Pb, Hg) in view of human health effects through intake of drinking water.

-Critical concentrations of free metal ions in soil solution (Cd, Pb) in view of ecotoxicological effects on soil micro-organisms, plants and invertebrates.

-Critical metal contents in the soil (Hg) in view of ecotoxicological effects on soil micro-organisms and invertebrates in the forest humus layer.

The critical total dissolved concentration ofa heavy metal in the soil drainage water([M]dis,sdw(crit)) includes both the free metalions and the metals bound to dissolved inorganic and organic complexes (eq. 5.92).

The derivation of the critical total dissolvedconcentrations to be applied in eq. 5.90 isexplained below.

5.5.2.2.1 Critical dissolved concentrations ofCd, Pb and Hg in view of critical plant metalcontents

Starting from the idea to derive critical totalCd, Pb and Hg concentrations in soil solutionrelated to human health effects on the basisof critical limits for plant metal contents(food quality criteria) for food crops onarable land De Vries et al. (2003) provided anoverview on selected soil-plant relationshipsof Cd, Pb and Hg. It shows that only for Cdsignificant relationships (R2 of ³ 0.5) are available.

Cadmium

Starting with a critical Cd content in plantone may derive a critical dissolved metalconcentration by a plant – soil solution relationship. Such a relationship was derivedby applying a regression of Cd contents inwheat in the Netherlands to calculated soilsolution concentrations, that were derivedby using measured total soil contents andsoil properties and application of a transferfunction, relating total concentrations insolution to the soil metal content (Romkenset al. 2004). By applying such a function,regression relationships were derived for Cdin plant (wheat grains) as a function of Cd insoil solution and vice versa as described inTable 5.20. The best estimate of a critical Cdconcentration might be the mean of bothestimates.

The EU regulation (EG) No.466/2001 uses alimit for Cd of 0.2 mg kg-1 fresh weight in wheatgrains. This limit was derived with the principle “As Low As ReasonablyAchievable” (ALARA) and is therefore notbased on effects. There are however manyindications that from the viewpoint of protection of human health, the critical limitof 0.1 mg kg-1 fresh weight, which was used inthe EU before 2001, is more appriate (for these arguments see De Vries et al. 2003,De Vries et al. 2004a,b). Table 5.20 provides





the parameters for the transfer functions aswell as results based on the critical limit of 0.1 mg kg-1 fresh weight (results for the EUlimit of 0.2 mg kg-1 fresh weight is given inbrackets). If the mean of both results oftransfer function application is used, theresulting critical total concentration isapproximately 0.8 mg m-3 (or 4 mg m-3). Themost conservative estimate equals approxi-mately 0.6 mg m-3 (or 1.75 mg m-3).

A more sophisticated and consistent waywould be to

- first derive a critical “pseudo” total soil metal content, by applying soil – plant relationships in the inverse way (derive a critical total soil content from a critical plant content)

- then apply a transfer function relating “pseudo”total metal contents to reactive metal contents (Annex 1, eq. A1.3).

- followed by a transfer function relating the free ion metal activity in solution to the reactive metal content (Annex 1, eq. A1.4 or eq. A1.5).

- followed by a calculation of total concentrations from free metal ion activities with a chemical speciation model (i.e. the W6S-MTC2 model, Section 5.5.2.2.3).

Please note that the current version of W6S-MTC2 is designed to calculate M(sdw)critbased only on the critical limits relating toecotoxicological effects and not to foodquality.

Lead and mercury

For Pb and Hg in food crops, back calcula-tion to soil content is not possible, becausethere are no relationships between contentof soil and contents in plants for those metals. For Pb and Hg, direct uptake from theatmosphere by plants has to be considered.Methods for such calculations, based ondata from De Temmerman and de Witte(2003a,b) are provided in Annex 5 of thebackground document (De Vries et al. 2004b).

5.5.2.2.2 Critical dissolved concentrations ofCd, Pb and Hg aiming at ground water protection

The critical total Cd, Pb and Hg concentrationin soil solution related to human healtheffects can also be based on quality criteria(critical limits) for drinking water (WHO 2004)for all terrestrial ecosystems (see Table 5.17).In line with the decisions of the ExpertMeeting on Critical Limits (2002, in Berlin)the protection of ground water for potentialuse as drinking water resource should alsobe addressed in critical load calculations.The Technical Guidance Document for RiskAssessment (http://ecb.jrc.it) suggests inchapter 3.1.3 that in the first instance theconcentration in soil pore water can be usedas an estimate of the concentration inground water. The WHO guideline includesthe following quality criteria for Cd, Pb andHg in view of drinking water quality:

Pb: 10 mg m-3

Cd: 3 mg m-3

Hg: 1 mg m-3

Table 5.20. Values for the intercept (int) and the parameter a in the regression relationships relating Cd in plant (wheat grains) as a function of Cd in soil solution and vice versa. The table also gives the percentage variationexplained (R2), the standard error of the result (se) and the resulting critical total dissolved Cd concentration whenapplying a critical Cd content in wheat of 0.1 mg kg-1 fresh weight (0.12 mg kg-1 dry weight) and in brackets the valuewhen applying the limit of 0.2 mg kg-1 fresh weight (EG No 466/2001).

1 log(Cd plant) = Int + a*log(Cd soil solution)2 log(Cd soil solution) = Int + a*log(Cd plant)

These values can directly be included as[M]dis,sdw(crit) in the critical load calculation.

5.5.2.2.3 Critical dissolved concentrations ofCd and Pb related to ecotoxicological effects

Critical limits related to the ecotoxicologicaleffects of Cd and Pb are related to impactson soil micro-organisms, plants and inverte-brates for both agricultural land (arable land,grassland) and non-agricultural land (forests,natural non-forested ecosystems; see Table 5.17). The critical concentrations usedin this manual are based on the followingapproach:

-Use of ecotoxicological data (NOEC and LOEC data) for the soil metal content using experiments with information on soil properties (clay and organic matter content and soil pH) as well;

-Calculation of critical free metal ion concentrations (critical limits) in soil solution on the basis of the ecotoxico-logical soil data (NOECs and LOECs) andsoil properties, using transfer functions relating the reactive soil metal content to the free metal ion concentration;

-Calculation of the critical total dissolved metal concentrations [M]dis,sdw(crit) from critical limits for free metal ion concen-trations using a chemical speciation model.

Calculation of critical free metal ion concen-trations from critical soil reactive metal contents

Soil toxicity data collated and acceptedunder the terms of current EU RiskAssessment procedures (Draft RiskAssessment Report Cd (July 2003) seehttp://ecb.jrc.it, Voluntary Risk Assessmentfor Pb), were used. The data covered chronic population-level effects on soilplants, soil-dwelling invertebrates andmicrobial processes. The toxicity endpointswere quoted mainly in terms of an addedmetal dose. In using added doses, theassumption is made that the added metal isentirely in reactive forms over the course ofthe toxicity experiment.

The transfer functions for the calculation offree metal ion concentration from reactivesoil metal content, used in the derivation offree ion critical limit functions, are given inAnnex 1. Soil properties needed in this function are organic matter and soil solutionpH. In the derivation, soil pH values measured by chemical extraction (by H2O,KCl or CaCl2) were used to estimate soilsolution pH by application of regressionsgiven in Annex 10 of the background docu-ment (De Vries et al. 2004b), assuming thatthe pH in soil solution equals pHsdw). EU RiskAssessment procedures do not require theorganic matter content of the soil to bespecified for data to be accepted. However,such data were not usable for the calculationof critical free metal ion concentrations fromcritical soil metal contents, since the usedtransfer functions do require these data (seeAnnex 1) and were thus removed from thedatabases.

The bioavailability of metals does not onlydepend on the free metal ion concentrationbut also on the concentration of othercations, particularly H+. This was taken intoaccount in deriving critical limits as a function of the pH in soil drainage water(pHsdw). The derived critical limit functionswere:

(5.94)

(5.95)

More information on the approach and thetoxicity data is given in Lofts et al. (2004) andin De Vries et al. (2004a). A summary can befound in the background document (De Vrieset al. 2004b).

Calculation of total dissolved metal concen-trations from free metal ion concentrations

To calculate critical loads for soils from thecritical limit functions, it is necessary toknow the total concentration of metal in soil





drainage water that corresponds to the freeion critical limit. In Annex 2, an overview isgiven of the calculation procedure using theWHAM model. Results thus obtained withthis model for an assumed standard CO2pressure of 15 times the atmospheric pres-sure of 0.3 mbar ( 4.5 mbar) are given inTables 5.21 and 5.22. WHAM includes alsothe fraction of suspended particulate matter,which strictly is not part of the soil solution.The total concentration is therefore relatedto soil drainage water. When [SPM]sdw= 0, the value of [M]tot,sdw(crit) equals that of[M]dis,sdw(crit) (see eq. 5.93). For reasons of consistency with the other approaches (seebefore), in which the critical value refers to[M]dis,sdw(crit), it is advocated to apply theresults with [SPM]sdw= 0. Furthermore, thereare high uncertainties in the data on SPM insoil solution. Table 5.21 furthermore showsthat in most cases, the impact of suspended

particulate matter on the total Cd concen-tration in soil drainage water (even at a concentration of 50 mg l-1) is small, but for Pbit can be large (Table 5.22).

Use of pH and DOC values to be consideredin the calculation of critical metal concentra-tions

Some parameters in the critical load calcula-tion depend on the status of the soil, in particular the acidification status (pH) andthe concentration of DOC (see also thetables 5.21 and 5.22). In the following recommendations are provided, which status of soil conditions should be considered, when Mdis,sdw(crit) is derived fromcritical limits for free metal ion concentrations, as presented in the tables5.21 and 5.22.

Table 5.21: Look-up table to derive values of the total critical Cd concentrations in soil drainage water [Cd]tot,sdw(crit)at a CO2 pressure that equals 15 times the CO2 pressure of the air

pH values: In principle the pH at steady stateconditions assuming Gothenborg Protocolimplementation, can best be taken as abasis. This may cause problems, as it has tobe determined using dynamic models.Instead the pH at the critical acid load can beused. This pH is easier to calculate but it maystrongly deviate from the pH at steady stateassuming Gothenburg Protocol implementa-tion. Furthermore, the calculation of the critical load pH is rather uncertain dependingon arbitrary choices to be made. Thereforethe use of the critical load pH is not

recommended.

Assuming that it is likely that present pH is(almost) equal to future pH at steady state(under Göteborg Protocol implementationconditions), the present pH is advised to usefor pragmatic reasons. Because the presentpH in soil solution is not always available, butrather measured as pH in water or in saltextracts, regression functions to relate several pH measurements to soil solution pHwere derived. Relations are given in Table5.22, assuming no effect of soil type on the



Table 5.22: Look-up table to derive values of the total critical Pb concentrations in soil drainage water [Pb]tot,sdw(crit)at a CO2 pressure that equals 15 times the CO2 pressure of the air

Table 5.23: Results of linear regression analyses of the pH in soil solution against pH-H2O, pH-CaCl2 and pH-KCl

1) All coefficients are significant at p > 0,999



relationship. These relations can be used tocalculate the soil solution pH which is need-ed in the critical load calculations and also inthe transfer functions relating reactive metalcontents to free metal ion concentrations.

More detailed information is given in Annex10 in the background document (De Vries et al. 2004b). This includes relationships as afunction of soil type. Ranges in the presentand steady-state critical soil pH for variouscombinations of land use, soil type and soildepth are also provided there.

DOC concentrations: The concentration ofdissolved organic matter (DOM) in soils isnowadays frequently determined in climate-related studies. Concentrations of DOM areusually determined by analysis of carbon(DOC) which accounts for half of the weightof soil organic matter (DOM = 2 × DOC).However, long-term data on soil solutionsare rarely available at sufficient density formapping region-specific means and variability’s, and may need to be estimatedfrom studies elsewhere. Ranges in DOCvalues for major forest types and soil layers,by means of the 5-, 50- and 95 percentiles,are presented in Annex 11 of the backgrounddocument (De Vries et al. 2004b) on the basisof DOC values from approximately 120Intensive Monitoring plots in Europe. In general, the results show a clear decrease inDOC concentrations going from the humuslayer (median value of 40 mg l-1) into the mineral subsoil. Furthermore, the values areslightly higher in coniferous forest comparedto deciduous forests.

Relationships of DOC concentrations withvegetation type, hydrology, growth conditions or soil properties may be expected, which would be useful to improve estimates for different sites and regions. Thedata for the mineral soil (De Vries et al.2004b) were thus used to derive relation-ships with available site characteristics andsoil data that may affect the DOC concentra-tions, including the type of forest, (coniferous or deciduous forests), textureclass (indication for soil type), temperature,pH and the contents of C and N, including theC/N ratio. Results thus obtained are given in

the background document. The results showa good relationship with the site and soilcharacteristics in the subsoil (below 30cm)but the relationships were much worse in thetopsoil (above 30cm). In the topsoil therewas a clear positive relationship with C/Nratio and temperature, while the correlatedvalues of the individual C and N concentra-tions were negatively and positively relatedto DOC, respectively. The relationships are,however, too weak to be very useful. This isin line with the limited number of studies inthe literature, from which no significant relationship could be discerned (Michalzik et al. 2001).

Based on the available data the followingdefault values for calculating critical loads ofPb and Cd, or critical levels of atmosphericHg pollution, respectively, are suggested(see background document, Annex 11):

Forest organic layer (O horizon):[DOC]sdw = 35 mg l-1 ([DOM]sdw = 70 mg l-1).

Forest mineral topsoil (0-10 cm):[DOC]sdw = 20 mg l-1 ([DOM]sdw = 40 mg l-1)).

Grass land (0-10) cm:[DOC]sdw = 15 mg l-1 ([DOM]sdw = 30 mg l-1).

Arable land (0-30) cm:[DOC]sdw = 10 mg l-1 ([DOM]sdw = 20 mg l-1).

5.5.2.2.4 Critical dissolved concentrations ofHg related to ecotoxicological effects in soils

Critical limit for the soil: With respect to Hg,critical limits refer only to effects on soilmicro-organisms and invertebrates in thehumus layer of forests. The suggested critical limit for Hg is that the concentrationin the humus layer (O-horizon) of forest soilsafter normalization with respect to theorganic matter content should not exceed0.5 mg kg (org)-1 (Meili et al. 2003a). Becauseof the strong association of Hg with organicmatter leaving virtually no free ions, theexposure of biota to Hg is controlled by the

competition between biotic and other organic ligands, and the contamination of alltypes of organic matter is determined by thesupply of organic matter relative to the supply of Hg at a given site (Meili 1991a,1997, cf. biodilution). Therefore, the criticallimit for Hg in soils is set for the organicallybound Hg rather than for the free ion concen-tration, also in solution.

Critical total mercury concentrations in soilsolution can be calculated by using a transfer function for Hg from soil to soil solution, while assuming a similar criticalHg/org ratio in the solid phase and in the liquid phase, at least in oxic environmentswhere binding to sulphides is negligible.Various reasons supporting this are given inMeili (1991a, 1997, 2003b), De Vries et al. (2003), and Åkerblom et al. (2004).

Transfer function for mercury: The criticalleaching of Hg from the humus layer (Mle(crit)in eq. 5.88) is related to the mobility and Hgcontent of dissolved organic matter becauseof the strong affinity of Hg for living and deadorganic matter and the resulting lack of competition by inorganic ligands in this layer(e.g. Meili 1991, 1997). Because of the strongassociation of Hg with organic matter leaving virtually no free ions (apparently farless than one per km2 of topsoil, based onSkyllberg et al. 2003), the biogeochemicalturnover of Hg is controlled by the competi-tion between biotic and other organic ligands. Therefore, Hg/OM ratios are a usefultool for calculating critical limits and loadsand associated transfer functions (Meili et al. 2003a). This is the basis of the transfer function to derive total Hg concentrations inpercolating (top)soil water ([M]dis,sdw(crit) in eq. 5.90, mg m-3) as follows:

(5.96)

where [Hg]dis,sdw(crit) = critical total Hg concentra-

tion in soil drainage water (mg m-3)

[Hg]OM(crit) = critical limit for Hg concen-tration in solid organicmatter (OM), or the Hg/OMratio in organic (top)soils([Hg]OM(crit)=0.5 mg (kg OM)-1).

ff = fractionation ratio, describing the Hg contamination of organic matter in solution (DOM) relative to that insolids (OM) (–),

[DOM]sdw = concentration of dissolved organic matter in soil drainage water (g m-3),

csdw = 10-3 kg g-1, factor for appropri-ate conversion of mass units.

The scale-invariant fractionation or transferfactor ff describes the Hg partitioningbetween organic matter in solids and organic matter in solution and is defined asthe ratio between the Hg content of DOM andthat of OM (Meili et al. 2003a, Meili et al.2003b). Preliminary studies in Sweden sug-gest that the Hg concentration in DOM is ofsimilar magnitude as that in OM, and that 1may be used as a default value for ff untildeviations from unity prove to be significant(Åkerblom et al. 2004).

Critical concentration for the soil drainagewater: Based on the Hg limit of 0.5 mg kg-1 OMand a DOM concentration of 70 mg l-1

(DOC = 35 mg l-1), the critical steady stateconcentration of total Hg in soil drainagewater is 35 ng l-1 or 0.035 ug l-1 (see eq. 5.96).This concentration is consistent with thatderived by a different approach at the water-shed scale (Meili et al. 2003a) and is similarto high-end values presently observed in soilsolutions and surface freshwaters (Meili,1997; Meili et al. 2003b; Åkerblom et al.2004). Note that this ecosystem limit for soilwater is much lower than the drinking waterlimit above, but still higher than that for surface freshwaters where Hg limits for fishconsumption usually are exceeded at surface water concentrations of 1-5 ng l-1.





5.5 3. Aquatic ecosystems

5.5.3.1 Critical loads of cadmium and lead

5.5.3.1.1 Simple steady-state mass balancemodel and related input data

In principle, the simple steady-state massbalance approach can be used for Cd, Pband Hg but it has been decided to restrict theapproach in first instance to Cd and Pb anduse a different, precipitation based approachfor Hg, as described in Section 5.5.3.2.

Steady-state mass balance model in streamwaters

As with terrestrial ecosystems, the criticalload of Cd and Pb for freshwaters is theacceptable total load of anthropogenicheavy metal inputs corresponding to thesum of tolerable outputs from the catchmentby harvest and outflow, minus the naturalinputs by weathering release in the catch-ment but adding the retention in the surfacewater (De Vries et al. 1998). There is no needto consider net release in catchment soils, ifthe net weathering (weathering minus occlusion) is negligible. Since the estimationof net release in soils includes high uncertainties, it is preliminarily assumed tobe negligible.

In the initial manual on the calculation of critical loads of heavy metals for aquaticecosystems (De Vries et al. 1998), the defaultmethod presented to calculate critical loadsof heavy metals for soils included in-lakemetal retention, including all relevant metalfluxes, namely sedimentation, resuspensionand exchange processes in the lake (infiltra-tion, diffusion and bioirrigation), whileassuming a steady state situation (DeVries et al.1998). To keep the approach as simpleas possible, and also to stay as close aspossible to the simple mass balanceapproach for nitrogen and acidity, this modelcan be simplified by neglecting weatheringin the catchment and lumping transientexchange processes at the sediment-waterinterface and the net effect of sedimentationand resuspension in one retention term

according to De Vries et al. (1998):

(5.97)

where:Mu = removal of heavy metal by bio-

mass harvesting or net uptake in the catchment (g ha-1a-1)

Mret(crit) = net retention of heavy metal inthe lake at critical load(g ha-1a-1)

Mlo(crit) = critical lateral outflow of heavy metal from the whole catch-ment (g ha-1a-1)

Al = lake area (ha)Ac = catchment area (ha)

When critical loads of Cd and Pb for streamwaters are calculated, there is no need toconsider net retention, leading to the follow-ing critical load calculation:

(5.98)

Because the estimation of net retention forlakes includes high uncertainties, it is rec-ommendable to calculate preliminarilyaquatic critical loads for stream waters only,for which the retention in surface water isnegligible. It furthermore leads to the lowestcritical loads and thus implies the protectionof lakes as well. Finally, when calculatingcritical loads for lakes, one may also assumethat net retention of metals in lakes is negligible, implying the assumption that theoverall release or retention of metals in acatchment, including the lake sediment, isnegligible.

Heavy metal removal by net uptake

The assessment of these data is comparablefor those in terrestrial ecosystems (see eq. 5.89), but now the uptake or releaserefers to the complete catchment. Thisimplies that no further reduction factorsneed to be applied to relate the uptake in the

root zone/catchment to the mineral topsoil.The equation for net uptake is thus equal to eq. 5.89 with fMu being equal to 1.

Critical output of heavy metals from theaquatic system

The critical lateral outflow can be describedas the product of the lateral outflow flux ofwater and the critical limit for the total concentration of the heavy metal in the sur-face water according to:

(5.99)

where:Qlo = lateral outflow flux of water

from the whole catchment area (m a-1)

[M]tot,sw(crit) = critical limit for the total concentration (dissolved and in suspended particles) of heavy metal in surface water (mg m-3)

Qlo, which sometimes is denoted as thehydraulic load in the literature can bederived for a lake on the basis of the flowfrom the aquatic system, Q (m3a-1) divided bythe catchment area (m2). The total concen-tration of metals can be calculated as:

(5.100)

where:[M]dis,sw(crit) = critical dissolved concentra-

tion of a heavy metal insurface water (mg m-3)

[M]SPM,sw(crit) = critical total content of a heavy metal in suspended particles (mg kg-1)

[SPM]sw = concentration of suspended particulate matter in surface water (kg m-3)

Data on the lateral outflow of lakes can bederived from the S&N critical loads

database. The critical load depends on thecritical limit used. In the initial manual foraquatic ecosystems (De Vries et al. 1998), itwas argued that critical limits referring to thefree metal ion activity in surface water aremost appropriate. This idea has been furtherdeveloped by Lofts et al. (unpublished data),but has not been adopted here, for reasonswhich will be given in 5.5.3.1.2. Instead, critical limits referring to total dissolvedmetal concentrations have been adopted. Itis necessary to include a solid-solutiontransfer function (see Annex 1) to calculatethe critical metal concentration in suspended particles and hence the criticaltotal aqueous metal concentration.

Information on how to estimate the criticalnet in-lake retention when calculating criticalmetal loads for lakes is given in the background document to this manual (De Vries et al. 2004b). Like for terrestrialecosystems it is recommendable to calculate weathering rates (here at least for adepth of 1 m) to account for the influence ofnatural processes in comparison to atmospheric deposition in order to evaluatecritical loads and critical limits exceedances.Information on how to calculate weatheringwithin the catchment is given in Annex 6 ofthe background document.

5.5.3.1.2 Critical total dissolved cadmiumand lead concentrations in aquatic ecosystems

Critical limits for total dissolved concen-trations

Analysis of aquatic ecotoxicological data byLofts et al. (unpublished) suggested overlapbetween aquatic and terrestrial toxic endpoint concentrations at a given pH.Hence it was suggested that common critical limits be applied for both soils andfreshwaters, by using the critical limit functions derived in 5.5.2.2 for toxic effectson the soil ecosystem. However, althoughthere is no theoretical reason why the sensi-tivities of soil and water organisms to metalsshould not be similar (assuming that uptakeof the free ion from the aqueous phase is the





significant mechanism leading to toxicity)this approach has not been adopted for thefollowing reasons:

1. The aquatic toxicity data for Cdcovered a more restricted pH range than for the terrestrial toxicity data (pH6.9 to 8.7 compared to pH 3.2 to 7.9). Therefore, although overlap of points was seen within the pH range covered by the aquatic toxicity data, no data were available to validate the theory of overlap below pH 6.9.

2. Observed overlapping of points for Pbwas less than for any of the metals studied (Cu and Zn in addition to Cd and Pb). Most of the aquatic toxicity data gave free Pb endpoints higher than those observed for soils.

For these reasons, it was decided not to usethe free ion approach for aquatic critical limits and instead to express the critical limits as the total dissolved metal (mg m-3). A summary of preliminary effect-based critical limits is given in Table 5.24. The values for Cd are based on the EU RiskAssessment Report for Cd (Risk assessmentCadmium metal CAS-No. 7440-43-9) Thevalues for Pb are based on Crommentuijn et al. (1997) for the value suggested for usein the 2004 call for data, and on a substancedata sheet on Pb and its compounds (2003)for the value to be used when updatedAnnex 3 is available. The reason of needingan update of Annex 3 is described below.The suggested substitude for Annex 3 is provided in Annex 12 of the background

document (DeVries et al. 2004b) includingdetailed calculation examples. The valuesare all related to ecotoxicological effects.There are also critical limits related to secondary poisoning, but these values arenot yet recommended for use because theydo require further substantiation and discussion.

The value of 0.38 mg m-3, taken from EU RiskAssessment Report for Cd, is based on the5-percentile cut-off value of chronic toxicitydata from 168 reliable tests on singlespecies and 9 multi-species studies. Anassessment factor of 2 is further introducedin the report, leading to a critical limit of 0.19mg m-3 , but this approach was not acceptedin this manual. For Cd, a relationship withwater hardness has also been found. In theEU Risk Assessment Report. Recently, it wasalso accepted to take the influence of hardness on the toxicity of cadmium intoaccount, using 3 hardness classes (withhardness H in mg CaCO3 l-1) according to 0.16 mg m-3 if H <100, 0.30 mg m-3 if 100<H <200 and 0.50 mg m-3 if H >200, usingno assessment factor (see also the back-ground document to the manual).

For Pb, the critical limit of 11 mg m-3 is basedon Crommentuijn et al. (1997), whereas thevalue of 5 mg m-3 (range of 2.1-9.3 mg m-3) isbased on the 5-percentile cut-off value ofchronic toxicity data, calculated with themethod of Aldenberg & Jaworska, using 3data sets of selected (i) freshwater and saltwater NOECs/EC10s (30 values), (ii)

Table 5.24: Recommended critical limits for dissolved Cd and Pb concentrations surface waters

1 A comparable critical limit is suggested in the RAR on Cd for the protection of top predators, namely 0.26 mg m-3.This value is based on a critical limit for the intake of Cd of 160 µg Cd /kg food (wet weight) of the predator, beingthe quality standard for biota tissue with respect to secondary poisoning. However, this value is yet considered

too uncertain to be used in the critical load calculations2 H = hardness in mg CaCO3 l-1

freshwater NOECs/EC10s (19 values) and (iii)saltwater NOECs/EC10s (11 values). In thesubstance data sheet on Pb, an assessmentfactor of 3 is further introduced, but thisapproach was not accepted in this manual.At a workshop of ICP Waters on heavy metals, 2002, in Lillehammer (Skjelkvale andUlstein, 2002) a range of 1 - 11 mg m-3 wassuggested in dependence on water chemistry, with low values referring to clearsoftwaters. The critical limit of 5 mg m-3 is inthe middle of this range and thus consistent.A much lower critical limit is suggested insubstance data sheet on Pb for the protec-tion of human health using a critical limit of200 µg Pb kg-1 muscle meat of fish (food standard set by Commission Regulation (EC)No. 466/2001) and the protection of preda-tors in freshwater and saltwater environ-ments from secondary poisoning (near0.4 µg Pb l-1). However, this value is yet considered to uncertain to be used in thecritical load calculations.

Although not presently used, a preliminarycritical limit for Hg can be found in the sub-stance data sheet on Hg and its compounds(2003). As with Pb , this value is based on the5-percentile cut-off value of chronic toxicitydata, using 3 data sets of selected (i) fresh-water and saltwater, (ii) freshwater and (iii)saltwater, leading to a value of 0.142 mg m-3

(90 percentile range of 0.056 - 0.281 mg m-3). Inthe substance data sheet on Hg, an assess-ment factor of 4 is further introduced, butthis approach was not accepted in this manual. A reliable quality standard to protect top predators from secondary poisoning can not be given, but the value ismuch lower than those for ecotoxicologicaleffects. The value of 0.035 mg m-3 presentedearlier for soils is likely to be an upper limitfor secondary poisoning.

Calculation of critical limits for total aqueousconcentrations

In order to calculate critical loads of metalsfor freshwater ecosystems it is necessary toknow the total aqueous concentration at thecritical limit, i.e. the concentration of dissolved metal and of metal bound to

suspended particulate matter (SPM). Thereare various possible approaches to deriveadsorbed metal contents on suspended particles ([M]SPMsw) from total dissolvedmetal concentrations in surface water([M]tot,sw). The simplest approach is a empirical linear approach (Kd-value) relatingboth contents and concentrations, whileaccounting for the impact of major properties of the suspended particles influencing the sorption relationship.However, Kd values for a given metal mayvary substantially from place to place and sothe Kd approach is not appropriate when calculating metal contents on suspendedparticles from a large number of differentlocations.

An alternative approach, which uses as faras possible data and models used elsewherein this manual, is to take a two-stageapproach:

1. Calculate the critical free ion concen-tration from the critical dissolved metalconcentration.

2. Calculate the critical particle-bound metal from the critical free ion.

3. Sum the critical particle-bound and dissolved metal to obtain the critical total metal.

Step 1 uses a complexation model (e.gWHAM) to calculate the critical free ion concentration from the critical dissolvedmetal concentration. Step 2 uses a transferfunction to calculate the particle-boundmetal from the free ion. This transfer functionis given in Annex 2. The calculation of thecritical total aqueous concentration is presented in Annex 3.

In Annex 3, the procedure given applies onlyto the values of 0.38 mg m-3 for Cd and 11 mg m-3 for Pb. Use of different values (for Cdas a function of hardness and for Pb 5instead of 11 implies a rerun of the WHAMmodel. This will be done and these values can be used, as soon as the updatedAnnex 3 is adopted.





Surface water chemistry data

Data needed to calculate the total dissolvedmetal concentration are the concentration ofsuspended particles in the water compart-ment, [SPM]sw, the pH and DOC concentra-tions of surface water. The concentration ofSPM in the surface water (kg m-3 or g l-1)depends on the turbulence of the water,which in turn depends on the geological setting (incl. land use) and water flow velocity (i.e. wind speed for lakes). The concentration of suspended particles maythus vary considerably and generally rangesfrom 1 to 100 g m-3. The average concentra-tion for Dutch surface waters, for example, is30 g m-3, and for a dataset of lowland UKrivers (n = 2490) it is 30.6 g m-3 with a range of <0.1 to 890 g m-3, while Scandianavianwaters typically show much lower values.

pH and DOC values for lakes largely dependon the landscape surrounding the lakesincluding the parent material (its sensitivityto acid inputs). Typical DOC values for clearwater lakes are below 5 mg l-1, whereas forhumic lakes, values can be higher than 50 mg l-1

. Values for the pH generally varybetween 5 and 7. Both pH and DOC are standard measurements in lake surveys anda wealth of data can be derived from thosesurveys.

When calculating in-lake retention in deriving critical loads for lakes, data oncharacteristics such as the lake and catchment area and the net retention rateare needed. For more information we refer tothe background document (De Vries et al. 2004b) and an earlier manual (De Vrieset al. 1998).

5.5.3.2 Critical levels of mercury in precipitation

Critical loads of atmospheric pollution foraquatic ecosystems (lakes and rivers) maybe approached by a mass balance approachinvolving a wide variety of processes bothwithin the water column and in the surround-ing watershed. Alternatively, the steady state

partitioning of pollutants in a constant environment can be formulated without anyneed for mass balance considerations ordetailed understanding of ecosystemprocesses. This can be achieved by linkingcritical receptors such as fish directly to themain immissions through transfer functions(TF) describing the relationship of their Hgconcentrations at steady state, as describedbelow.

5.5.3.2.1 Derivation of critical levels of mercury in precipitation referring to a standard fish

Basic concept

Hg concentrations in fish show a wide varia-tion, about 30-fold both within and amongsites (Meili 1997). A standardized value for agiven site (lake or river) can be obtained byreferring to a commonly caught piscivorousfish with a total body weight of 1 kg, in particular pike (Esox lucius). Using a 1-kgpike as a standard receptor, the mean Hgconcentration in fish flesh can be related tothe mean Hg concentration in precipitationat a given site as follows:

(5.101)

where:[Hg]Pike = Hg concentration in the flesh of

1-kg pike (mg kg-1 fw)[Hg]Prec = Hg concentration in precipita-

tion (ng l-1)TFHgSite = site-specific transfer function

(l kg-1 fw) referring to the trans-fer of atmospheric Hg to fishflesh in a watershed at steady state

cbp = 10-6 mg ng-1, factor for appropri-ate conversion of units.

The critical level of atmospheric pollution([Hg]Prec(crit)) can thus be calculated as follows:

(5.102)

where:[Hg]Pike(crit) = critical Hg concentration in the

flesh of 1-kg pike (0.3 mg kg-1 fw)

[Hg]Prec(crit) = critical Hg concentration in precipitation (ng l-1)

cbp = 10-6 mg ng-1, factor for appro-priate conversion of flux units.

Regarding the critical limit for mercury inpike of 0.3 mg kg-1 fw, we refer to the background document of the manual (DeVries et al. 2004b).

The transfer function TFHgSite

TFHgSite addresses the wide variation of Hgconcentrations among ecosystems inresponse to a given atmospheric Hg input atsteady state. It accounts for a variety ofcomplex processes including both terrestrialand aquatic aspects related to the biogeo-chemistry of Hg in lakes and rivers (Meili et al. 2003a), thus accounting for both fluxesand transformations of Hg (e.g. sorption,volatilization, net methylation, bioavailability,biodilution, biomagnification). For mappingof watershed sensitivity, TFHgSite is preferablyexpressed as a function of basic physical-chemical parameters. Hg concentrations infish are generally highest in nutrient-poorsoftwaters in acidic watersheds rich in wetlands (e.g. Verta et al. 1986, Håkanson et al. 1988, Meili 1991a, 1994, 1996a, 1997).Such differences can be described by empirical relationships to address regionaland local differences in watershed biogeo-chemistry, based on variables for which dataare commonly available (e.g. from otherstudies under CLRTAP), such as surface

water pH or concentrations of organic carbon or nutrients (the latter being of particular relevance for mercury). Two alternative formulations capturing part of thelarge variation in TFHgSite are:

(5.103a)

(5.103b)

where:[TOC]sw = concentration of total organic

carbon in surface water (mg l-1)[TP]sw = concentration of total phos-

phorus in surface water (mg l-1)pHsw = pH in surface water TFHgRun = transfer function (l kg-1 fw) refer-

ring to the transfer of atmo-spheric Hg to fish flesh via runoff in a reference watershed at steady state.

The first formulation (16a) is most appropri-ate and should be used when concentra-tions of total organic carbon and total phos-phorus in surface water are available, whichis often the case from routine monitoring ofsurface waters. The alternative formulationbased on pH alone (16b) is less adequate butcan be used if data access is limited.

TFHgRun can be quantified from adequatedata sets in various ways (see Annex 13 ofthe background document, De Vries et al. 2004b). If such data are not available, avalue of 250 000 l kg-1 fw can be used forTFHgRun referring to the standard fish (1 kg, inparticular pike, Esox lucius) at steady state(Meili et al. 2003a, cf. Verta et al. 1986, Meili1991a). An important aspect to considerwhen quantifying TFHgRun (or other steadystate parameters) from field data is thatpresent environmental Hg concentrations arenot in steady state with the present level ofatmospheric pollution.





5.5.3.2.2 Derivation of critical levels of mercury in precipitation referring to otherorganisms

Basic concept

The Hg concentration in any fish or otherorganism, serving as food for humans andfish-based wildlife such as birds and mammals, can be related to the Hg concen-tration in 1-kg pike according to:

(5.104)

where:[Hg]Bio = Hg concentration in any biota, e.g.

fish flesh (mg kg-1 fw)TFHgBio = organism-specific transfer func-

tion addressing the typical Hgpartitioning within food webs (-)

The critical level of atmospheric pollution([Hg]Prec(crit)) can thus be calculated from acombination of eq. 5.102 and eq. 5.104 asfollows:

(5.105)

where:[Hg]Bio(crit) = critical Hg concentration in

any biota, e.g. fish flesh (mg kg-1 fw)

cbp = see above

TFHgBio is useful for two purposes: (1) to estimate values for 1-kg pike forsites/regions in which only mercury concen-trations in other organisms are available, (2)to convert critical load maps referring to 1-kg pike into maps for other target organisms of local/regional interest.

The transfer function TFHgBio

TFHgBio addresses the wide variation of Hgconcentrations among organisms within

food webs, by describing the typical deviation from the standard fish. Amongcommonly available variables, body weightis the most powerful single predictor of fishHg levels, also across species. The variationin TFHgBio can be described as follows:

(5.106)

where:fHgY = value for very young fish and other

small animals (–); fHgY »0.13fHgW = species-specific slope coefficient (–);

fHgW » 0.2...2 (Table 5.25)W = total body fresh weight (kg fw)

For many freshwater fish used for humanconsumption, this will generate estimates ofmean Hg concentrations at a given fish sizethat differ less than 2-fold from observedmeans. Species-specific slope coefficients(fHgW) for some common freshwater fish aregiven in Table 5.25 for the typical case thatthe value for very young fish and other small animals (fHgY) can be maintained at 0.13. Forany fish species (e.g. for unexplored sites orfor unknown future fish populations), a firstapproximation differing less than 3-fold fromobserved size-class means can be madebased on body weight alone, using theparameter for the standard fish, pike (fHgW = 0.87, Table 5.25). If fish weight dataare not available, total body weight (W in kg)can be estimated from total body length byapplying a species-specific shape factor(fLW, Table 5.25) according to:

(5.107)

where:L = length of the fish (cm)

Table 5.25 is meant as a reference that canbe expanded and adapted for local use,based on additional field data from systemswhere several coexisting species have been analyzed. Note that for compatibility oftransfer functions and for inter-regional comparisons, the value of TFHgBio refers to a1-kg pike, which should be maintained as areference receptor with a value of TFHgBio = 1.

5.5.4. Limitations in the present approach and possible future refinements

In general the uncertainties in measurementas well as in modelling are higher withrespect to trace elements than for mainnutrient elements. In particular the followinguncertainties of the models should be men-tioned:

-The steady-state of metal inputs and outputs on the level of the critical limit is a theoretical situation. In dependence of the actual status of a site (or area) it may take years to centuries (e.g. for calcareous soils) to reach this steady-state. This should be considered, when critical loads and their exceedances are to be interpreted. To consider the processes of metal accumulation or loss from soils over time, dynamic approaches would be needed. Although such models are already suggested, they are not yet considered here, because they still need further sophisti-cation. There is some inconsistencybetween the calculation of the critical leaching and the tolerable removal of the

metals with biomass, because types of critical limits and their mode of use are different for both fluxes.

-The uptake of heavy metals by plants is not constant over time but varies strongly with changes in pollution and is at present likely lower than indicated above at steady state at the level of critical concentrations,

-Possible effects of thinning of the metal concentration due to high mass fluxes of biomass harvest (high yields) are not considered due to missing knowledge,

-The delivery of heavy metals to the available pools of soils and surface waters is excluded from the mass balance equation due to high uncertain-ties of the available calculationapproach. However since the same approach is used to identify sites with high natural inputs it may happen that one site is excluded, while another site with an insignificant lower weathering rate will stay in the database,

-The approaches taken to calculate critical limits for ecotoxicological effects are different for terrestrial and aquatic ecosystems. Given the likelihood that terrestrial and freshwater organisms (with the exception of surface-dwelling soil invertebrates such as snails) are exposed to metal in a similar manner (i.e. via the solution phase), a common approach to deriving critical limits, if not common values or functions for the limits, is scientifically desirable,



Table 5.25: Coefficients for size conversion (fLW) and normalization of Hg concentrations (fHgW) in freshwater fish,some standard fish weights (W) for consumption and the related value for TFHgBio



-The critical limit derivation includes several uncertainties, as e.g. differences between results from laboratory or field, which are (deviating e.g. from OECD methodologies) not taken into account by the use of ”uncertainty factors”,

-Organisms can be affected by different pathways, this could only partly considered here,

-The vertical flux of metals bound to particulate matter suspended in the drainage water, may be remarkable in certain soils, this holds in particular for Pb. It was, however, not recommended to consider this, in order to be consistent with other parts of the manual,

-The seasonal variation of soil para-meters such as pH, DOC cannot be accounted for in the models.

Achermann B, Bobbink R (eds) (2003)Empirical Critical Loads for Nitrogen.Environmental Documentation No. 164,Swiss Agency for the Environment,Forests and Landscape (SAEFL), BerneSwitzerland, 327 pp.

Aherne J, Kelly-Quinn M, Farrell EP (2002) Asurvey of lakes in the Republic of Ireland:Hydrochemical characteristics and acidsensitivity. Ambio 31: 452-459.

Aherne J, Curtis CJ (2003) Critical loads ofacidity for Irish lakes. Aquatic Sciences65: 21-35.

Aherne J, Posch M, Dillon PJ, Henriksen A(2004) Critical loads of acidity for surfacewaters in south-central Ontario, Canada:Regional application of the First-orderAcidity Balance (FAB) model. Water, Airand Soil Pollution: Focus 4: 25-36.

Åkerblom S, Meili M, Bringmark L,Johansson K (2004). Determination of thefractionation factor (ff) in forest soildescribing the Hg content of organic matter in solution relative to that in solidsbased on field data from Sweden.Background Document on Critical Loadsof Heavy Metals, UN/ECE-CLRTAP-ICPModelling and Mapping, 6 p.( h t t p : / / w w w . i c p m a p p i n g . c o m ,http://www.oekodata.com/pub/mapping/workshops/ ws_potsdam/Akerblom.pdf).

Alriksson A, Eriksson H, Karltun E, Lind T,Olsson M (2002) Carbon pools andsequestration in soil and trees in Sweden,based on data from national soil and forest inventories. In: M Olsson (ed): LandUse Strategies for Reducing NetGreenhouse Gas Emissions (LUSTRA),Progress Report 1999–2002, pp. 30-36.Swedish University of AgriculturalSciences, Uppsala, Sweden

Baker LA, Brezonik PL (1988) Dynamicmodel of in-lake alkalinity generation.Water Resources Research 24: 65-74.

Batterbee RW, Allot TEH, Juggins S, KreiserAM (1995) Estimating the base critical

load: The diatom model. In: CLAG (1995)op. cit., pp.3-6.

Berendse F, Beltman B, Bobbink R, KwantM, Schmitz MB (1987) Primary productionand nutrient availability in wet heathlandecosystems. Acta Oec./Oecol. Plant. 8:265-276.

Bindler R, Renberg I, Klaminder J, EmterydO (2004). Tree rings as Pb pollutionarchives? A comparison of 206Pb/207Pbisotope ratios in pine and other environ-mental media. Sci. Total Environ. 319:173-183.

Bobbink R, Boxman D, Fremstad E, Heil G,Houdijk A, Roelofs J (1992) Critical loadsfor nitrogen eutrophication of terrestrialand wetland ecosystems based uponchanges in vegetation and fauna. In:Grennfelt and Thörnelöf (1992), op. cit.,pp. 111-159.

Bobbink R, Hornung M, Roelofs JGM (1996)Empirical nitrogen critical loads for natu-ral and semi-natural ecosystems. In: UBA(1996) op. cit., Annex III (54 pp).

Bobbink R, Hornung M, Roelofs JGM (1998)The effects of air-borne pollutants onspecies diversity in natural and semi-nat-ural European vegetation. Journal ofEcology 86: 717-738.

Bobbink R, Ashmore M, Braun S, FlückigerW, Van den Wyngaert IJJ (2003) Empiricalnitrogen critical loads for natural andsemi-natural ecosystems: 2002 update.In: Achermann and Bobbink (2003), op.cit., pp 43-170.

Boggero A, Barbieri A, De Jong J, MarchettoA, Mosello R (1998) Chemistry and criticalloads of Alpine lakes in Canton Ticino(southern central Alps). Aquatic Science60: 300-315.

Bolt GH, Bruggenwert MGM (eds) (1976) SoilChemistry. Part A. Basic Elements.Elsevier, Amsterdam, The Netherlands,281 pp.

Bouten W, De Vre FM, Verstraten JM,Duysings JJHM (1984) Carbon dioxide inthe soil atmosphere: simulation modelparameter estimation from field measure-ments. In: E Eriksson (ed): Hydrochemical



References



Balances of Freshwater Systems. IAHS-AISH 150, Uppsala, Sweden, pp. 23-30.

Brakke DF, Henriksen A, Norton SA (1989)Estimated background concentrations ofsulfate in dilute lakes. Water ResourcesBulletin 25(2): 247253.

Brakke DF, Henriksen A, Norton SA (1990) Avariable F-factor to explain changes inbase cation concentrations as a functionof strong acid deposition. Verh. Internat.Verein. Limnol. 24: 146-149.

Breeuwsma A, Chardon JP, Kragt JF, DeVries W (1991) Pedotransfer functions fordenitrification. Final Report of the project‘Nitrate in Soils’, DG XII, EuropeanCommunity, Brussels, pp. 207-215.

Brook GA, Folkoff ME, Box EO (1983) Aworld model of carbon dioxide. EarthSurface Processes and Landforms 8: 79-88.

Burman R, Pochop LO (1994) Evaporation,Evapotranspiration and Climatic Data.Developments in Atmospheric Science22, Elsevier, Amsterdam, 278 pp.

CLAG (Critical Loads Advisory Group) (1995)Critical loads of acid deposition for UnitedKingdom freshwaters. ITEEdinburgh/Department of theEnvironment, London, United Kingdom,80 pp.

Crommentuijn T, Polder MD, Van dePlassche EJ (1997) Maximum permissibleconcentrations and negligible concentra-tions for metals, taking background con-centrations into account. Report 601501001, National Institute for Public Healthand the Environment, Bilthoven, TheNetherlands, 157 pp.

Cronan CS, Walker WJ, Bloom PR (1986)Predicting aqueous aluminium concentra-tions in natural waters. Nature 324: 140-143.

Davies CE, Moss D (2002) EUNIS habitatclassification, Final Report. CEH MonksWood, United Kingdom.

De Temmerman L, de Witte T (2003a).Biologisch onderzoek van de verontreini-ging van het milieu door zware metalen teHoboken groeiseizoen 2001. Internal

report Centrum voor onderzoek in dierge-neeskunde en agrochemie C.O.D.A,Tervuren.

De Temmerman L, de Witte T (2003b)Biologisch onderzoek van de verontreini-ging van het milieu door kwik en chloridente Tessenderlo en van kwik teBerendrecht. Internal report Centrum vooronderzoek in diergeneeskunde en agrochemie C.O.D.A, Tervuren. EG No.466/2001: COMMISSION REGULATION(EC) No 466/2001 of 8 March 2001 settingmaximum levels for certain contaminants in foodstuffs, OfficialJournal of the European Commission No.L 77;

De Vries W (1988) Critical deposition levelsfor nitrogen and sulphur on Dutch forestecosystems. Water, Air and Soil Pollution42: 221-239.

De Vries W, Hol A, Tjalma S, Voogd JC (1990)Literature study on the amounts and resi-dence times of elements in forest ecosys-tems (in Dutch). Rapport 94, DLO WinandStaring Centre, Wageningen, TheNetherlands, 205 pp.

De Vries W (1991) Methodologies for theassessment and mapping of critical loadsand of the impact of abatement strategieson forest soils. Report 46, DLO WinandStaring Centre, Wageningen, TheNetherlands, 109 pp.

De Vries W, Posch M, Reinds GJ, Kämäri J(1993) Critical loads and their exceedanceon forest soils in Europe. Report 58(revised version), DLO Winand StaringCentre, Wageningen, The Netherlands,116 pp.

De Vries W (1994) Soil response to acid dep-osition at different regional scales. PhDthesis, Agricultural UniversityWageningen, Wageningen, TheNetherlands, 487 pp.

De Vries W, Reinds GJ, Posch M (1994)Assessment of critical loads and theirexceedances on European forests using aone-layer steady-state model. Water, Airand Soil Pollution 72: 357-394.

De Vries W, Bakker DJ (1998) Manual for calculating critical loads of heavy metals

for terrestrial ecosystems. Guidelines forcritical limits, calculation methods andinput data. DLO Winand Staring Centre,Report 166, Wageningen, TheNetherlands, 144 pp.

De Vries W, Bakker DJ, Sverdrup HU (1998)Manual for calculating critical loads ofheavy metals for aquatic ecosystems.Guidelines for critical limits, calculationmethods and input data. DLO WinandStaring Centre, Report 165, Wageningen,The Netherlands, 91 pp.

De Vries W, Posch M (2003) Derivation ofcation exchange constants for sandloess, clay and peat soils on the basis offield measurements in the Netherlands.Alterra-Rapport 701, Alterra Green WorldResearch, Wageningen, The Netherlands,50 pp.

De Vries W, Reinds GJ, Posch M, Sanz MJ,Krause GHM, Calatayud V, Renaud JP,Dupouey JL, Sterba H, Vel EM, DobbertinM, Gundersen P, Voogd JCH (2003)Intensive Monitoring of ForestEcosystems in Europe. Technical Report2003, EC-UNECE, Brussels and Geneva,161 pp.

De Vries W, Schütze G, Lofts S, Meili M,Römkens PFAM, Farret R, DeTemmerman L, Jakubowski M (2003)Critical limits for cadmium, lead and mercury related to ecotoxicologicaleffects on soil organisms, aquatic organisms, plants, animals and humans.In: Schütze et al. (2003) op. cit., pp. 29 -78

De Vries W, Lofts S, Tipping E., Meili M.,Groenenberg JE, Schütze G (2004a).Critical limits for cadmium, lead, copper,zinc and mercury related to ecotoxico-logical effects on terrestrial and aquaticorganisms. Water, Air and Soil Pollution(in prep)

De Vries W, Schütze G, Lofts S, Tipping E,Meili M, Groenenberg JE, Römkens PFAM(2004b). Calculation of critical loads forcadmium, lead and mercury. Backgrounddocument to Mapping Manual Chapter5.5, ..pp, www.icpmapping.org

Dillon PJ, Molot LA (1990) The role of ammo-nium and nitrate retention in the acidifica-tion of lakes and forested catchments.Biogeochemistry 11: 23-43.

Driscoll CT, Lehtinen MD, Sullivan TJ (1994)Modeling the acid-base chemistry oforganic solutes in Adirondack, New York,lakes. Water Resources Research 30:297-306.

Duan L, Hao J, Xie S, Du K (2000) Criticalloads of acidity for surface waters inChina. Science of the Total Environment246: 1-10.

Dutch J, Ineson P (1990) Denitrification of anupland forest site. Forestry 63: 363-377.

Eurosoil (1999) Metadata: Soil GeographicalData Base of Europe v.3.2.8.0. JointResearch Centre, Ispra, Italy.

FAO (1981) FAO-Unesco Soil Map of theWorld, 1:5.000.000; Volume V Europe,Unesco-Paris, 199 pp.

Farret R (2003) General Methodology. In:Schütze et al. (2003) op. cit., pp. 103-107.

Grennfelt P, Thörnelöf E (eds) (1992) CriticalLoads for Nitrogen. Nord 92:41. NordicCouncil of Ministers, Copenhagen, 428pp.

Groenenberg JE, Römkens PFAM, Tipping E,Pampura T, De Vries W (2004) Transferfunctions for the solid solution partition-ing of Cd, Cu, Pb and Zn – Review, synthe-sis of datasets, derivation and validation.(in prep.)

Gunn J, Trudgill ST (1982) Carbon dioxideproduction and concentrations in the soilatmosphere: a case study from NewZealand volcanic ash soils. Catena 9: 81-94.

Håkanson L, Nilsson Å, Andersson T (1988)Mercury in fish in Swedish lakes.Environmental Pollution 49: 145-162.

Hall J, Bull K, Bradley I, Curtis C, Freer-SmithP, Hornung M, Howard D, Langan S,Loveland P, Reynolds B, Ullyett J, Warr T(1998) Status of UK critical loads andexceedances. Part 1: Critical loads andcritical load maps. Report prepared underDETR/NERC Contract EPG1/3/116.http://critloads.ceh.ac.uk





Hall J, Ullyett J, Heywood E, Broughton R,Fawehinmi J & UK experts (2003) Statusof UK critical loads: critical loads meth-ods, data and maps. Report preparedunder Defra/NERC contract EPG 1/3/185.http://critloads.ceh.ac.uk

Hall J, Ashmore M, Curtis C, Doherty C,Langan S, Skeffington R (2001) UN/ECEexpert workshop: Chemical criteria andcritical limits. In: Posch et al. (2001) op.cit., pp. 67-71.

Hall J, Davies C, Moss D (2003)Harmonisation of ecosystem definitionsusing the EUNIS habitat classification. In:Achermann and Bobbink (2003) op .cit.,pp 171-195.

Henriksen A (1984) Changes in base cationconcentrations due to freshwater acidifi-cation. Verh. Internat. Verein. Limnol. 22:692-698.

Henriksen A, Lien L, Traaen TS, Sevaldrud IS,Brakke DF (1988) Lake acidification inNorway - Present and predicted chemicalstatus. Ambio 17: 259266.

Henriksen A, Lien L, Rosseland BO, TraaenTS, Sevaldrud IS (1989) Lake acidificationin Norway - Present and predicted fishstatus. Ambio 18: 314321.

Henriksen A, Kämäri J, Posch M, Wilander A(1992) Critical loads of acidity: Nordic sur-face waters. Ambio 21: 356-363.

Henriksen A, Forsius M, Kämäri J, Posch M,Wilander A (1993) Exceedance of criticalloads for lakes in Finland, Norway andSweden: Reduction requirements fornitrogen and sulfur deposition. Acid RainResearch Report 32/1993, NorwegianInstitute for Water Research, Oslo,Norway, 46 pp.

Henriksen A, Posch M, Hultberg H, Lien L(1995) Critical loads of acidity for surfacewaters – Can the ANClimit be considered

variable? Water, Air and Soil Pollution 85:2419-2424.

Henriksen A, Posch M (2001) Steady-statemodels for calculating critical loads ofacidity for surface waters. Water, Air andSoil Pollution: Focus 1: 375-398.

Henriksen A, Dillon PJ (2001) Critical load ofacidity in south-central Ontario, Canada:I. Application of the Steady-State WaterChemistry (SSWC) model. Acid RainResearch Report 52/01, NorwegianInstitute for Water Research, Oslo,Norway, 31 pp.

Henriksen A, Dillon PJ, Aherne J (2002)Critical loads of acidity for surface watersin south-central Ontario, Canada:Regional applications of the Steady-StateWater Chemistry (SSWC) model.Canadian Journal of Fisheries andAquatic Sciences 59: 1287-1295.

Hettelingh J-P, Slootweg J, Posch M,Dutchak S, Ilyin I (2002) Preliminary modelling and mapping of critical loads ofcadmium and lead in Europe. Report259101011, CCE & EMEP MSC-East,RIVM, Bilthoven, The Netherlands, 127pp.

Hindar A, Posch M, Henriksen A, Gunn J,Snucins E (2000) Development and appli-cation of the FAB model to calculate criti-cal loads of S and N for lakes in theKillarney Provincial Park (Ontario,Canada). Report SNO 4202-2000,Norwegian Institute for Water Research,Oslo, Norway, 40 pp.

Hindar A, Posch M, Henriksen A (2001)Effects of in-lake retention of nitrogen oncritical load calculations. Water, Air andSoil Pollution 130: 1403-1408.

Hornung M, Sutton MA, Wilson RB (eds)(1995) Mapping and Modelling of CriticalLoads for Nitrogen: A Workshop Report.Proceedings of the Grange-over-SandsWorkshop 24-26 October 1994. Institutefor Terrestrial Ecology, United Kingdom,207 pp.

Hornung M, Bull KR, Cresser M, Hall J,Langan SJ, Loveland P, Smith C (1995) Anempirical map of critical loads of acidityfor soils in Great Britain. EnvironmentalPollution 90: 301-310.

Ilyin I, Ryaboshapko A, Afinogenova O, BergT, Hjellbrekke AG (2001) Evaluation oftransboundary transport of heavy metalsin1999. Trend analysis. EMEP Report3/2001, EMEP MeteorologicalSynthesizing Centre - East, Moscow.

Jacobsen C, Rademacher P, Meesenburg H,Meiwes KJ (2002) Gehalte chemischerElemente in Baumkompartimenten,Niedersächsische Forstliche Versuchs-anstalt Göttingen, im Auftrag desBundesministeriums für Verbraucher-schutz, Ernährung und Landwirtschaft(BMVEL), Bonn, 80 pp.

Johansson M, Tarvainen T (1997) Estimationof weathering rates for critical load calcu-lations in Finland. Environmental Geology29(3/4): 158-164.

Johnson DW, Cole DW, Gessel SP (1979)Acid precipitation and soil sulfate adsorp-tion properties in a tropical and in a tem-perate forest soil. Biotropica 11: 38-42.

Johnson DW, Henderson GS, Huff DD,Lindberg SE, Richter DD, Shriner DS,Todd DE, Turner J (1982) Cycling of organ-ic and inorganic sulphur in a chestnut oakforest. Oecologia 54: 141-148.

Johnson DW (1984) Sulfur cycling in forests.Biogeochemistry 1: 29-43.

Joki-Heiskala P, Johansson M, Holmberg M,Mattson T, Forsius M, Kortelainen P, HallinL (2003) Long-term base cation balancesof forest mineral soils in Finland. Water,Air and Soil Pollution 150: 255-273.

Kaste Ø, Dillon PJ (2003) Inorganic nitrogenretention in acid-sensitive lakes in south-ern Norway and southern Ontario,Canada – a comparison of mass balancedata with and empirical N retentionmodel. Hydrological Processes 17: 2393-2407.

Kelly CA, Rudd JWM, Hesslein RH, SchindlerDW, Dillon PJ, Driscoll CT, Gherini SA,Hecky RE (1987) Prediction of biologicalacid neutralization in acid-sensitive lakes.Biogeochemistry 3: 129-140.

Kimmins JP, Binkley D, Chatarpaul L, DeCatanzaro J (1985) Biogeochemistry oftemperate forest ecosystems: Literatureon inventories and dynamics of biomassand nutrients. Information Report PI-X-47E/F, Petawawa National ForestryInstitute, Canada, 227 pp.

Larssen T, Høgåsen T (2003) Critical loadsand critical load exceedances in Norway

(in Norwegian with English Appendix).NIVA Report 4722-2003, NorwegianInstitute for Water Research, Oslo,Norway, 23 pp.

Lien L, Raddum GG, Fjellheim A, Henriksen A(1996) A critical limit for acid neutralizingcapacity in Norwegian surface waters,based on new analyses of fish and inver-tebrate responses. The Science of theTotal Environment 177: 173193.

Lofts S, Spurgeon DJ, Svendsen C, TippingE (2003) Soil critical limits for Cu, Zn, Cdand Pb: a new free ion-based approach,Env. Science and Technology, accepted).

Lydersen E, Larssen T, Fjeld E (2004) Theinfluence of TOC on the relationshipbetween Acid Neutralizing Capacity (ANC)and fish status in Norwegian lakes. TheScience of the Total Environment (inpress).

Meili M (1991a) The coupling of mercury andorganic matter in the biogeochemicalcycle – towards a mechanistic model forthe boreal forest zone. Water, Air and SoilPollution 56: 333-347.

Meili M (1991b) Mercury in boreal forest lakeecosystems. Acta UniversitatisUpsaliensis 336, Chapter 8.

Meili M (1994) Aqueous and biotic mercuryconcentrations in boreal lakes: modelpredictions and observations. In: CJWatras and JW Huckabee (eds) MercuryPollution: Integration and Synthesis. CRCPress, Lewis Publishers Inc., Boca RatonFL, Chapter I.8, pp. 99-106.

Meili M, Malm O, Guimarães JRD, ForsbergBR, Padovani CR (1996a) Mercury concentrations in tropical (Amazon) andboreal freshwater fish: natural spatial vari-ability and pollution effects. 4th

International Conference on Mercury as aGlobal Pollutant, Hamburg, Germany, 4–8Aug. 1996. Book of Abstracts, GKSS, p.403.

Meili M (1997) Mercury in lakes and rivers.Metal Ions in Biological Systems 34: 21-51.

Meili M, Bishop K, Bringmark L, JohanssonK, Munthe J, Sverdrup H, De Vries W





(2003a) Critical levels of atmospheric pol-lution: criteria and concepts for opera-tional modelling of mercury in forest andlake ecosystems. The Science of the TotalEnvironment 304: 83-106.

Meili M, Åkerblom S, Bringmark L,Johansson K, Munthe J (2003b) Criticalloads and limits of heavy metals inecosystems: Some Swedish contributionsto European modelling efforts,Background document presented at theEditorial Meeting of the Expert Panel onCritical Loads of Heavy Metals underUNECE-CLRTAP-ICP Modelling andMapping, Paris, 9–10 April 2003, 22 pp.http://www.icpmapping.com /work-shops/ws_berlin/sweden.pdf.

Michalzik B, Kalbitz K, Park JH, Solinger S,Matzner E (2001) Fluxes and concentra-tions of dissolved organic carbon andnitrogen – a synthesis for temperateforests. Biogeochemistry 52: 173-205.

Moiseenko T (1994). Acidification and criticalload in surface waters: Kola, NorthernRussia. Ambio 23: 418-424.

Monteith JL, Unsworth M (1990) Principles of

Environmental Physics (2nd edition).Arnold, London, 291 pp.

Mulder J, Stein A (1994) The solubility of alu-minum in acidic forest soils: long-termchanges due to acid deposition.Geochimica et Cosmochimica Acta 58:85-94.

Nagel H-D, Gregor H-D (eds) (1999) Ökolo-gische Belastungsgrenzen – CriticalLoads & Levels (in German). Springer,Berlin, 259 pp.

Nagel H-D, Becker R, Eitner H, Kunze F,Schlutow A, Schütze G (2000) Kartierungvon Critical Loads für den Eintrag vonSäure und eutrophierenden Stickstoff inWaldökosysteme und naturnahe waldfreieÖkosysteme zur Unterstützung vonUNECE-Protokollen, Abschlussberichtzum Forschungsprojekt 297 73 011 imAuftrag des Umweltbundesamtes Berlin.

NEG/ECP (2001) Protocol for Assessmentand Mapping of Forest Sensitivity toAtmospheric S and N Deposition, pre-

pared by the NEG/ECP Forest MappingGroup, New England Governors/EasternCanadian Premiers, ‘Acid Rain ActionPlan 2001, Action Item 4: Forest MappingResearch Project’, 79 pp.

Nilsson J, Grennfelt P (eds) (1988) CriticalLoads for Sulphur and Nitrogen.Environmental Report 1988:15 (Nord1988:97), Nordic Council of Ministers,Copenhagen, 418 pp.

Oliver BG, Thurman EM, Malcolm RL (1983)The contribution of humic substances tothe acidity of colored natural waters.Geochimica et Cosmochimica Acta 47:2031-2035.

Olsson M, Rosén K, Melekrud P-A (1993)Regional modelling of base cation lossesfrom Swedish forest soils due to whole-tree harvesting. Applied Geochemistry,Suppl. Issue No.2, pp. 189-194.

Paces T (1983) Rate constants of dissolutionderived from the measurements of massbalance in hydrological catchments.Geochimica et Cosmochimica Acta 47:1855-1863.

Posch M, Forsius M, Kämäri J (1993) Criticalloads of sulfur and nitrogen for lakes I:Model description and estimation ofuncertainty. Water, Air and Soil Pollution66: 173-192.

Posch M, Hettelingh J-P, Sverdrup HU, BullK, De Vries W (1993) Guidelines for thecomputation and mapping of criticalloads and exceedances of sulphur andnitrogen in Europe. In: RJ Downing, J-PHettelingh, PAM de Smet (eds)Calculation and Mapping of Critical Loadsin Europe. CCE Status Report 1993, RIVMReport 259101003, Bilthoven, TheNetherlands, pp. 25-38. See alsowww.rivm.nl/cce

Posch M, Kämäri J, Forsius M, Henriksen A,Wilander A (1997) Exceedance of criticalloads for lakes in Finland, Norway andSweden: Reduction requirements foracidifying nitrogen and sulfur deposition.Environmental Management 21(2): 291-304.

Posch M (2000) Critical loads of acidity:Possible modifications. In: M Holmberg

(ed) Critical Loads Calculations:Developments and TentativeApplications. TemaNord 2000:566, NordicCouncil of Ministers, Copenhagen, pp.8-19.

Posch M, De Smet PAM, Hettelingh J-P,Downing RJ (eds) (2001) Modelling andMapping of Critical Thresholds in Europe.CCE Status Report 2001, RIVM Report259101010, Bilthoven, Netherlands,iv+188 pp. See also www.rivm.nl/cce

Posch M, Hettelingh J-P, Slootweg J (eds)(2003a) Manual for dynamic modelling ofsoil response to atmospheric deposition.RIVM Report 259101012, Bilthoven, TheNetherlands, 69 pp. See alsowww.rivm.nl/cce

Posch M, Reinds GJ, Slootweg J (2003b) TheEuropean background database. In: MPosch, J-P Hettelingh, J Slootweg, RJDowning (eds) Modelling and Mapping ofCritical Thresholds in Europe. CCE StatusReport 2003, RIVM Report 259101013,Bilthoven, The Netherlands, pp. 37-44.See also www.rivm.nl/cce

Reinds GJ, Posch M, De Vries W (2001) Asemi-empirical dynamic soil acidificationmodel for use in spatially explicit integrat-ed assessment models for Europe. AlterraReport 084, Alterra Green WorldResearch, Wageningen, The Netherlands,55 pp.

Reuss JO (1983) Implications of the calcium-aluminum exchange system for the effectof acid precipitation on soils. Journal ofEnvironmental Quality 12(4): 591-595.

Reuss JO, Johnson DW (1986) AcidDeposition and the Acidification of Soilsand Waters. Ecological Studies 59,Springer, New York, 119 pp.

Römkens PFAM, Groenenberg JE, Bril J, DeVries W (2001) Derivation of partition relationships to calculate Cd, Cu, Pb, Niand Zn solubility and activity in soil solu-tions: an overview of experimental results.Summary of an Alterra Report.

Rosén K (1990) The critical load of nitrogento Swedish forest ecosystems.Department of Forest Soils, SwedishUniversity of Agricultural Sciences,

Uppsala, Sweden, 15 pp.

Rosén K, Gundersen P, Tegnhammar L,Johansson M, Frogner T (1992) Nitrogenenrichment in Nordic forest ecosystems –The concept of critical loads. Ambio 21:364-368.

Ryaboshapko A, Ilyin I, Gusev A,Afinogenova O, Berg T, Hjellbrekke AG1999: Monitoring and modelling of lead,cadmium and mercury transboundarytransport in the atmosphere of Europe.EMEP Report 3/99, MeteorologicalSynthesizing Centre - East, Moscow.

Santore RC, Driscoll CT, Aloi M (1995) Amodel of soil organic matter and its func-tion in temperate forest soil development.In: WW McFee, JM Kelly (eds) CarbonForms and Functions in Forest Soils. SoilScience Society of America, Madison,Wisconsin, pp.275-298.

Sauvé S, McBride M, Hendershot W et al.(1998). Soil solution speciation of lead(II):effects of organic matter and pH Soil Sci.Soc. Am. J. , 62, 618 - 621

Sauvé S, Norvell W, McBride, Hendershop W(2000). Speciation and complexation ofcadmium in extracted soil solutionsEnviron. Sci. Technol., 34, 291 – 296

Schütze G, Nagel H-D (1998) Kriterien für dieErarbeitung von Immissionsminderungs-zielen zum Schutz der Böden undAbschätzung der langfristigen räumlichenAuswirkungen anthropogener Stoffein-träge auf die Bodenfunktionen, Abschluß-bericht UBA-FKZ 104 02 825, in UBA-Texte 19/98, Umweltbundesamt, Berlin.

Schütze G, Lorenz U, Spranger T (2003)Expert meeting on critical limits for heavymetals and methods for their application,2–4 December 2002 in Berlin,Proceedings, UBA Texte 47/2003,Umweltbundesamt, Berlin.

SEPA (2000) Environmental quality criteria:lakes and running waters. SwedishEnvironmental Protection Agency, Report5050, 102 pp.

Skjelkvåle BL, Ulstein M (2002) Proceedingsfrom the workshop on heavy metals (Pb,Cd, Hg) in surface waters: Monitoring and





biological impact. 18–20 March 2002,Lillehammer, Norway, ICP Waters Report67/2002, Norwegian Institute for WaterResearch, NIVA, Oslo.

Skyllberg U, Qian J, Frech W, Xia K, BleamWF, (2003). Distribution of mercury, methylmercury and organic sulphur species insoil, soil solution and stream of a borealforest catchment. Biogeochemistry 64:53–76.

Sogn TA, Stuanes AO, Abrahamsen G (1999)The capacity of forest soil to absorbanthropogenic N. Ambio 28: 346-349.

Steenvorden J (1984) Influence of changes inwater management on water quality (inDutch). Report 1554, Institute for Landand Water Management, Wageningen,The Netherlands.

Sverdrup H, Warfvinge P (1988) Weatheringof primary silicate minerals in the naturalsoil environment in relation to a chemicalweathering model. Water, Air and SoilPollution 38: 387-408.

Sverdrup HU (1990) The Kinetics of BaseCation Release due to ChemicalWeathering. Lund University Press, Lund,Sweden, 246 pp.

Sverdrup H, De Vries W, Henriksen A (1990)Mapping Critical Loads. EnvironmentalReport 1990:14 (NORD 1990:98), NordicCouncil of Ministers, Copenhagen, 124pp.

Sverdrup H, Ineson P (1993) Kinetics of den-itrification in forest soils. Compuscript, 18pp.

Sverdrup H, Warfvinge P (1993) The effect ofsoil acidification on the growth of trees,grass and herbs as expressed by the(Ca+Mg+K)/Al ratio. Reports in Ecologyand Environmental Engineering 2, LundUniversity, Lund, Sweden, 177 pp.

Sverdrup H, De Vries W (1994) Calculatingcritical loads for acidity with the simplemass balance method. Water, Air and SoilPollution 72: 143-162.

Technical Guidance Document on RiskAssessment in support of CommissionRegulation (EC) No 1488/94 on RiskAssessment for existing substances, TGD

European Chemicals Bureau, Institute forHealth and Consumer Protection,European Commission Joint ResearchCentre

Tipping E (1994) WHAM – A chemical equi-librium model and computer code forwaters, sediments, and soils incorporat-ing a discrete site/electrostatic model ofion-binding by humic substances.Computers & Geosciences 20(6): 973-1023.

Tipping E (1998) Humic ion-binding Model IV:an improved description of the interac-tions of protons and metal ions withhumic substances. Aquatic Geochemistry4: 3-48.

Tipping E, Lofts S, Smith EJ, Shotbolt L,Ashmore MR, Spurgeon D, Svendsen C(2003a) Information and proposedmethodology for determining criticalloads of cadmium and lead; a UK contri-bution. Backround document presentedat the Editorial Meeting of the ExpertPanel on Critical Loads of Heavy Metalsunder ICP Modelling and Mapping, Paris,9–10 April 2003.

Tipping E, Rieuwerts J, Pan G, Ashmore MR,Lofts S, Hill MTR, Farrago ME, Thornton I(2003b) The solid-solution partitioning ofheavy metals (Cu, Zn, Cd, Pb) in uplandsoils of England and Wales.Environmental Pollution 125: 213-225.

UBA (1996) Manual on Methodologies andCriteria for Mapping Critical Levels/Loadsand Geographical Areas where they areexceeded. UNECE Convention on Long-range Transboundary Air Pollution,Federal Environmental Agency, Berlin.

UNECE (1995) Calculation of critical loads ofnitrogen as a nutrient. Summary report onthe development of a library of defaultvalues. Document EB.AIR/WG.1/R.108,United Nations Economic Commission forEurope, Geneva, 7 pp.

UNECE (2001) Workshop on chemical crite-ria and critical limits. DocumentEB.AIR/WG.1/2001/13, United NationsEconomic Commission for Europe,Geneva, 8 pp.

Ulrich B, Sumner ME (eds) (1991) Soil Acidity.Springer, Berlin, 224 pp.

Utermann J, Duewel O, Gaebler H-E, HindelR (2000) Beziehung zwischenTotalgehalten und königswasser-extrahierbaren Gehalten vonSchwermetallen in Böden. In: RosenkranzD, Einsele G, Bachmann G, Harreß M(Eds): Handbuch Bodenschutz,Loseblattsammlung, Erich Schmidt-Verlag, Berlin, Kennzahl 1600.

Van Dam D (1990) Atmospheric depositionand nutrient cycling in chalk grassland.PhD Thesis, University of Utrecht,Utrecht, The Netherlands, 119 pp.

Van der Salm C, Köhlenberg L, De Vries W(1998) Assessment of weathering rates inDutch loess and river-clay soils at pH 3.5,using laboratory experiments. Geoderma85: 41-62.

Van der Salm C, De Vries W (2001) A reviewof the calculation procedure for criticalacid loads for terrestrial ecosystems. TheScience of the Total Environment 271: 11-25.

Verta M, Rekolainen S, Mannio J, Surma-AhoK (1986) The origin and level of mercury inFinnish forest lakes. Finnish NationalBoard of Waters, Helsinki, Publications ofthe Water Research Institute 65: 21-31.

Walther B (1998) Development of a process-based model to derive methane emis-sions from natural wetlands for climatestudies. Dissertation im FachbereichGeowissenschaften der UniversitätHamburg, Examensarbeit Nr. 60, Max-Planck-Institut für Meteorologie,Hamburg.

Warfvinge P, Sverdrup H (1992) Calculatingcritical loads of acid deposition with PRO-FILE - A steady-state soil chemistrymodel. Water, Air and Soil Pollution 63:119-143. See also www2.chemeng.lth.se

Warfvinge P, Sverdrup H (1995) Critical loadsof acidity to Swedish forest soils. Reportsin Ecology and EnvironmentalEngineering 5, Lund University, Lund,Sweden, 104 pp.

Weng LP, Temminghoff EJM, Lofts S, Tipping

E, Van Riemsdijk WH (2002) Environ.Complexation with dissolved organicmatter and solubility control of metals in asandy soil. Sci. Technol. 36, 4804-4810

WHO (2000) Air Quality Guidelines forEurope, Second Edition. WHO RegionalPublications, European Series, No. 91.World Health Organisation, RegionalOffice for Europe, Copenhagen, 273 pp.

WHO (2004): Guidelines for Drinking WaterQuality - Third Edition, Vol. 1 –Recommendations, World HealthOrganisation, Geneva.

Wilander A (1994) Estimation of backgroundsulphate concentrations in natural sur-face waters in Sweden. Water Air and SoilPollution 75: 371-387.

Witkamp M (1966) Decomposition of leaf lit-ter in relation to environment, microflora,and microbial respiration. Ecology 47(2):194-201.

Wright RF, Lie MC (eds) (2002) Workshop onmodels for biological recovery from acidi-fication in a changing climate, 9-11September 2002 in Grimstad, Norway.Acid Rain Research Report 55/02,Norwegian Institute for Water Research(NIVA), Oslo, Norway, 42 pp.

Závodský D, Babiaková G, Mitosinková M,Pukanèíková K, Roneák P, Bodis D,Rapant S, Mindás J, Skvarenina J,Cambel B, Rehák S, Wathne BM,Henriksen A, Sverdrup H, Tørseth K,Semb A, Aamlid D (1995) Mapping criticallevel/loads for the Slovak Republic. AcidRain Research Report 37/1995.Norwegian Institute for Water Research(NIVA), Oslo, Norway, 74 pp.



Need of transfer functions in deriving criticaldissolved metal concentrations

In principle, transfer functions are not needed in performing a critical load calculation. Transfer functions have beenused to derive critical limits for free metal ionconcentrations from NOEC data, referring toreactive soil metal contents. When applyingcritical limits for free metal ion concentra-tions, related to ecotoxicological effects, notransfer function is needed any more, since[M](sdw)crit can be obtained directly, either byreference to the look up tables or by use ofthe W6S-MTC2 program (see Section5.5.2.2.3). In case of ground water protec-tion, total dissolved critical concentrationscan be used directly (see Section 5.5.2.2.2).In the case of using critical limits referring tothe metal content in plants, an empiricalrelationship can be used to derive total dissolved critical concentrations in soil solution, at least for Cd (See Table 4).

Using the more sophisticated and consistentway to derive soil solution concentrationsfrom critical plant contents does howeverrequire transfer functions according to thefollowing:

- first derive a critical “pseudo” total soil metal content, by applying soil–plant relationships in the inverse way (derive a critical total soil content from a critical plant content)

- then apply a transfer function relating pseudo- total metal contents to reactive metal contents (Annex 1, eq. A1.3).

- followed by a transfer function relating the free ion metal activity in solution to the reactive metal content (Annex 1, eq. A1.4 or eq. A1.5).

Furthermore, all the transfer functions listed

below are needed for the calculation of acritical soil limit (from a given critical limitfunction for the soil solution) and to comparethis to the present soil metal content toassess the critical limit exceedance in thepresent situation. This requires a map of thepresent soil metal content in the country.Inversely, one may calculate the present dissolved metal concentration from the present soil metal content, using the transfer functions described below and compare thisto the critical limit function for the soil solution (see section 5.5.1.4).

Transfer functions to calculate pseudo-totalfrom total contents of Cd and Pb

In some countries true total metal concen-trations are measured, whereas most ornearly all countries use “pseudo-total” concentrations. Utermann et al. (2000) provided transfer functions to calculatepseudo-total contents of heavy metals (hereaqua regia extract [M]AR) from total contents (here [M]HF), according to:

(A1.1)

where:[M]HF = total content of heavy metal M in

soil, provided as HF-extraction (mg kg-1)

[M]AR = pseudo-total content of heavy metal M in soil provided as Aqua Regia extraction (mg kg-1)

Values for a0 and a1 are given in Tables A1.1and A1.2. The correlations are depending onmetal and substrate. In general, total andpseudo-total contents are very similar. Forback-calculations of total contents frompseudo-total contents, different functionsare to be used (see background document,De Vries et al 2004b, Annex 7). These functions are not provided here, since thosecalculations are not needed in the presentcalculation of critical loads.


Mapping Manual 2004 • Chapter V Mapping Critical Loads - Annex 1-33 Page V - A1

Annex 1: Transfer functions for lead and cadmium for the conversion of metal concentrations in different soil phases


Mapping Manual 2004 • Chapter V Mapping Critical Loads - Annex 1-33Page V - A2

Transfer functions to calculate reactive contents from pseudo-total contents of Cdand Pb

The reactive metal concentration [M]re (mol kg-1) can be related to the pseudo-total concentration extracted withAqua Regia [M]AR (mol kg-1) according to:

(A1.3)

Regression relations were derived from aDutch dataset containing 630 soil sampleswhich were both extracted with 0.43 Mol l-1

HNO3 and Aqua Regia. The dataset consistsof large variety of soil types with a wide vari-ety in soil properties such as the organicmatter and clay content. The dataset com-prises both polluted and unpolluted soils.Results are shown in Table A1.3 and suggestthat reactive contents typically are morethan half of pseudo-total contents.

Table A1.1: Relationship between cadmium (Cd) content in soils extractable by aqua regia (AR) and total contentsin dependence on the parent material.

Table A1.2: Relationship between lead (Pb) content in soils extractable by aqua regia (AR) and total contentsextractable by HF in dependence on the parent material.

Transfer functions to calculate free Cd andPb ion concentrations from reactive Cd andPb contents used in the derivation of criticallimits for free Cd and Pb ion concentrations

Critical concentrations of soil metal are frequently higher than ambient soil concen-trations. Therefore, a transfer functionshould if possible be calibrated over a rangeof soil metal concentrations which is thewhole range of critical receptor concen-trations observed. This is relevant since thederived critical limit functions are dependentupon the transfer functions.

For calibration of direct transfer functions forCd and Pb, data were drawn from foursources:

-Sauvé et al. (1998). Soil metal and labile Pb in Pb-contaminated soils of various origins. Free Pb concentrations were estimated by measurement of labile Pbusing differential pulse anodic stripping voltammetry (DPASV) and speciation calculations.

-Sauvé et al. (2000). Soil metal and labile Cd in Cd-contaminated soils of various origins. Free Cd concentrations were estimated by measurement of labile Cdusing differential pulse anodic stripping voltammetry (DPASV) and speciation calculations.

-Weng et al. (2002). Soil metal and free ion concentrations in sandy Dutch soils.

Free Cd and Pb concentrations were estimated by the Donnan membrane technique.

-Tipping et al. (2003a). Soil metal and free ion concentrations in UK upland soils. Free Cd and Pb were estimated by using the WHAM6 speciation model (Tipping, 1998) to speciate the soil solution.

The data were fitted to the following transferfunction (termed as c-Q relationship:

(A1.4)

where:[M]free,sdw = the free metal ion concentration

(mol l-1)[M]re = the reactive metal content in the

solid phase (mol l-1)[OM]s = organic matter (%)pHsdw = soil drainage water pH

Calculated values of the parameters aregiven in Table A1.4.



Table A1.2: Relationship between lead (Pb) content in soils extractable by aqua regia (AR) and total contentsextractable by HF in dependence on the parent material.

1) The standard error of the y-estimate on a logarithmic basis

Table A1.4: Values for the regression coefficients for the free ion concentration - reactive metal content relation-ship (eq. A1.4) and statistical measures R2 and se(Y) based on results of studies carried out in Canada, theNetherlands and the UK. Values in brackets are the standard errors for the coefficients.



Transfer functions to calculate reactive Cdand Pb contents from free Cd and Pb ion concentrations used in the derivation of critical Cd and Pb contents on suspendedparticles in aquatic ecosystems

This transfer function (termed as Q-c relationship) has been derived using thesame soil data set used to calculate thetransfer function relating the free ion to thesoil reactive metal (See Table A1.4). Theexpression for the Q-c relation is:

(A1.5)

where:[M]free,sdw = the free metal ion concentration

in surface water (mol l-1)[M]re = the reactive metal content in the

solid phase (mol l-1)[OM]s = organic matter (%), here the

organic matter content of the suspended particles

pHsw = the pH of the surface water

Calculated values of the parameters aregiven in Table A1.5.

Use of transfer functions in the manual

The direct transfer function for the calculation of the free ion concentration from the soil reactive metal content (the c-Q relation) is used for the calculation of the pH-dependent critical limit functions (seeSection 5.5.2.2.3), in order to express theendpoint metal dose in toxicity experimentsas the free ion concentration. The transferfunction for the calculation of the soil reactive metal content from the free metalion concentration (the Q-c relation) is usedto calculate the critical SPM-bound metal([M]SPM (crit)) in surface waters (see Section5.5.2.2.3 and Annex 2).

Table A1.5: Values for the regression coefficients for the reactive metal content - free ion concentration relationship(eq. 8) and statistical measures R2 and se(Y) based on results of studies carried out in Canada, the Netherlands andthe UK. Values in brackets are the standard errors for the coefficients.

The metal in soil drainage water comprisesthe following metal species

Metal species Symbol

Metal free ion M2+ [M]free,sdw

Inorganic complexesMOH+, MHCO3

+, MCl+ etc [M]DIC,sdw

Metal bound to DOM [M]DOM,sdw

Metal bound to SPM [M]SPM,sdw

Here, DOM is dissolved organic matter, andSPM is suspended particulate matter. Thetotal concentration of metal in soil drainagewater does not refer simply to dissolvedcomponents ([M]free,sdw, [M]DIC,sdw, and[M]DOM,sdw), but also includes [M]SPM,sdw.Data on SPM concentration in soil drainagewaters may be scarce, and in many casesthe contribution of SPM to the metal leachingis only small. Thus this flux can be neglectedpreliminarily. The calculation model includes,however, the possibility of metal beingleached from the soil in association with par-ticulates.

Given the activity or concentration of M2+, theconcentrations of the other metal speciescan be estimated by applying an equilibriumspeciation model. The calculation has totake into account the dependence of themetal speciation on pH and competitiveeffects due to major cationic species of Mg,Al, Ca and Fe. For this purpose a customversion of the Windermere Humic AqueousModel version 6 (WHAM6; Tipping 1998)speciation model, termed W6S MTC2, hasbeen produced. A more detailed descriptionof the model calculation steps is given in thebackground document (De Vries et al. 2004b). NFCs may calculate critical dissolved metal concentrations from the freeion concentration by one of three methods:

1. Linear interpolation in the look-up tables (chapter 5.5.2.2.3). The look-up tables list critical dissolved metal concentrations (calculated using W6S-MTC2) for various combinations of pH, concentrations of soil organic matter, dissolved organic carbon ([DOC]sdw) and suspended particulate matter (SPM) and partial CO2 pressure (pCO2).

2. Sending suitably formatted files to the Centre for Ecology & Hydrology (CEH), Lancaster, Ed Tipping ([email protected]), who will perform the computations with W6S-MTC2. Instructions for preparing suitably formatted files for this purpose are given below.

3. Using the W6S-MTC2 program themselves. Instructions for use are given with the program, which can be obtained by contacting Ed Tipping (see above).

NFCs that wish values of Mtot,sdw(crit) to becalculated by should submit files to the CEHLancaster, Ed Tipping ([email protected]).The data should simply be entered into anExcel workbook, under the following headings.

code the user’s identifier of the sitepH soil solution pH% OM the soil organic matter contentpCO2 the soil pCO2 expressed as a

multiple of the atmospheric valueDOC concentration of dissolved

organic carbon in mg l-1

SPM concentration of suspended particulate matter in mg l-1.

• Please see the background document (Annex 8 and 9) regarding the selection of pH and pCO2 values. If data on DOCconcentration are not available, a standard value of 20 mg l-1 will be assumed.



Annex 2: Calculation of total metal concentration from free metal ion concentrations using the WHAM model

code pH % OM pCO2 DOC SPM



• If data on pCO2 are not available, a value of 15 x atmospheric will be ass-umed.

• If data on SPM are not available, a value of zero will be assumed.

Please note that it is necessary to recal-culate values of soil pH (measured in KCl,CaCl2, H2O) to soil solution pH, as mentionedin the main text, before applying the look-uptables or creating input files for W6S-MTC2.

The calculation of the critical total aqueousconcentration comprises the followingsteps:

1. Estimate the critical free metal ion concentration from the critical dissolved concentration.

2. Calculate the metal bound per unit mass of SPM.

3. Sum the total dissolved and particulate concentrations.

Step 1

The free ion concentrations are calculatedusing WHAM6, for waters of different pH,DOC and pCO2, making the same assump-tions as are used for calculating total metalfrom free-ion critical limits (for the Look UpTables). In the calculations the critical dissolved concentrations used were 0.38 mg m-3 for Cd and 11 mg m-3 for Pb. Notethat, here, all waters are assumed to be “normal” with respect to dissolved Al (i.e.acid bog-waters are not included).

The free ion activities calculated withWHAM6 can be expressed in terms of multiple regression equations at different pHvalues. Thus;

(A3.1)

where:[DOC]sw is in mg l-1 and pCO2 is a multiple ofthe atmospheric pCO2. The regression coefficients are given in Tables A3.1 andA3.2. Linear interpolation can be performedto obtain coefficients for intermediate pHvalues.

Step 2

The critical SPM–bound metal ([M]SPM,sw(crit),mol g-1) is calculated using the Q-c relationsderived in Annex 1, eq. A1.5 (Table A1.5).Before proceeding to Step 3 [M]SPM,sw(crit)must be converted to units of mg kg-1:

(A3.2a)

(A3.2b)

Step 3

The total aqueous metal at the critical limit isgiven by:

(A3.3)

where:[M]dis,sw(crit) is the critical dissolved concentration (mg m-3 or µg l-1) (Table 10),[M]SPM,sw(crit) is the critical concentration



Annex 3: Calculation of the critical total aqueous concentra-tion from the critical dissolved concentration using the WHAM model

Table A3.1: Regression coefficients for estimatingfree Cd2+ concentrations

Table A3.2: Regression coefficients for estimatingfree Pb2+ concentrations



bound to SPM calculated in Step 2 (mg kg-1),and [SPM]sw is the concentration of SPM inthe surface water (kg m-3 or g l-1).

Calculation examples

For a water of pH 6 with [DOC] = 8 mg l-1 apCO2 four times the atmospheric value and[SPM]sw = 0.050 g l-1 (50 mg l-1) with 20%organic matter content:

(A3.4)

and

(A3.5)

using the critical limits listed in Table 7 forthe critical limits recommended for the callfor data 2004.

An update of this annex will be made soonby applying critical Cd limits as a function ofwater hardness and applying a critical Pblimit of 5 mg m-3.

Dynamic modelling is the logical extensionof critical loads. Critical loads are based ona steady-state concept, they are the con-stant depositions an ecosystem can toleratein the long run, i.e., after it has equilibratedwith these depositions. However, manyecosystems are not in equilibrium with pres-ent or projected depositions, since there areprocesses (‘buffer mechanisms’) at work,which delay the reaching of an equilibrium(steady state) for years, decades or evencenturies. By definition, critical loads do notprovide any information on these timescales. Dynamic models are needed toassess time delays of recovery in regionswhere critical loads cease being exceededand time delays of damage in regions wherecritical loads continue to be exceeded.

The purpose of this Chapter is to explain theuse (and constraints) of dynamic modellingin support of the effects-oriented work underthe LRTAP Convention. This Chapter is ashortened and updated version of a‘Dynamic Modelling Manual’ published earli-er by the CCE (Posch et al. 2003).

For the sake of simplicity and in order toavoid the somewhat vague term ‘ecosys-tem’, we refer in the sequel to non-calcare-ous (forest) soils. However, most of the con-siderations hold for surface water systemsas well, since their water quality is stronglyinfluenced by properties of and processes incatchment soils. A separate report dealingwith the dynamic modelling of surfacewaters on a regional scale has been pre-pared under the auspices of the ICP Waters(Jenkins et al. 2002).

6.1.1 Why dynamic modelling?

In the causal chain from deposition of strongacids to damage to key indicator organismsthere are two major links that can give rise todelays. Biogeochemical processes candelay the chemical response in soil, and bio-logical processes can further delay theresponse of indicator organisms, such asdamage to trees in forest ecosystems. Thestatic models to determine critical loads

consider only the steady-state condition, inwhich the chemical and biological responseto a (new) (constant) deposition is complete.Dynamic models, on the other hand, attemptto estimate the time required for a new(steady) state to be achieved.

With critical loads, i.e. in the steady-state sit-uation, only two cases can be distinguishedwhen comparing them to deposition: (1) thedeposition is below critical load(s), i.e. doesnot exceed critical loads, and (2) the deposi-tion is greater than critical load(s), i.e. thereis critical load exceedance. In the first casethere is no (apparent) problem, i.e. no reduc-tion in deposition is deemed necessary. Inthe second case there is, by definition, anincreased risk of damage to the ecosystem.Thus a critical load serves as a warning aslong as there is exceedance, since it statesthat deposition should be reduced. However,it is often assumed that reducing depositionto (or below) critical loads immediatelyremoves the risk of ‘harmful effects’, i.e. thechemical criterion (e.g. the Al/Bc-ratio1) thatlinks the critical load to the (biological)effect(s), immediately attains a non-critical(‘safe’) value, and that there is immediatebiological recovery as well. But the reactionof soils, especially their solid phase, tochanges in deposition is delayed by (finite)buffers, the most important being the cationexchange capacity (CEC). These buffermechanisms can delay the attainment of acritical chemical parameter, and it might takedecades or even centuries, before an equi-librium (steady state) is reached. These finitebuffers are not included in the critical loadformulation, since they do not influence thesteady state, but only the time to reach it.Therefore, dynamic models are needed toestimate the times involved in attaining acertain chemical state in response to depo-sition scenarios, e.g. the consequences of‘gap closures’ in emission reduction negoti-ations. In addition to the delay in chemicalrecovery, there is likely to be a further delaybefore the ‘original’ biological state isreached, i.e. even if the chemical criterion ismet (e.g. Al/Bc<1), it will take time before biological recovery is achieved.

6 Dynamic Modelling

6.1 Introduction

Mapping Manual 2004 • Chapter VI Dynamic Modelling Page VI - 1

1 In Chapter 5 (and elsewhere) the Bc/Al-ratio is used.However, this ratio becomes infinite when the Al concentra-

tion approaches zero. To avoid this inconvenience, itsinverse, the Al/Bc-ratio, is used here.

6 Dynamic Modelling

Mapping Manual 2004 • Chapter VI Dynamic ModellingPage VI - 2

Figure 6.1 summarises the possible develop-ment of a (soil) chemical and biological variable in response to a ‘typical’ temporaldeposition pattern. Five stages can be distinguished:

Stage 1: Deposition was and isbelow the critical load (CL) and the chemicaland biological variables do not violate theirrespective criteria. As long as depositionstays below the CL, this is the ‘ideal’ situa-tion.

Stage 2: Deposition is above the CL,but (chemical and) biological criteria are notviolated because there is a time delay beforethis happens. No damage is likely to occur atthis stage, therefore, despite exceedance ofthe CL. The time between the firstexceedance of the CL and the first violationof the biological criterion (the first occur-rence of actual damage) is termed theDamage Delay Time (DDT=t3–t1).

Stage 3: The deposition is above theCL and both the chemical and biological criteria are violated. Measures (emissionreductions) have to be taken to avoid a (fur-ther) deterioration of the ecosystem status.

Stage 4: Deposition is below the CL,but the (chemical and) biological criteria are

still violated and thus recovery has not yetoccurred. The time between the first non-exceedance of the CL and the subsequentnon-violation of both criteria is termed theRecovery Delay Time (RDT=t6–t4).

Stage 5: Deposition is below the CLand both criteria are no longer violated. Thisstage is similar to Stage 1 and only at thisstage can the ecosystem be considered tohave recovered.

Stages 2 and 4 can be subdivided into twosub-stages each: Chemical delay times(DDTc=t2–t1 and RDTc=t5–t4; dark grey inFigure 6.1) and (additional) biological delaytimes (DDTb=t3–t2 and RDTb=t6–t5; light grey).Very often, due to the lack of operational bio-logical response models, damage andrecovery delay times mostly refer to chemical recovery alone and this is used asa surrogate for overall recovery. It is alsoimportant to note that recovery does not follow the same, (inverse) path of damage,since there is a so-called hysteresis in thesenatural systems (see, e.g., Warfvinge et al.1992).

Figure 6.1: ‘Typical’ past and future development of the acid deposition effects on a soil chemical variable (Al/Bc-ratio) and the corresponding biological response in comparison to the critical values of those variables and the critical load derived from them. The delay between the (non)exceedance of the critical load, the (non)violation of thecritical chemical criterion and the crossing of the critical biological response is indicated in grey shades, high-lighting the Damage Delay Time (DDT) and the Recovery Delay Time (RDT) of the system.

6.1.2 Constraints for dynamic modelling under the LRTAP Convention

Steady-state models (critical loads) havebeen used to negotiate emission reductionsin Europe. In this context, an emissionreduction is judged successful if non-exceedance of critical loads is attained.To gain insight into the time delay betweenthe attainment of non-exceedance and actu-al chemical (and biological) recovery,dynamic models are needed. Thus if dynamic models are to be used to assess recovery under the LRTAP Convention, they should be compatible with the steady-state models used for calculatingcritical loads. In other words, when criticalloads are used as input to the dynamicmodel, the (chemical) parameter chosen asthe criterion in the critical load calculationhas to attain the critical value (after thedynamic simulation has reached steadystate). But this also means that conceptsand equations used in the dynamic modelhave to be an extension of the concepts andequations employed in deriving the steady-state model. For example, if critical loads arecalculated with the Simple Mass Balance(SMB) model (see Chapter 5), this modelshould be the steady-state version of thedynamic model used (e.g., the VSD model,see below).

Due to a lack of (additional) data, it may beimpossible to run dynamic models on allsites in a country for which critical loadshave been calculated. The selection of thesubset of sites, at which dynamic models areapplied, has to be representative enough toallow comparison with results obtained withcritical loads.

Dynamic models of acidification are basedon the same principles as steady-state mod-els: The charge balance of the ions in the soilsolution, mass balances of the various ions,and equilibrium equations. However, where-as in steady-state models only infinitesources and sinks are considered (such asbase cation weathering), the inclusion of the

finite sources and sinks of major ions intodynamic models is crucial, since they deter-mine the long-term (slow) changes in soil(solution) chemistry. The three most impor-tant processes involving finite buffers andtime-dependent sources/sinks are cationexchange, nitrogen retention and sulphateadsorption.

A short description of the models mentioned in this section, such as VSD,MAGIC, SAFE and SMART, can be found inSection 6.3.

6.2.1 Charge and mass balances

As mentioned above, we consider as‘ecosystem’ non-calcareous forest soils,although most of the considerations holdalso for non-calcareous soils covered by(semi-)natural vegetation. Since we are inter-ested in applications on a large regionalscale (for which data are scarce) and longtime horizons (decades to centuries with atime step of one year), we make the samesimplifying assumption as for the SMBmodel (see Chapter 5). We assume that thesoil is a single homogeneous compartmentand its depth is equal to the root zone. Thisimplies that internal soil processes (such asweathering and uptake) are evenly distrib-uted over the soil profile, and all physico-chemical constants are assumed uniform inthe whole profile. Furthermore we assumethe simplest possible hydrology: The waterleaving the root zone is equal to precipitationminus evapotranspiration; more precisely,percolation is constant through the soil profile and occurs only vertically.

As for the SMB model, the starting point isthe charge balance of the major ions in thesoil water, leaching from the root zone (cf. eq.5.9):

(6.1)

where BC=Ca+Mg+K+Na and RCOO standsfor the sum of organic anions. Eq. 6.1 alsodefines the acid neutralising capacity, ANC.

6 Dynamic Modelling


6.2 Basic Concepts and Equations

6 Dynamic Modelling


The leaching term is given by Xle=Q·[X]where [X] is the soil solution concentration(eq/m3) of ion X and Q (m/yr) is the waterleaving the root zone.

The concentrations [X] of an ion in the soilcompartment, and thus its leaching Xle, areeither obtained from equilibrium equationswith [H], such as [Al], [HCO3] and [RCOO](see eqs. 5.42, 5.43 and 5.45), or from massbalance equations. The latter describe thechange over time of the total amount of ionX per unit area in the soil matrix/soil solution,Xtot (eq/m2):

(6.2)

where Xin (eq/m2/yr) is the net input of ion X(sources minus sinks, except leaching).

With the simplifying assumptions used in thederivation of the SMB model, the net input ofsulphate and chloride is given by theirrespective deposition:

(6.3)

For base cations the net input is given by(Bc=Ca+Mg+K):

(6.4)

where the subscripts dep, w and u stand fordeposition, weathering and net uptake,respectively. Note, that S adsorption andcation exchange reactions are not includedhere, they are included in Xtot and describedby equilibrium equations (see below). Fornitrate and ammonium the net input is givenby:

(6.5)

(6.6)

where the subscripts ni, i and de stand fornitrification, net immobilisation and denitrifi-cation, respectively. In the case of completenitrification one has NH4,in=0 and the netinput of nitrogen is given by:

(6.7)

6.2.2 From steady state (critical loads) to dynamic models

Steady state means there is no change overtime in the total amounts of ions involved,i.e. (see eq.6.2):

(6.8)

From eq.6.7 the critical load of nutrient nitro-gen, CLnut(N), is obtained by specifying anacceptable N-leaching, Nle,acc. By specifyinga critical leaching of ANC, ANCle,crit, andinserting eqs. 6.3, 6.4 and 6.7 into the chargebalance (eq. 6.1), one obtains the equationdescribing the critical load function of S andN acidity, from which the three quantitiesCLmax(S), CLmin(N) and CLmax(N) can bederived (see Chapter 5).

To obtain time-dependent solutions of themass balance equations, the term Xtot in eq. 6.2, i.e. the total amount (per unit area) ofion X in the soil matrix/soil solution systemhas to be specified. For ions, which do notinteract with the soil matrix, Xtot is given bythe amount of ion X in solution alone:

(6.9)

where z (m) is the soil depth under consider-ation (root zone) and Q (m3/m3) is the (annualaverage) volumetric water content of the soil

compartment. The above equation holds forchloride. For every base cation Y participat-ing in cation exchange, Ytot is given by:

(6.10)

where r is the soil bulk density (g/cm3), CECthe cation exchange capacity (meq/kg) andEY is the exchangeable fraction of ion Y.

The (long-term) changes of the soil N poolare mostly caused by net immobilisation,and Ntot is given by:

(6.11)

If there is no ad/desorption of sulphate,SO4,tot is given by eq. 6.9. If sulphate adsorp-tion cannot be neglected, it is given by:

(6.12)

When the rate of Al leaching is greater thanthe rate of Al mobilisation by weathering ofprimary minerals, the remaining part of Alhas to be supplied from readily available Alpools, such as Al hydroxides. This causesdepletion of these minerals, which mightinduce an increase in Fe buffering which inturn leads to a decrease in the availability ofphosphate (De Vries 1994). Furthermore, thedecrease of those pools in podzolic sandysoils may cause a loss in the structure ofthose soils. The amount of aluminium is inmost models assumed to be infinite and thusno mass balance for Al is considered. TheSMART model, however, includes an Albalance, and the terms in eq. 6.2 areAlin=Alw, and Altot is given by:

(6.13)

where Alox (meq/kg) is the amount of oxalate

extractable Al, the pool of readily available Alin the soil.

Inserting these expressions into eq. 6.2 andobserving that Xle=Q·[X], one obtains differ-ential equations for the temporal develop-ment of the concentration of the differentions. Only in the simplest cases can theseequations be solved analytically. In general,the mass balance equations are discretisedand solved numerically, with the solutionalgorithm depending on the model builders’preferences.

6.2.3 Finite buffers

Finite buffers of elements in the soil are notincluded in the derivation of critical loads,since they do not influence the steady state.However, when investigating the state soilsover time as a function of changing deposi-tion patterns, these finite buffers govern thelong-term (slow) changes in soil (solution)chemistry. In the following we describe themost important ones in turn.

6.2.3.1 Cation exchange

Generally, the solid phase particles of a soilcarry an excess of cations at their surfacelayer. Since electro-neutrality has to bemaintained, these cations cannot beremoved from the soil, but they can beexchanged against other cations, e.g. thosein the soil solution. This process is known ascation exchange; and every soil (layer) ischaracterised by the total amount ofexchangeable cations per unit mass(weight), the so-called cation exchangecapacity (CEC, measured in meq/kg). If X andY are two cations with charges m and n, thenthe general form of the equations used todescribe the exchange between the liquid-phase concentrations (or activities) [X] and[Y] and the equivalent fractions EX and EY atthe exchange complex is

(6.14)

6 Dynamic Modelling


6 Dynamic Modelling


where KXY is the so-called exchange (or selectivity) constant, a soil-dependentquantity. Depending on the powers i and jdifferent models of cation exchange can bedistinguished: For i=n and j=m one obtainsthe Gaines-Thomas exchange equations,whereas for i=j=mn, after taking the mn-throot, the Gapon exchange equations areobtained.

The number of exchangeable cations considered depends on the purpose andcomplexity of the model. For example,Reuss (1983) considered only the exchangebetween Al and Ca (or divalent base cations).In general, if the exchange between N ionsis considered, N–1, exchange equations (andconstants) are required, all the other(N–1)(N–2)/2 relationships and constants canbe easily derived from them. In the VSD,SMART and SAFE model the exchangebetween aluminium, divalent base cationsand protons is considered. The exchange ofprotons is important, if the cation exchangecapacity (CEC) is measured at high pH-val-ues (pH=6.5). In the case of the Bc-Al-Hsystem, the Gaines-Thomas equations read:

(6.15)

where Bc=Ca+Mg+K, with K treated as divalent. The equation for the exchange ofprotons against Al can be obtained from eqs. 6.15 by division:

(6.16)

The corresponding Gapon exchange equa-tions read:

(6.17)

Again, the H-Al exchange can be obtainedby division (with kHAl=kHBc/kAlBc). Charge

balance requires that the exchangeable frac-tions add up to one:

(6.18)

The sum of the fractions of exchangeablebase cations (here EBc) is called the basesaturation of the soil; and it is the time development of the base saturation, which isof interest in dynamic modelling. In theabove formulations the exchange of Na, NH4(which can be important in high NH4 deposi-tion areas) and heavy metals is neglected (orsubsumed in the proton fraction).

Care has to be exercised when comparingmodels, since different sets of exchangeequations are used in different models.Whereas eqs. 6.15 are used in the SMARTmodel (but with Ca+Mg instead of Bc, K-exchange being ignored in the current version), the SAFE model employs theGapon exchange equations (eqs. 6.17), how-ever with exchange constants k’X/Y=1/kXY. Inthe MAGIC model the exchange of Al with allfour base cations is modelled separatelywith Gaines-Thomas equations, withoutexplicitly considering H-exchange.

6.2.3.2 Nitrogen immobilisation

In the calculation of critical loads the(acceptable, sustainable) long-term netimmobilisation (i.e. the difference betweenimmobilisation and mineralisation) isassumed to be constant. However, it is wellknown, that the amount of N immobilised is(at present) in many cases larger than thislong-term value. Thus a submodel describ-ing the nitrogen dynamics in the soil is partof most dynamic models. For example, theMAKEDEP model, which is part of the SAFEmodel system (but can also be used as astand-alone routine) describes the N-dynamics in the soil as a function of forestgrowth and deposition.

According to Dise et al. (1998) andGundersen et al. (1998) the forest floor C/N-ratios may be used to assess risk fornitrate leaching. Gundersen et al. (1998)

suggested threshold values of >30, 25 to 30,and <25 to separate low, moderate, and highnitrate leaching risk, respectively. This infor-mation has been used in several models,such as SMART and MAGIC to calculatenitrogen immobilisation as a fraction of thenet N input, linearly depending on the C/N-ratio in the mineral topsoil.In addition to the long-term constant netimmobilisation, Ni,acc, the net amount of Nimmobilised is a linear function of the actualC/N-ratio, CNt, between a prescribed maximum, CNmax, and a minimum C/N-ratio,CNmin:

(6.19)

where Nin,t is the available N (e.g., Nin,t=Ndep,t–Nu,t–Ni,acc). At every time step theamount of immobilised N is added to theamount of N in the top soil, which in turn isused to update the C/N-ratio. The totalamount immobilised at every time step isthen Ni=Ni,acc+Ni,t. The above equationstates that when the C/N-ratio reaches theminimum value, the annual amount of Nimmobilised equals the acceptable valueNi,acc (see Figure 6.2). This formulation iscompatible with the critical load formulationfor t®¥.

6.2.3.3 Sulphate adsorption

The amount of sulphate adsorbed, SO4,ad(meq/kg), is often assumed to be in equilibri-

um with the solution concentration and istypically described by a Langmuir isotherm(e.g., Cosby et al. 1986):

(6.20)

where Smax is the maximum adsorptioncapacity of sulphur in the soil (meq/kg) andS1/2 the half-saturation concentration(eq/m3).

6.2.4 From soils to surface waters

The processes discussed so far areassumed to occur in the soil solution while itis in contact with the soil matrix. To calculatesurface water concentrations it is assumedthat the water leaves the soil matrix and isexposed to the atmosphere (Cosby et al.1985, Reuss and Johnson 1986). When thisoccurs, excess CO2 in the water degasses.This shifts the carbonate-bicarbonate equi-libria and changes the pH (see eq. 5.43).Surface water concentrations are thus cal-culated by resolving the system of equationspresented above at a lower partial pressureof CO2 (e.g. mean pCO2 of 8·10-4 atm for 37lakes, Cole et al. 1994) while ignoringexchange reactions, nitrogen immobilisationand sulphate adsorption. Since exchangeswith the soil matrix are precluded, the con-centration of the base cations and the strongacid anions (SO4, NO3 and Cl) will not changeas the soil water becomes surface water. Assuch, ANC is conservative (see eq. 6.1).

6 Dynamic Modelling


0 50 100 150 200 250 3000

0.1

0.2

0.3

0.4

0.5

years

Ni(e

q/m

2 /yr

)

0 50 100 150 200 250 3000

5

10

15

20

25

years

C/N

rat

io (

g/g)

Figure 6.2: Amount of N immobilised (left) and resulting C/N-ratio in the topsoil (right) for a constant net input of N

of 1 eq/m2/yr (initial Cpool = 4000 gC/m

2), Ni,acc = 1 kg/ha/yr). Figure 6.2: Amount of N immobilised (left) and resulting C/N-ratio in the topsoil (right) for a constant net input of N

of 1 eq/m2/yr (initial Cpool = 4000 gC/m2, Ni,acc = 1 kg/ha/yr).

6 Dynamic Modelling


6.2.5 Biological response models

Just as there are delays between changes inacid deposition and changes in surface (orsoil) water chemistry, there are delaysbetween changes in chemistry and the bio-logical response. Because the goal in recov-ery is to restore good or healthy populationof key indicator organisms, the time lag inresponse is the sum of the delays in chemi-cal and biological response (see Figure 6.1).Thus dynamic models for biologicalresponse are needed; and in the following asummary is provided of existing models andideas.

6.2.5.1 Terrestrial ecosystems

A major drawback of most dynamic soilacidification models is the neglect of bioticinteractions. For example, vegetationchanges are mainly triggered by a change inN cycling (N mineralisation; Berendse et al.1987). Furthermore the enhancement of dis-eases by elevated N inputs, such as heatherbeetle outbreaks, may stimulate vegetationchanges. Consequently, dynamic soil-vege-tation models, which include such process-es, have a better scientific basis for theassessment of critical and target N loads.Examples of such models are CALLUNA(Heil and Bobbink 1993) and ERICA(Berendse 1988). The model CALLUNA inte-grates N processes by atmospheric deposi-tion, accumulation and sod removal, withheather beetle outbreaks and competitionbetween species, to establish the critical Nload in lowland dry-heathlands (Heil andBobbink 1993). The wet-heathland modelERICA incorporates the competitive relation-ships between the species Erica andMolinia, the litter production from bothspecies, and nitrogen fluxes by accumula-tion, mineralisation, leaching, atmosphericdeposition and sheep grazing. At presentthere are also several forest-soil models thatdo calculate forest growth impacts inresponse to atmospheric deposition andother environmental aspects, such as meteorological changes (precipitation,temperature) and changes in CO2 concentra-tion. Examples are the models NAP (VanOene 1992), ForSVA (Oja et al. 1995) andHybrid (Friend et al. 1997).

To date, biological dose/response modelsrelated to impacts on species diversity in ter-restrial ecosystems did not focus on time-dynamic aspects. Instead, statistical modelshave been developed to assess the relation-ship between the species diversity of theecosystem and abiotic aspects related toacidification and eutrophication. An exampleis the vegetation model MOVE (Latour andReiling 1993), that predicts the occurrenceprobability of plant species in response toscenarios for acidification, eutrophicationand desiccation. Input to the model comesfrom the output of the soil model SMART2(Kros et al. 1995), being an extension ofSMART. The SMART2 model predictschanges in abiotic soil factors indicatingacidification (pH), eutrophication (N avail-ability) and desiccation (moisture content) inresponse to scenarios for acid depositionand groundwater abstraction, including theimpact of nutrient cycling (litterfall, minerali-sation and uptake). MOVE predicts theoccurrence probability of ca 700 species asa function of three abiotic soil factors,including nitrogen availability, using regres-sion relationships. Since combined samplesof vegetation and environmental variablesare rare, the indication values of plantspecies by Ellenberg (1985) are used toassess the abiotic soil conditions. Deductionof values for the abiotic soil factors from thevegetation guarantees ecological relevance.Combined samples of vegetation with envi-ronmental variables are used exclusively tocalibrate Ellenberg indication values withquantitative values of the abiotic soil factors.A calibration of these indication values toquantitative values of the abiotic soil factorsis necessary to link the soil module to thevegetation module.

A comparable statistical model is the NTMmodel (Wamelink et al. 2003, Schouwenberget al. 2000), that was developed to predictthe potential conservation value of naturalareas. Normally conservation values are cal-culated on the basis of plant species or veg-etation types. As with MOVE, NTM has thepossibility to link the vegetation and the siteconditions by using plant ecological indica-tor values. NTM uses a matrix of the habitatsof plant species defined on the basis of

moisture, acidity and nutrient availability.The model was calibrated using a set of160,252 vegetation relevees. A value indexper plant species was defined on the basisof rarity, decline and international impor-tance. This index was used to determine aconservation value for each relevee. Thevalue per relevee was then assigned to eachspecies in the relevee and regressed on theEllenberg indicator values for moisture, acid-ity and nutrient availability (Ellenberg 1985)using a statistical method (P-splines). Themodel has these three Ellenberg indicationvalues as input for the prediction of thepotential conservation value. A potentialconservation value is calculated for a combi-nation of the abiotic conditions and vegeta-tion structure (ecotope). Therefore four veg-etation types are accounted for, each repre-sented by a submodel of NTM: heathland,grassland, deciduous forest and pine-forest.Use of those models in dynamic modellingassessments is valuable to gain more insightin the effect of deposition scenarios on terrestrial ecosystems.

6.2.5.2 Aquatic ecosystems

As with terrestrial ecosystems, biologicaldose/response models for surface watershave not generally focussed on the time-dynamic aspects. For example, the relation-ship between lake ANC and brown trout population status in Norwegian lakes usedto derive the critical limit for surface watersis based on synoptic (once in time) surveysof ANC and fish status in a large number oflakes. Similarly the invertebrate indices(Raddum 1999) and diatom response models(Allot et al. 1995, Battarbee et al. 1996) donot incorporate time-dynamic aspects.Additional information on dose/responsecomes from traditional laboratory studies oftoxicity (chronic and acute) and reproductivesuccess.

Information on response times for variousorganisms comes from studies of recoveryfollowing episodes of pollution, for example,salmon population following chemical spill ina river. For salmon full recovery of the popu-lation apparently requires about 10 yearsafter the water chemistry has been restored.

There are currently no available time-dynamic process-oriented biologicalresponse models for effects of acidificationon aquatic and terrestrial organisms, com-munities or ecosystems. Such models arenecessary for a full assessment of the lengthof time required for recovery of damage fromacidification.

There are several types of evidence that canbe used to empirically estimate the timedelays in biological recovery. The whole-lakeacidification and recovery experiments conducted at the Experimental Lakes Area(ELA), north-western Ontario, Canada, provide such information at realistic spatialand temporal scales. These experimentsdemonstrate considerable lag timesbetween achievement of acceptable waterquality following decrease in acid inputs,and achievement of acceptable biologicalstatus. The delay times for various organisms are at least several years. In thecase of several fish species irreversiblechanges may have occurred (Hann andTurner 2000, Mills et al. 2000).

A second source of information on biologicalrecovery comes from liming studies. Overthe years such studies have producedextensive empirical evidence on rate ofresponse of individual species as well ascommunities following liming. There hasbeen little focus, however, on the processesinvolved.

Finally there is recent documentation ofrecovery in several regions at which aciddeposition has decreased in the 1980s and1990s. Lakes close to the large point-sourceof sulphur emissions at Sudbury, Ontario,Canada, show clear signs of chemical andbiological recovery in response to substan-tial decreases in emissions beginning in thelate 1970s (Keller and Gunn 1995). Lakes inthe nearby Killarney Provincial Park (Ontario,Canada) also show clear signs of biologicalrecovery during the past 20 years (Snucinset al. 2001). Here there are several biologicalfactors that influence the rate of biologicalrecovery such as:(1) fish species composition and density(2) dispersal factors such as distance to

intact population and ability to disperse

6 Dynamic Modelling


6 Dynamic Modelling


(3) existence of resting eggs (for such organisms such as zooplankton)

(4) existence of precluding species – i.e. the niche is filled.

Recently, a workshop held under the auspices of the UNECE reviewed the currentknowledge on models for biological recoveryin surface waters (see Wright and Lie 2002).

Future work on biological response modelsmust also include consideration of the fre-quency and severity of harmful episodes,such as pH shocks during spring snowmelt,or acidity and aluminium pulses due tostorms with high seasalt inputs. These linksbetween episodic water chemistry and bio-logical response at all levels (organisms,community, and ecosystem) are poorlyquantified and thus not yet ready to be incor-porated into process-oriented models.

6.3 Available Dynamic Models

In the previous sections the basic processesinvolved in soil acidification have been sum-marised and expressed in mathematicalform, with emphasis on slow (long-term)processes. The resulting equations, or gen-eralisations and variants thereof, togetherwith appropriate solution algorithms andinput-output routines have over the past 15 years been packaged into soil acidifica-tion models, mostly known by their (more orless fancy) acronyms.

There is no shortage of soil (acidification)models, but most of them are not designedfor regional applications. A comparison of 16 models can be found in a special issue ofthe journal ‘Ecological Modelling’ (Tiktak andVan Grinsven 1995). These models emphasise either soil chemistry (such asSMART, SAFE and MAGIC) or the interactionwith the forest (growth). There are very fewtruly integrated forest-soil models. An example is the forest model series ForM-S(Oja et al. 1995), which is implemented not ina ‘conventional’ Fortran code, but is realisedin the high-level modelling software STELLA.

The following selection is biased towardsmodels which have been (widely) used andwhich are simple enough to be applied on a(large) regional scale. Only a short descrip-tion of the models is given, but details can

be found in the references cited. It should beemphasised that the term ‘model’ used hererefers, in general, to a model system, i.e. aset of (linked) software (and databases)which consists of pre-processors for inputdata (preparation) and calibration, post-processors for the model output, and – ingeneral the smallest part – the actual modelitself.

An overview of the various models is given inTable 6.1 and a short description below. Thefirst three models are soil models of increas-ing complexity, whereas the MAGIC model is

Table 6-1: Overview of dynamic models that have been (widely) applied on a regional scale.

Model Essential process descriptions Layers Essential model inputs Contact

VSD ANC charge balance

Mass balances for BC and N

(complete nitrification assumed)

One CL input data +

CEC, base saturation

C/N-ratio

M Posch

SMART VSD model +

SO4 sorption

Mass balances for CaCO3 and Al

Separate mass balances for NH4

and NO3, nitrification

Complexation of Al with DOC

One VSD model +

Smax and S1/2

Ca-carbonate, Alox

nitrification fraction,

pK values

W de Vries

SAFE VSD model +

Separate weathering calculation,

Element cycling by litterfall,

Root decay,

Mineralisation and root uptake

Several VSD model +

Input data for PROFILE,

litterfall rate,

parameters describing

mineralisation and root uptake

H Sverdrup

MAGIC VSD model +

SO4 sorption,

Al speciation/complexation,

Aquatic chemistry

Several

(mostly

one)

VSD model +

Smax and S1/2

pK values for several Al reactions

parameters for aquatic chemistry

RF Wright

Table 6.1: Overview of dynamic models that have been (widely) applied on a regional scale.

generally applied at the catchment level.Application on the catchment level, insteadon a single (forest) plot, has implications forthe derivation of input data. For example,weathering rates have to represent the aver-age weathering of the whole catchment,data that is difficult to obtain from soilparameters. Thus in MAGIC catchmentweathering is calibrated from water qualitydata.

6.3.1 The VSD model

The basic equations presented in section 6.2have been used to construct a Very SimpleDynamic (VSD) soil acidification model. TheVSD model is designed as the simplestextension of the SMB model for criticalloads. It only includes cation exchange andN immobilisation, and a mass balance forcations and nitrogen as described above, inaddition to the equations included in theSMB model. It resembles the model pre-sented by Reuss (1980) which, however,does not consider nitrogen processes.

In the VSD model, the various ecosystemprocesses have been limited to a few keyprocesses. Processes that are not taken intoaccount, are: (i) canopy interactions, (ii)nutrient cycling processes, (iii) N fixation andNH4 adsorption, (iv) interactions (adsorption,uptake, immobilisation and reduction) ofSO4, (v) formation and protonation of organicanions, (RCOO) and (vi) complexation of Alwith OH, SO4 and RCOO.

The VSD model consists of a set of massbalance equations, describing the soil input-output relationships, and a set of equationsdescribing the rate-limited and equilibriumsoil processes, as described in section 6.2.The soil solution chemistry in VSD dependssolely on the net element input from theatmosphere (deposition minus net uptakeminus net immobilisation) and the geochem-ical interaction in the soil (CO2 equilibria,weathering of carbonates and silicates, andcation exchange). Soil interactions aredescribed by simple rate-limited (zero-order)reactions (e.g. uptake and silicate weather-ing) or by equilibrium reactions (e.g. cationexchange). It models the exchange of Al, H

and Ca+Mg+K with Gaines-Thomas orGapon equations. Solute transport isdescribed by assuming complete mixing ofthe element input within one homogeneoussoil compartment with a constant densityand a fixed depth. Since VSD is a single layersoil model neglecting vertical heterogeneity,it predicts the concentration of the soil waterleaving this layer (the rootzone). The annualwater flux percolating from this layer is takenequal to the annual precipitation excess. The

time step of the model is one year, i.e. seasonal variations are not considered. Themodel is available from the CCE website(www.rivm.nl/cce); and a detailed description will be found in Posch andReinds (2005).

6.3.2 The SMART model

The SMART model (Simulation Model forAcidification’s Regional Trends) is similar tothe VSD model, but somewhat extended andis described in De Vries et al. (1989) andPosch et al. (1993). As with the VSD model,the SMART model consists of a set of massbalance equations, describing the soil input-output relationships, and a set of equationsdescribing the rate-limited and equilibriumsoil processes. It includes most of theassumptions and simplifications given forthe VSD model; and justifications for themcan be found in De Vries et al. (1989). SMARTmodels the exchange of Al, H and divalentbase cations using Gaines-Thomas equa-tions. Additionally, sulphate adsorption ismodelled using a Langmuir equation (as inMAGIC) and organic acids can be describedas mono-, di- or tri-protic. Furthermore, itincludes a balance for carbonate and Al;thus allowing the calculation from calcare-ous soils to completely acidified soils that donot have an Al buffer left. In this respect,SMART is based on the concept of bufferranges expounded by Ulrich (1981). Recentlya description of the complexation of alumini-um with organic acids has been included.The SMART model has been developed withregional applications in mind, and an earlyexample of an application to Europe can befound in De Vries et al. (1994).

6 Dynamic Modelling


6 Dynamic Modelling


6.3.3 The SAFE model

The SAFE (Soil Acidification in ForestEcosystems) model has been developed atthe University of Lund (Warfvinge et al. 1993)and a recent description of the model can befound in Alveteg (1998) and Alveteg andSverdrup (2002). The main differences to theSMART and MAGIC models are: (a) weather-ing of base cations is not a model input, butit is modelled with the PROFILE (sub-)model,using soil mineralogy as input (Warfvingeand Sverdrup 1992); (b) SAFE is oriented tosoil profiles in which water is assumed tomove vertically through several soil layers(usually 4), (c) Cation exchange between Al,H and (divalent) base cations is modelledwith Gapon exchange reactions, and theexchange between soil matrix and the soilsolution is diffusion limited. The standardversion of SAFE does not include sulphateadsorption although a version, in which sulphate adsorption is dependent on sulphate concentration and pH has recentlybeen developed (Martinson et al. 2003).

The SAFE model has been applied to manysites and more recently also regional applications have been carried out forSweden (Alveteg and Sverdrup 2002) andSwitzerland (SAEFL 1998, Kurz et al. 1998,Alveteg et al. 1998).

6.3.4 The MAGIC model

MAGIC (Model of Acidification of Ground-water In Catchments) is a lumped-parametermodel of intermediate complexity, devel-oped to predict the long-term effects ofacidic deposition on soils and surface waterchemistry (Cosby et al. 1985a,b,c, 1986). Themodel simulates soil solution chemistry andsurface water chemistry to predict themonthly and annual average concentrationsof the major ions in lakes and streams.MAGIC represents the catchment withaggregated, uniform soil compartments (oneor two) and a surface water compartmentthat can be either a lake or a stream. MAGICconsists of (1) a section in which the concen-trations of major ions are assumed to begoverned by simultaneous reactions involv-ing sulphate adsorption, cation exchange,dissolution-precipitation-speciation of

aluminium and dissolution-speciation ofinorganic and organic carbon, and (2) amass balance section in which the flux ofmajor ions to and from the soil is assumed tobe controlled by atmospheric inputs, chemi-cal weathering inputs, net uptake in biomassand losses to runoff. At the heart of MAGICis the size of the pool of exchangeable basecations in the soil. As the fluxes to and fromthis pool change over time owing to changesin atmospheric deposition, the chemicalequilibria between soil and soil solution shiftto give changes in surface water chemistry.The degree and rate of change in surfacewater acidity thus depend both on flux factors and the inherent characteristics ofthe affected soils.

The soil layers can be arranged vertically orhorizontally to represent important verticalor horizontal flowpaths through the soils. If alake is simulated, seasonal stratification ofthe lake can be implemented. Time steps aremonthly or yearly. Time series inputs to themodel include annual or monthly estimatesof: (1) deposition (wet plus dry) of ions fromthe atmosphere; (2) discharge volumes andflow routing within the catchment; (3) biological production, removal and trans-formation of ions; (4) internal sources andsinks of ions from weathering or precipita-tion reactions; and (5) climate data. Constantparameters in the model include physicaland chemical characteristics of the soils andsurface waters, and thermodynamic constants. The model is calibrated usingobserved values of surface water and soilchemistry for a specified period.

MAGIC has been modified and extendedseveral times from the original version of1984. In particular organic acids have beenadded to the model (version 5; Cosby et al.1995a) and most recently nitrogen processeshave been added (version 7; Cosby et al.2001).

The MAGIC model has been extensivelyapplied and tested over a 15-year period atmany sites and in many regions around theworld. Overall, the model has proven to berobust, reliable and useful in a variety of scientific and managerial activities.

Running a dynamic model is usually the leasttime- or resource-consuming step in anassessment. It takes more time to interpretmodel output, but most time-consuming isthe acquisition and preparation of inputdata. Rarely can laboratory or literature databe directly used as model inputs. They haveto be pre-processed and interpreted, oftenwith the help of other models. Especially forregional applications not all model inputs areavailable (or directly usable) from measure-ments at sites, and interpolations and transfer functions have to be derived andused to obtain the necessary input data.When acquiring data from different sourcesof information, it is important to keep arecord of the ‘pedigree’, i.e. the entire chainof information, assumptions and (mental)models used to produce a certain number.Also the uncertainty of the data should beassessed, recorded and communicated.

As with critical loads, for the policy supportof the effects-oriented work under theLRTAP Convention output of dynamic models will most usefully represent not aparticular site, but a larger area, e.g. a forestinstead of a single tree stand. Therefore, certain variables should be ‘smoothed’ torepresent that larger area. For example, (projected) growth uptake of nutrients (nitrogen and base cations) should reflectthe (projected) average uptake of the forestover that area, and not the succession ofharvest and re-growth at a particular spot.

6.4.1 Input data

The input data required to run dynamic models depend on the model, but essentially all of them need the following(minimum) data, which can be roughlygrouped into in- and output fluxes, and soilproperties. Note that this grouping of theinput data depends on the model considered. For example, weathering has tobe specified as a (constant) input flux in theSMART and MAGIC model, whereas in the

SAFE model it is internally computed fromsoil properties and depends on the state ofthe soil (e.g. the pH). Some of the input dataare also needed in the SMB model to calcu-late critical loads, and are described inChapter 5. This chapter thus focuses onadditional data and parameters needed torun dynamic models. The most importantsoil parameters are the cation exchangecapacity (CEC), the base saturation and theexchange (or selectivity) constants describ-ing cation exchange, as well as parametersdescribing nitrogen retention and sulphatead/desorption, since these parametersdetermine the long-term behaviour (recovery) of soils.

Ideally, all input data are directly derivedfrom measurements. This is usually not feasible for regional applications, in whichcase input data have to be derived from relationships (transfer functions) with basic(map) information. In this chapter we provideinformation on the input data needed for running the VSD model and thus, by extension, also other models. Descriptionsand technical details of the input data forthose models can be found in Posch et al.(1993) for the SMART model, in Cosby et al.(1985a) for the MAGIC model and in Alvetegand Sverdrup (2002) for the SAFE model.

In most of the (pedo-)transfer functions presented here, soils – or rather soil groups – are characterised by a few properties, mostly organic carbon and claycontent of the mineral soil (see also Figure6.3) . The organic carbon content, Corg, canbe estimated as 0.5 or 0.4 times the organicmatter content in the humus or mineral soillayer, resp. If Corg>15%, a soil is considereda peat soil. Mineral soils are called sand(or sandy soil) here, if the clay content isbelow 18% (coarse textured soils; see alsoTable 5.6), otherwise it is called a clay(or clayey/loamy soil). Loess soils are soilswith more than 50% silt, i.e. clay+sand<50%(since clay+silt+sand=100%).

6 Dynamic Modelling


6.4 Input Data and Model Calibration

6 Dynamic Modelling


6.4.1.1 Averaging soil properties

For single layer soil models, such as VSD,SMART or MAGIC, the profile averages ofcertain soil parameters are required, and inthe sequel formulae for the average bulkdensity, cation exchange capacity and basesaturation are derived.

For a given soil profile it is assumed thatthere are measurements of bulk density rl(g/cm3), cation exchange capacity CECl(meq/kg) and base saturation EBC,l for n

(homogeneous) soil horizons with thicknesszl (l=1,..., n). Obviously, the total thickness(soil depth) z is given by:

(6.21)

The mean bulk density r of the profile isderived from mass conservation (per unitarea):

(6.22)

The average cation exchange capacity CEChas to be calculated in such a way that thetotal number of exchange sites (per unitarea) is given by z × r × CEC. This implies thefollowing formula for the profile averagecation exchange capacity:

(6.23)

And for the profile average base saturationEBC one then gets:

(6.24)

Note that for aquatic ecosystems, theseparameters have to be averaged over theterrestrial catchment area as well.

6.4.1.2 Data also used for critical load calculations

In this section we describe those input datawhich are also used in critical load calcula-tions (see Chapter 5) and for details thereader is referred to that Chapter, especially

Figure 6.3: Illustration of the basic composition of a soil profile: soil water, organic matter (organic carbon) and themineral soil, characterised by its clay, silt and sand fraction (clay+silt+sand=100%).

sections 5.3.1.3 and 5.3.2.3. Whereas forcritical loads and exceedance calculationsdata are needed at a specific point in time (orat steady-state), their past and future temporal development is needed for dynamic modelling.

Deposition:Non-anthropogenic (steady-state) basecation and chloride deposition are incorpo-rated into the definition of the critical load ofacidity. For dynamic models times series ofpast and future depositions are needed.However, at present there are no projectionsavailable for these elements on a Europeanscale. Thus in most model applications(average) present base cation and chloridedepositions are assumed to hold also in thefuture (and past).

Sulphur and nitrogen depositions enter onlythe exceedance calculations of criticalloads. In contrast, their temporal develop-ment is the driving force of every dynamicmodel. Time series for the period 1880-1990of S and N deposition on the EMEP-150 gridhave been recently computed using published estimates of historical emissions(Schöpp et al. 2003) and 12-year averagetransfer matrices derived from theEMEP/MSC-W lagrangian atmospherictransport model. Scenarios for future sul-phur and nitrogen deposition are providedby integrated assessment modellers, basedon atmospheric transport modelling byEMEP.

In case the deposition model provides onlygrid average values, a local deposition(adjustment) model could compute the localdeposition from the grid average values,especially for forest soils, where the actual(larger) deposition depends on the type andage of trees (via the ‘filtering’ of depositionby the canopy). An example of such a modelis the MAKEDEP model, which is also part ofthe SAFE model system (Alveteg andSverdrup 2002).

Uptake:Long-term average values of the net growth-uptake of nitrogen and base cations byforests are also needed to calculate critical

loads; and data sources and calculation procedures are given in Chapter 5. In simpledynamic models these processes aredescribed as a function of actual and projected forest growth. To this end, additional information on forest age andgrowth rates is needed, and the amount ofdata needed depends largely on whether thefull nutrient cycle is modelled or whetheronly net sources and sinks are considered.

Considering net removal by forest growth, asin the VSD, SMART and MAGIC models, theyield (forest growth) at a certain age can bederived from yield tables for the consideredtree species. The element contents in stems(and possibly branches) should be the sameas used in the critical load calculations (see,e.g., Table 5.2). If the nutrient cycle is modelled, as in the SAFE model, data areneeded on litterfall rates, root turnover rates,including the nutrient contents in litter(leaves/needles falling from the tree), andfine roots. Such data are highly dependenton tree species and site conditions.Compilations of such data can be found inDe Vries et al. (1990) and Jacobsen et al. (2002).

Water flux and soil moisture:Water flux data that are needed in one-layermodels are limited to the precipitation surplus leaving the root zone (see Chapter5), whereas multi-layer models require waterfluxes for each soil layer down to the bottomof the root zone. For simple dynamic models, water fluxes could be calculated bya separate hydrological model, running on adaily or monthly time step with aggregationto annual values afterwards. An example ofsuch a model is WATBAL (Starr 1999), whichis a capacity-type water balance model forforested stands/plots running on a monthlytime step and based on the following waterbalance equation:

(6.25)

where: Q = precipitation surplus, P = precip-itation, ET= evapotranspiration and ±DSM =changes in soil moisture content. WATBAL

6 Dynamic Modelling


6 Dynamic Modelling


uses relatively simple input data, which iseither directly available (e.g., monthly precip-itation and air temperature) or which can bederived from other data using transfer functions (e.g., soil available water capacity).

In any dynamic model, which includes amass balance for elements, also informationon the soil moisture content is needed. Thisis also output from a hydrological model (seeSM in eq. 6.25) or has to be estimated fromother site properties. An approximate annualaverage soil moisture content θ (m3/m3) canbe obtained as a function of the clay content(see Brady 1974):

(6.26)

i.e., for clay contents above 30% a constantvalue of 0.27 is assumed. It should be notedthat most models are quite insensitive to thevalue of θ.

Base cation weathering:The various possibilities to assess weather-ing rates of base cations, which are a keyinput also to critical load calculations, arediscussed and cited in Chapter 5.

Mineralisation and (de-)nitrification:Rate constants (and possibly additionalparameters) for mineralisation, nitrificationand denitrification are needed in detailedmodels, but simple models mostly use fac-tors between zero and one, which computenitrification and denitrification as fraction ofthe (net) input of nitrogen.

As in the calculation of critical loads (SMBmodel), in the VSD model complete nitrifica-tion is assumed (nitrification fraction equals1.0), and denitrification fractions can befound in Table 5.3. Mineralisation is not considered explicitly, but included in the netimmobilisation calculations.

For the complete soil profile, the nitrificationfraction in forest soils varies mostly between0.75 and 1.0. This is based on measure-ments of NH4/NO3 ratios below the rootzoneof highly acidic Dutch forests with very high

NH4 inputs in the early nineties, which werenearly always less than 0.25 (De Vries et al.1995). Generally, 50% of the NH4 input isnitrified above the mineral soil in the humuslayer (Tietema et al. 1990). Actually, the nitrification fraction includes the effect ofboth nitrification and preferential ammoniumuptake.

Al-H equilibrium:The constants needed to quantify the equilibrium between [Al] and [H] in the soilsolution are discussed and presented inChapter 5 (Table 5.9), since they are alsoneeded for critical load calculations. Formodels including the complexation of Al withorganic anions, such as MAGIC and SMART,relevant parameters can be found, e.g., inDriscoll et al. (1994).

6.4.1.3 Data needed to simulate cationexchange

In all dynamic soil models, cation exchangeis a crucial process (see section 6.2.3.1).Data needed to allow exchange calculationsare:• The pool of exchangeable cations, being

the product of layer thickness, bulk density, cation exchange capacity (CEC)and exchangeable cation fractions

• Cation exchange constants (selectivity coefficients)

Preferably these data are taken from meas-urements. Such measurements are generallymade for several soil horizons. For single-layer models, such as VSD, these data haveto be properly averaged over the entire soildepth (rooting zone; see eqs. 6.21 - 6.24).

In the absence of measurements, the variousdata needed to derive the pool of exchange-able cations for major forest soil types canbe derived by extrapolation of point data,using transfer functions between bulk density, CEC and base saturation and basicland and soil characteristics, such as soiltype, soil horizon, organic matter content,soil texture, etc.

Soil bulk density:If no measurements are available, the soilbulk density r (g/cm3) can be estimated fromthe following transfer function:

(6.27)

where Corg is the organic carbon content andclay the clay content (both in %). The topequation for mineral soils is based on databy Hoekstra and Poelman (1982), the bottomequation for peat(y) soils is derived from VanWallenburg (1988) and the central equationis a linear interpolation (for clay=0) betweenthe two (Reinds et al. 2001).

Cation exchange capacity (CEC):The value of the CEC depends on the soil pHat which the measurements are made.Consequently, there is a difference betweenunbuffered CEC values, measured at theactual soil pH and buffered values measuredat a standard pH, such as 6.5 or 8.2. In theVSD (and many other) models the exchangeconstants are related to a CEC that is measured in a buffered solution in order tostandardise to a single pH value (e.g.,

pH=6.5, as upper limit of non-calcareoussoils). The actual CEC can be calculatedfrom pH, clay and organic carbon contentaccording to (after Helling et al. 1964):

(6.28)

where CEC is the cation exchange capacity(meq/kg), clay is the clay content (%) andCorg the organic carbon content (%). The pHin this equation should be as close as possible to the measured soil solution pH.For sandy soils the clay content can be setto zero in eq. 6.28. Typical average clay contents as a function of the texture class,presented on the FAO soil classification(FAO 1981), are given in Table 6.2. Values forCorg range from 0.1% for arenosols (Qc) to50% for peat soils (Od).

6 Dynamic Modelling


Table 6-2: Average clay contents and typical base saturation as a function of soil texture classes (see Table 5-6).

Texture

class

Name Definition average

clay content (%)

typical

base saturation (%)

1 coarse clay < 18% and sand ≥ 65% 6 5

2 medium clay < 35% and sand ≥ 15%;

but clay ≥ 18% if sand ≥ 65%

20 15

3 medium fine clay < 35% and sand < 15% 20 20

4 fine 35% ≤ clay < 60% 45 50

5 very fine clay ≥ 60% 75 50

9 organic soils Soil types O 5 10-70

Computing CEC(pHmeasured), i.e. the CEC from eq.6.28, using measured (site-specific) Corg, 0

clay and pH does not always match the measured CEC, CECmeasured, and thus computing CEC

at pH=6.5, CEC(6.5), would not be consistent with it. Nevertheless, eq.6.28 can be used to

scale the measured CEC to a value at pH=6.5, i.e. the value needed for modelling, in the

following manner:

5

(6.29) )(

)5.6(

5.6measuredpHCEC

CECmeasuredpH CECCEC ⋅==

Table 6.2: Average clay contents and typical base saturation as a function of soil texture classes (see Table 5.12).

6 Dynamic Modelling


Computing CEC(pHmeasured), i.e. the CEC fromeq. 6.28, using measured (site-specific) Corg,clay and pH does not always match themeasured CEC, CECmeasured, and thus computing CEC at pH=6.5, CEC(6.5), wouldnot be consistent with it. Nevertheless, eq.6.28 can be used to scale the measured CECto a value at pH=6.5, i.e. the value needed formodelling, in the following manner:

(6.29)

This method of scaling measured data withthe ratio (or difference) of model output iswidely used in global change work to obtain,e.g., climate-changed (meteorological) dataconsistent with observations.

Exchangeable base cation fraction (basesaturation):In most models, a lumped expression isused for the exchange of cations, distinguishing only between H, Al and basecations (VSD, SMART and SAFE). As with theclay content, data for the exchangeablecation fractions, or in some cases only thebase saturation, can be based on information on national soil information systems, or in absence of these, on the FAOsoil map of Europe (FAO 1981). Base satura-tion data vary from 5-25% in relatively acidforest soils to more than 50% in wellbuffered soils. A very crude indication of thebase saturation as a function of the textureclass of soils is given in the last column ofTable 6.2. This relationship is based on datafrom forest soils given in FAO (1981) and inGardiner (1987). A higher texture classreflects a higher clay content implying anincrease in weathering rate, which implies ahigher base saturation. For organic soils thebase saturation is put equal to 70% for eutrichistosols (Oe) and 10% for dystric histosols(Od).

Ideally, only measured CEC and exchange-able cation data are used. However, whendata on the initial base saturation of soils arenot available for regional (national) modelapplications, one may derive them from arelationship with environmental factors.

Such an exercise was carried out using aEuropean database with approximately 5300soil chemistry data for the organic layer andthe forest topsoil (0-20 cm) collected on asystematic 16×16 km2 grid (ICP Forest level-I grid; Vanmechelen et al. 1997). Theregression relationship for the estimatedbase saturation EBC (expressed as a fractionwith values between 0 and 1) is:

(6.30a)

with

(6.30b)

where ‘ln’ is the natural logarithm,lt(x)=ln(x/(1–x)), and the ak’s are the regres-sion coefficients. The regression analysiswas carried out using a so-called Select-pro-cedure. This procedure combines qualitativepredictor variables, such as tree speciesand/or soil type, with quantitative variablesand it combines forward selection, startingwith a model including one predictor vari-able, and backward elimination, starting witha model including all predictor variables. The‘best’ model was based on a combination ofthe percentage of explained variance, thatshould be high and the number of predictorvariables that should be low. More informa-tion on the procedure is given in Klap et al.(2004). Results of the analyses are given inTable 6.3. The explained variance for basesaturation was approx. 45%.

Note: When data are not available, one mayalso calculate base saturation as the maxi-mum of (i) a relation with environmental fac-tors as given above and (ii) an equilibriumwith present deposition levels of SO4, NO3,NH4 and BC. Especially in southern Europe,where acid deposition is relatively low andbase cation input is high, the base saturationin equilibrium with the present load can behigher than the value computed according toeq. 6.30.

Exchange constants (selectivity coeffi-cients):In many exchange models the cations arelumped to H, Al and base cations (as in VSD,SMART and SAFE), but in MAGIC every basecation (Ca, Mg, K, Na) is modelled separately.Furthermore, cation exchange in SMARTand MAGIC is based upon Gaines-Thomasequations, in SAFE it is described by Gaponexchange reactions, whereas in the VSDmodel the user can chose between the two.Exchange constants can be derived from thesimultaneous measurement of the majorcations (H, Al, Ca, Mg, K and Na) at theadsorption complex and in the soil solution.

Using more than 800 such measurementsfrom Dutch soils, extensive tables of

exchange constants have been derived forsand, loess, clay and peat soils, togetherwith their standard deviations and correla-tions for all combinations of H, Al and basecations (De Vries and Posch 2003). The datashow the high affinity of the complex for protons compared to all other monovalentcations, and that the relative contributions ofK, Na and NH4 on the adsorption complexare very low. Results for the logarithms(log10) of the exchange constants used in theVSD model, both for the ‘Gaines-Thomasmode’ and the ‘Gapon mode’, together withtheir standard deviation (‘stddev’) are givenin Tables 6.4 to 6.7. For a conversion to otherunits see Annex III.

6 Dynamic Modelling


Predictor variable Base saturation

(mineral topsoil)

C/N-ratio

(organic layer)

C/N-ratio

(mineral topsoil)

Coefficients in

eqs.6.30 and 6.31

Constant 3.198 3.115 1.310 a0

Soil group:

sandy soils 0 0 0 a1

loamy/clayey soils 0.297 -0.807 -0.279 a1

peat soils 0.534 -0.025 -0.312 a1

Tree species:

pine 0 0 0 a2

spruce -0.113 -0.158 -0.093 a2

oak 0.856 -0.265 -0.218 a2

beech 0.591 -0.301 -0.218 a2

Site conditions:

Altitude [m] -0.00014 -0.00008 -0.000136 a3

Age [yr] 0 0.025 0.096 a4

Meteorology:

Temperature [°C] 0 -0.0078 -0.041 a5

Temperature2 [°C

2] 0 0.00095 0.0014 a6

Precipitation [mm/yr] 0 0.178 0.194 a7

Deposition:

Na [eq/ha/yr] -0.223 0 0.080 a8

N-tot (=NOy+NHz) [eq/ha/yr]:

sandy soils 0 -0.150 -0.019 a9

loamy/clayey soils 0 -0.032 0 a9

peat soils 0 -0.136 0 a9

Acid (=SOx*+N-tot) [eq/ha/yr] -1.025 0 0 a10

Bc* (=Ca

*+Mg

*+K

*) [eq/ha/yr] 0.676 0 0 a11

Deposition fractions:

NHz / Acid [-]:

sandy soils 0 0 0 a12

loamy/clayey soils -0.494 0 0 a12

peat soils -0.896 0 0 a12

NHz /N-tot [-] 0 0.102 0.120 a13

Ca*/Bc

* [-] 1.211 0 0 a14

Mg*/Bc

* [-] 0.567 0 0 a15

Table 6.3: Coefficients for estimating base saturation and the C/N-ratio in the mineral topsoil (0-20 cm) and theorganic layer (after Klap et al. 2004; Note: (a) the star denotes sea-salt corrected depositions, (b) depositions<0.1should be set to 0.1 to avoid underflow in the equations).

6 Dynamic Modelling


It should be noted that exchange constantsvary widely and are unknown for most sites.Therefore, in most models (SAFE, MAGIC,but also VSD) they are calibrated againstmeasurements of base saturation (and soilsolution concentrations).

6.4.1.4 Data needed for balances of nitrogen, sulphate and aluminium

C/N-ratio:Data for the C/N-ratio generally vary between15 in rich soils where humification has beenhigh to 40 in soils with low N inputs and lesshumification. Values can also be obtainedfrom results of a regression analysis similarto that of the base saturation according to:

Table 6-4: Mean and standard deviation of logarithmic Gaines-Thomas exchange constants of H against

Ca+Mg+K as a function of soil depth for sand, loess, clay and peat soils (mol/l)–1

.

Layer Sand Loess Clay Peat

(cm) Mean stddev Mean stddev Mean stddev Mean stddev

0-10 5.338 0.759 5.322 0.692 6.740 1.464 4.754 0.502

10-30 6.060 0.729 5.434 0.620 6.007 0.740 4.685 0.573

30-60 6.297 0.656 - - 6.754 0.344 5.307 1.051

60-100 6.204 0.242 5.541 0.579 7.185 - 5.386 1.636

0-30 5.236 0.614 5.386 0.606 6.728 1.373 4.615 0.439

0-60 5.863 0.495 - - 6.887 1.423 4.651 0.562

Table 6.4: Mean and standard deviation of logarithmic Gaines-Thomas exchange constants of H against Ca+Mg+Kas a function of soil depth for sand, loess, clay and peat soils (mol/l)–1.

Table 6-5: Mean and standard deviation of logarithmic Gaines-Thomas exchange constants of Al against 25 Ca+Mg+K as a function of soil depth for sand, loess, clay and peat soils (mol/l).



0-10 2.269 1.493 1.021 1.147 1.280 1.845 0.835 1.204

10-30 3.914 1.607 1.257 0.939 -0.680 1.152 0.703 0.968

30-60 4.175 1.969 - - -3.070 0.298 0.567 1.474

60-100 2.988 0.763 1.652 1.082 -2.860 - 0.969 1.777

0-30 2.306 1.082 0.878 1.079 0.391 1.555 0.978 0.805

0-60 2.858 1.121 - - -0.973 1.230 0.666 0.846

Table 6.5: Mean and standard deviation of logarithmic Gaines-Thomas exchange constants of Al against Ca+Mg+Kas a function of soil depth for sand, loess, clay and peat soils (mol/l).

Table 6-6: Mean and standard deviation of logarithmic Gapon exchange constants of H against Ca+Mg+K a

function of soil depth for sand, loess, clay and peat soils (mol/l)–1/2

.



0-10 3.178 0.309 3.138 0.268 3.684 0.568 2.818 0.199

10-30 3.527 0.271 3.240 0.221 3.287 0.282 2.739 0.175

30-60 3.662 0.334 - - 3.521 0.212 2.944 0.382

60-100 3.866 0.125 3.232 0.251 3.676 - 3.027 0.672

0-30 3.253 0.311 3.170 0.206 3.620 0.530 2.773 0.190

0-60 3.289 0.340 - - 3.604 0.654 2.694 0.170

Table 6.6: Mean and standard deviation of logarithmic Gapon exchange constants of H against Ca+Mg+K as a function of soil depth for sand, loess, clay and peat soils (mol/l)–1/2.

Table 6-7: Mean and standard deviation of logarithmic Gapon exchange constants of Al against Ca+Mg+K 5 function of soil depth for sand, loess, clay and peat soils (mol/l)

1/6.



0-10 0.306 0.440 0.190 0.546 -0.312 0.738 -0.373 0.350

10-30 0.693 0.517 0.382 0.663 -0.463 0.431 -0.444 0.255

30-60 0.819 0.527 - - -1.476 0.093 -0.740 0.336

60-100 1.114 0.121 0.390 0.591 -1.795 - -0.867 0.401

0-30 0.607 0.472 0.221 0.647 -0.609 0.731 -0.247 0.404

0-60 0.199 0.633 - - -1.054 0.362 -0.551 0.210

Table 6.7: Mean and standard deviation of logarithmic Gapon exchange constants of Al against Ca+Mg+K as a function of soil depth for sand, loess, clay and peat soils (mol/l)1/6.

(6.31)

where ‘ln’ is the natural logarithm and lt(x) = ln(x/(1–x)). Results of the analysis,which was performed with the same datasets as described in the section on base saturation, are given in Table 6.3; and moreinformation on the procedure is given in Klapet al. (2004).

Sulphate sorption capacity and half-satura-tion constant:Values for the maximum sorption capacityfor sulphate, Smax, can be related to the content of oxalate extractable Al (meq kg-1)according to (Johnson and Todd 1983):

(6.32)

Estimates for the oxalate extractable Alcontent are given below. Adsorption or half-saturation constants for sulphate, S1/2,can be derived from literature information(e.g. Singh and Johnson 1986, Foster et al.1986). A reasonable average value is 1.0eq/m3.

Al-hydroxide content:Data for the oxalate extractable Al content(the content of readily available Al-hydrox-ide) are often available in national soil information systems, such as the soil information system of the Netherlands. Insandy soils the Al-hydroxide content (inmeq/kg) mostly varies between 100-200 forA-horizons, between 200-350 for B-horizonsand between 50-150 for C-horizons (parentmaterial, De Vries 1991).

6.4.2 Model calibration

If all input parameters, initial conditions anddriving forces were known, the chosenmodel would describe the future develop

ment of the soil chemical status for anygiven deposition scenario. However, in mostcases several of the parameters are poorlyknown, and thus many models, i.e. the badlyknown parameters in the model, have to be‘calibrated’. The method of calibration varieswith the model and/or the application.

In standard applications of both the MAGICand SAFE model it is assumed that in pre-acidification times (say 1850) the input ofions is in equilibrium (steady state) with thesoil (solution) chemistry. Furthermore it isassumed that the deposition history of all(eight) ions is known (properly recon-structed).

In SAFE, weathering rates and uptake/netremoval of N and base cations are computedwithin the model (see above). Only simulated(present) base saturation is matched withobservations (in every soil layer) by adjustingthe cation exchange selectivitycoefficient(s). Matching simulated andobserved soil solution concentrations is notpart of the standard calibration procedure.

The calibration of MAGIC is a sequentialprocess whereby firstly the input and outputof those ions assumed to act conservativelyin the catchment are balanced (usually onlyCl). Next, the anion concentrations in surfacewaters are matched by adjusting catchmentnet retention (of N) and soil adsorption (of S)if appropriate. Thirdly, the four individualmajor base cation concentration in thestream and on the soil solid phase(expressed as a percentage of cationexchange capacity) are matched by adjust-ing the cation exchange selectivity coefficients and the base cation weatheringrates. Finally, surface water pH, Al andorganic anion concentrations are matchedby adjusting the aluminium solubility

6 Dynamic Modelling


6 Dynamic Modelling


coefficient and total organic acid concen-tration in surface water.

Both in MAGIC and SAFE automatic calibra-tion routines are an important part of theoverall model system. For the SMART modelno such automatic model calibration routineis presently available. For the VSD model thesame calibration routine as used in SAFEhas been implemented, and details can befound in Posch and Reinds (2005) and in thehelp-file of the ‘VSD-Studio’ software (avail-able at www.rivm.nl/cce).

As stated above, the most demanding part isnot the actual running of a model, but thederivation and preparation of input data(files) and the model initialisation/calibration.However, especially for regional applica-tions, i.e. runs for many sites, additionalwork is required to embed the model – oftendesigned for single site applications – into asuitable (data base) framework which allowsthe efficient handling of model inputs andoutputs.

6.5.1 Use of dynamic models in inte-grated assessment

Most usefully, for the review of theGothenburg Protocol, a link should be established between the dynamic modelsand integrated assessment (models). In thefollowing several modes of interaction withintegrated assessment (IA) models are iden-tified.

Scenario analyses:Deposition scenario output from IA modelsare used by the ‘effects community’ (ICPs)as input to dynamic models to analyse theirimpact on (European) soils and surfacewaters, and the results (recovery times etc.)are reported back.

Presently available dynamic models are wellsuited for this task. The question is how tosummarise the resulting information on a

European scale. Also, the ‘turn-around time’of such an analysis, i.e. the time betweenobtaining deposition scenarios and reportingback dynamic model results, may be long,as it could involve the work of several subsidiary bodies under the LRTAPConvention.

Response functions:Response functions are pre-processeddynamic model runs for a large number ofplausible future deposition patterns fromwhich the results for every (reasonable) dep-osition scenario can be obtained by interpo-lation. Such response functions encapsulatea site’s temporal behaviour to reach a certain(chemical) state and linking them to IA models allows to evaluate the site’sresponse to a broad range of deposition patterns.

An example is shown in Figure 6.4: It showsthe isolines of years (‘recovery isochrones’)in which Al/Bc=1 is attained for the first timefor a given combination of percent deposi-tion reduction (vertical axis) and implemen-tation year (horizontal axis). The reductionsare expressed as percentage of the deposi-tion in 2010 after implementation of theGothenburg Protocol and the implementa-tion year refers to the full implementation ofthat additional reduction. For example, a44% reduction of the 2010 deposition, fullyimplemented by the year 2020 will result in a(chemical) recovery by the year 2040(dashed line in Figure 6.4). Note that for thisexample site no recovery is possible, unlessdeposition is reduced more than 18% of the2010 level.

6.5 Model Calculations and Presen-tation of Model Results

Considering how critical loads have beenused in IA during the negotiations of protocols, it is unlikely that there will be awide variation in the implementation year ofa new reduction agreement (generally 5-10years after a protocol enters into force).Thus, for a fixed implementation year thequestion will be: What is the maximum dep-osition allowed to achieve recovery, i.e.,reach (and sustain!) a desired chemical state(e.g. Al/Bc=1) in a prescribed year? Such a

deposition is called a target load and, in thecase of a single pollutant, target loads can,in principle, be read from information as presented in Figure 6.4. In Figure 6.5 targetloads (expressed as percent depositionreductions from the 2010 level) are shownexplicitly as function of the target year forthe (fixed) implementation year 2020. In thecase of a single pollutant, this is the type ofinformation to be linked to IA models.

6 Dynamic Modelling


2020

2030

2040

20502060

208021002150

2015 2025 2035 2045 2055 20650

10

20

30

40

50

60

70

80

90

100

Implementation year

% D

ep-r

educ

tion

beyo

nd G

-Pro

toco

l

No recovery

Figure 6.4: Example of ‘recovery isochrones’ for a single site. The vertical axis gives the additional reduction in acid-ifying deposition after the implementation of the Gothenburg Protocol in 2010 (expressed as percentage of the 2010level) and the horizontal axis the year at which this additional reductions are fully implemented. The isolines arelabelled with the first year at which Al/Bc=1 is attained for a given combination of percent reduction and implemen-tation year.

2020 2030 2040 2050 2060 2070 2080 2090 21000

10

20

30

40

50

60

70

80

Target year

% D

ep-r

educ

tion

beyo

nd G

-Pro

toco

l

No recovery

Implementation year: 2020

Figure 6.5: Required deposition reductions (target loads) for a site as a function of the target year, i.e. the year inwhich recovery is achieved (see also Figure 6.4). The implementation year of the reductions is 2020. Note that forreductions above 74% the recovery happens already before the implementation year.

6 Dynamic Modelling


However, in the case of acidity, both N and Sdeposition determine the soil chemical stateand it will not be possible to obtain uniquepairs of N and S deposition to reach a prescribed target (similar to critical loadfunctions for acidity). Thus, target loadsfunctions have to be derived with dynamicmodels for a series of target years andagreed upon implementation years. Thesetarget load functions, or suitable statisticsderived from them, are passed on to the IAmodellers who evaluate their feasibility ofachievement (in terms of costs and techno-logical abatement options available).

The determination of response functions,such a target loads, requires no changes toexisting models per se, but rather additionalwork, since dynamic soil model have to runmany times and/or ‘backwards’, i.e. in aniterative mode. A further discussion of theproblems and possible pitfalls in the compu-tation of target loads is provided in the nextsection (see also Jenkins et al. 2003).

Integrated dynamic model:The ‘most intimate’ link would be the inte-gration of a dynamic model into an IA model(e.g. RAINS). In this way it could be an integral part of all scenario analyses andoptimisation runs. Widely used models, suchas MAGIC, SAFE and SMART, are not easily

incorporated into IA models, and they mightbe still too complex to be used in optimisa-tion runs. Alternatively, a very simple dynam-ic model could be incorporated into an IAmodel, capturing the essential, long-termfeatures of dynamic soil models. This wouldbe comparable to the process that led to thesimple ozone model included in RAINS,which was derived from the complex photo-oxidant model of EMEP. However, even thiswould require a major effort, not the least ofwhich is the creation of a European data-base to run the model.

6.5.2 Target load calculations

As outlined above, target loads, or targetload functions in the case of acidification,are a way to link dynamic models with inte-grated assessment models, not least due totheir similarity with critical load functions(see Chapter 5). If a target load exists, thereexists also an infinite variety of depositionpaths to reach that target load. To bringorder into this multitude and to make resultscomparable, we define a target load as adeposition path characterised by three num-bers (years): (i) the protocol year, (ii) theimplementation year, and (iii) the target year(see Figure 6.6). If needed, these terms areproceeded by the term ‘dynamic modelling’(‘DM’) to distinguish them from similar termsused in IA circles.

Figure 6.6: Deposition paths for calculating target loads by dynamic models (DM) are characterised by three keyyears. (i) The year up to which the (historic) deposition is fixed (protocol year); (ii) the year in which the emissionreductions leading to a target load are fully implemented (DM implementation year); and (iii) the years in which thechemical criterion is to be achieved (DM target years).

In contrast to scenario analyses, the compu-tation of target loads is not straightforward.After specifying the target year and the yearof implementation of the (yet unknown) target load, the dynamic model has to be runiteratively until the deposition (= target load)is found which is required to reach thedesired chemical status in the specified target year. The following examples demon-strate the different cases that can arise whencalculating target loads and what can happen when doing such calculations ‘blindly’. For simplicity we use a single pollutant (deposition), but the conclusionshold for target load functions as well.

As an example, Figure 6.7 shows the deposi-tion history (left) and the resulting molarAl/Bc-ratio (right) as simulated (by the VSD

model) for three different soils, solely distin-guished by their CEC (40, 60 and 80 meq/kg).In two cases the Al/Bc-ratio in the year 2010is above the critical value (=1), while forCEC=80 it stayed below it during the past. Toinvestigate the future behaviour of the soils,we let the deposition drop to the critical load(which is independent of the CEC) during the‘implementation period’ (marked by two vertical lines in Figure 6.7). Obviously, forCEC=80, the Al/Bc-ratio stays below one,whereas for CEC=60 it drops below one within the first decade and then slowly risesagain towards the critical value. For CEC=40,the Al/Bc-ratio stays well above the criticalvalue, approaching it asymptotically overtime. In all three cases the approach to thecritical value is very slow.

6 Dynamic Modelling


1970 1990 2010 2030 2050 20700

100

200

300

400

500

600

700

800

year

Dep

(eq

/ha/

yr)

CEC=40

CEC=60

CEC=80

1970 1990 2010 2030 2050 20700

0.25

0.50

0.75

1.00

1.25

1.50

1.75

2.00

year

Al/B

c (m

ol/m

ol)

Figure 6.7: Temporal development of acidifying deposition (left) and corresponding molar Al/Bc-ratio (right) for 3 soils varying in CEC. The two vertical lines separate 50 years of ‘history’, 10 years (2010-2020) of implementation,and the future. Also shown are the critical load and the critical value (Al/Bc)crit=1 as thin horizontal lines. The deposition drops to the critical load within the implementation period and the Al/Bc-ratios (slowly) approach the critical value.

Next we look at target load calculations forthese three soils. Figure 6.8 shows theresults of target load calculations for 40 years, i.e. achieving (Al/Bc)crit=1 in theyear 2050. For CEC=40 meq/kg the targetload is smaller than the critical load, as onewould expect. For CEC=60 and 80, however,the computed target loads are higher thanthe critical load. As Figure 6.8 illustrates, this

does not make sense: After reaching the critical limit, these two soils deteriorate andthe Al/Bc-ratio gets larger and larger. Sincetarget loads are supposed to protect alsoafter the target year, we stipulate that when-ever a calculated target load is higher thanthe critical load, it has to be set equal to thecritical load.

6 Dynamic Modelling


In the light of the above considerations wedefine that a target load is the depositionfor which a pre-defined chemical or bio-logical status is reached in the targetyear and maintained (or improved) there-after.

In view of this, the steps to be considered forcalculating a target load are shown in theflow chart in Figure 6.9. The first check at

every site is, whether the critical load (CL) isexceeded in the reference year (2010 in ourcase). If the answer is ‘yes’ (as for the soilswith CEC=40 and 60 in Figure 6.7), the nextstep is to run the dynamic model with thedeposition equal to the critical load. If in thetarget year the chemical criterion is nolonger violated (e.g. Al/Bc£1), the target loadequals the critical load (TL=CL).

CEC=40

CEC=60

CEC=80

1970 1990 2010 2030 2050 20700

100

200

300

400

500

600

700

800

year

Dep

(eq

/ha/

yr)

CEC=40

CEC=60

CEC=80

1970 1990 2010 2030 2050 20700

0.25

0.50

0.75

1.00

1.25

1.50

1.75

2.00

year

Al/B

c (m

ol/m

ol)

Figure 6.8: Target loads (with 2050 as target year) for three soils and the resulting Al/Bc-ratio (left). Note that forCEC=60 and 80 the target load is higher than the critical load, even when (Al/Bc)crit<1 at present (for CEC=80)! Clearly,in such cases target load calculations don’t make sense.

Exceedance of CL in 2010?

Yes No

No

Yes NoRun model with "zero"deposition until target year

Run model with 2010deposition until target year

Al/Bc>1 in target year?

Al/Bc>1 in 2010?


Compute TL

Yes

Target cannot be achieved No deposition reductions needed

No

Run model with CLuntil target year


TL = CL

Yes

NoYes

Figure 6.9: Flow chart of the procedure to calculate a target load, avoiding the pitfalls mentioned in the text (e.g.computing a target load that allows violation of the criterion after the target year).

If, after running the model with the criticalload as deposition, the criterion is still violated in the target year, the model has tobe run with “zero” deposition until the specified target year. “Zero” depositionmeans a deposition small enough as not tocontribute to acidification (or eutrophica-tion). In the case of nitrogen this would meanthat Ndep is set equal to CLmin(N), thus avoiding problems, e.g. negatively influenc-ing forest growth in case of zero N deposi-tion.

If, after running the model with “zero” deposition, the criterion is still violated in thetarget year, then the target cannot be met inthat year. In such a case recovery can onlybe achieved in a later year. Otherwise, a target load exists and has to be calculated;its value lies somewhere between zero andthe critical load.

If the critical load is not (or no longer)exceeded in 2010 (as for the soil withCEC=80 in Figure 6.7), this does not meanthat the risk of damage to the ecosystem isalready averted – it only means that eventu-ally, maybe after a very long time, the chem-ical criterion is no longer violated. Only if, inaddition, the chemical criterion is not violat-ed in 2010, no further emission reductionsare required for that ecosystem. Also, if themodel is run with the 2010 deposition untilthe target year and the criterion is no longerviolated in that year, no further emissionreductions are required.

In the implementation of the above procedure one could skip the step in whichthe model is run with the critical load as deposition (in case of exceedance in 2010)and immediately start with target load calculations (if a target load exists). And onlyafterwards check if this target load is greaterthan the CL (and set it equal to the CL) (seethe soil with CEC=60 in Figure 6.8). However,in view of the fact that TL calculationsrequire iterative model runs, and also toavoid surprises due to round-off errors, itmakes good sense to include that inter-mediate step.

An issue requiring attention in all target loadcalculations is the assumptions about finitenitrogen buffers. If it is, e.g., assumed that asoil can immobilise N for (say) the next 50years more than assumed in the critical loadcalculations, then target loads can be higherthan the critical load. This might cause confusion and demands careful explana-tions.

The above considerations hold also in thecase of two pollutants, such as S and N in thecase of acidification. The results is then nota single value for a target load, but a so-called target load function consisting ofall pairs of deposition (Ndep, Sdep) for whichthe target is achieved in the selected year.This concept is very similar to the criticalload function (see Chapter 5). In Figure 6.10examples of target load functions are shownor a set of target years.

6 Dynamic Modelling


6 Dynamic Modelling


When summarising target load calculationsfor ecosystems in a grid square (or region) itis important not only to report sites for whichtarget load (functions) could been derived,but all cases (and their areas), i.e. the sitesfor which (i) no further deposition reductionsare necessary, (ii) a target load has been cal-culated, and (iii) no target load exists (for thegiven target year). Note that in case (i), the2010 deposition has necessarily to be below(or equal to) the critical load.

6.5.3 Presentation of model results

For single site applications of dynamic mod-els the obvious way to present model outputare graphs of the temporal development of

the most relevant soil chemical variables,such as base saturation or the concentra-tions of ions in the soil solution (e.g. Al/Bc-ratio), in response to given depositionscenarios. In regional (European) applica-tions, however, this kind of information hasto be summarised. This can be done in several ways, e.g., by displaying the tem-poral development of selected percentiles ofthe distribution of the variable(s) of interest(see Figure 6.11). Another way is to show asequence of maps displaying the variable ofinterest in (say) five-year intervals (‘mapmovies’). These and other options are dis-cussed and illustrated in Evans et al. (2001),Jenkins et al. (2002) and Moldan et al. (2003).

Figure 6.10: Example of target load functions for a site for five different target years. Also shown is the critical loadfunction of the site (dashed line). Note that any meaningful target load function has to lie below the critical load function, i.e. require stricter deposition reductions than achieving critical loads.

Figure 6.11: Example of percentile traces of a regional dynamic model output. From it seven percentiles (5, 10, 25,50, 75, 90 and 95%) can be read for every time step.

Maps can represent single sites only if theirnumber does not become too large. If thenumber of sites reaches the thousands, statistical descriptors (means, percentiles)have to be used to represent the model output. For example, for a given target yearthe percentage of ecosystems in a gridsquare for which the target is achievedunder a given deposition scenario can bedisplayed in a map format, very much in thesame way as protection percentages(derived from protection isolines) have beendisplayed for critical load exceedances.Procedures for calculating percentiles and‘target load’ isolines can be found in Chapter 8.

Allott TEH, Battarbee RW, Curtis C, KreiserAM, Juggins S, Harriman R (1995) Anempirical model of critical acidity loadsfor surface waters based on palaeolimno-logical data. In: Hornung M, Sutton MA,Wilson RB (eds) Mapping and Modellingof Critical Loads for Nitrogen: AWorkshop Report. Institute of TerrestrialEcology, Penicuik, United Kingdom,pp.50-54.

Alveteg M (1998) Dynamics of forest soilchemistry. PhD thesis, Reports in Ecologyand Environmental Engineering 3:1998,Department of Chemical Engineering II,Lund University, Lund, Sweden, 81pp.+appendices.

Alveteg M, Sverdrup H, Kurz D (1998)Integrated assessment of soil chemicalstatus. 1. Integration of existing modelsand derivation of a regional database forSwitzerland. Water, Air and Soil Pollution105: 1-9.

Alveteg M, Sverdrup H (2002) Manual forregional assessments using the SAFEmodel (draft version 8 April 2002).Department of Chemical Engineering II,Lund University, Lund, Sweden. See alsowww2.chemeng.lth.se

Battarbee RW, Allott TEH, Juggins S, KreiserAM, Curtis C, Harriman R (1996) Criticalloads of acidity to surface waters – anempirical diatom-based palaeolimnologi-cal model. Ambio 25: 366-369.

Berendse F, Beltman B, Bobbink R, KwantM, Schmitz MB (1987) Primary productionand nutrient availability in wet heathlandecosystems. Acta Oec./Oecol. Plant. 8:265-276.

Berendse F (1988) The nutrient balance ofvegetation on dry sandy soils in the con-text of eutrophication via the air. Part 1: Asimulation model as an aid for the man-agement of wet heathlands (in Dutch).Centre for Agrobiological Research,Wageningen, The Netherlands, 51 pp.

Brady NC (1974) The Nature and Properties

of Soils (8th edition). MacMillan, NewYork, 693 pp.

Cole JJ, Caraco NF, Kling GW, Kratz TK(1994) Carbon dioxide supersaturation inthe surface waters of lakes. Science 265:1568-1570.

Cosby BJ, Hornberger GM, Galloway JN,Wright RF (1985a) Modeling the effects ofacid deposition: Assessment of a lumpedparameter model of soil water andstreamwater chemistry. Water ResourcesResearch 21(1): 51-63.

Cosby BJ, Wright RF, Hornberger GM,Galloway JN (1985b) Modeling the effectsof acid deposition: Estimation of long-term water quality responses in a smallforested catchment. Water ResourcesResearch 21(11): 1591-1601.

Cosby BJ, Hornberger GM, Galloway JN,Wright RF (1985c) Time scales of catch-ment acidification: A quantitative modelfor estimating freshwater acidification.Environmental Science & Technology19:1144-1149.

Cosby BJ, Hornberger GM, Wright RF,Galloway JN (1986) Modeling the effectsof acid deposition: Control of long-termsulfate dynamics by soil sulfate adsorp-tion. Water Resources Research 22(8):1283-1291.

6 Dynamic Modelling


References

6 Dynamic Modelling


Cosby BJ, Ferrier RC, Jenkins A, Wright RF(2001) Modelling the effects of acid depo-sition: refinements, adjustments andinclusion of nitrogen dynamics in theMAGIC model. Hydrology and EarthSystem Sciences 5(3): 499-517.

De Vries W, Posch M, Kämäri J (1989)Simulation of the long-term soil responseto acid deposition in various bufferranges. Water, Air and Soil Pollution 48:349-390.

De Vries W, Hol A, Tjalma S, Voogd JC (1990)Stores and residence times of elements ina forest ecosystem: a literature study (inDutch). DLO-Staring Centrum, Rapport94, Wageningen, The Netherlands, 205pp.

De Vries W (1991) Methodologies for theassessment and mapping of critical loadsand the impact of abatement strategieson forest soils. DLO Winand StaringCentre for Integrated Land, Soil andWater Research, Report 46, Wageningen,The Netherlands, 109 pp.

De Vries W, Reinds GJ, Posch M, Kämäri J(1994) Simulation of soil response toacidic deposition scenarios in Europe.Water, Air and Soil Pollution 78: 215-246.

De Vries W (1994) Soil response to acid dep-osition at different regional scales. Fieldand laboratory data, critical loads andmodel predictions. PhD Thesis,Agricultural University, Wageningen, TheNetherlands, 487 pp.

De Vries W, Van Grinsven JJM, Van BreemenN, Leeters EEJM, Jansen PC (1995)Impacts of acid atmospheric depositionon concentrations and fluxes of solutes inDutch forest soils. Geoderma 67: 17-43.

De Vries W, Posch M (2003) Derivation ofcation exchange constants for sandloess, clay and peat soils on the basis offield measurements in the Netherlands.Alterra-rapport 701, Alterra Green WorldResearch, Wageningen, The Netherlands,50 pp.

Dise NB, Matzner E, Gundersen P (1998)Synthesis of nitrogen pools and fluxesfrom European forest ecosystems. Water,Air and Soil Pollution 105: 143-154.

Driscoll CT, Lehtinen MD, Sullivan TJ (1994)Modeling the acid-base chemistry oforganic solutes in Adirondack, New York,lakes. Water Resources Research 30:297-306.

Ellenberg H (1985) Veränderungen der FloraMitteleuropas unter dem Einfluss vonDüngung und Immissionen.Schweizerische Zeitschift für dasForstwesen 136: 19-39.

Evans C, Jenkins A, Helliwell R, Ferrier R,Collins R (2001) Freshwater Acidificationand Recovery in the United Kingdom.Centre for Ecology and Hydrology,Wallingford, United Kingdom, 80 pp.

FAO (1981) FAO Unesco Soil Map of theWorld, 1:5,000,000. Volume V Europe.Unesco, Paris 1981, 199 pp.

Friend AD, Stevens AK, Knox RG, ChannelMGR (1997) A process-based terrestrialbiosphere model of ecosystem dynamics(Hybrid v3.0). Ecological Modelling 95:249-287.

Foster NW, Morrison IK, Nicolson JA (1986)Acid deposition and ion leaching from apodzolic soil under hardwood forest.Water, Air and Soil Pollution 31: 879-889.

Gardiner MJ (1987) Representative data formajor soil units in the EEC soil map. AnForas Taluntais, Ireland. Internal Report,486 pp.

Gundersen P, Callesen I, De Vries W (1998)Nitrate leaching in forest ecosystems iscontrolled by forest floor C/N ratio.Environmental Pollution 102: 403-407.

Hann BJ, Turner MA (2000) Littoral micro-crustacea in Lake 302S in theExperimental Lakes Area of Canada: acid-ification and recovery. Freshwater Biology43: 133-146.

Heil GW, Bobbink R (1993) ‘CALLUNA’ a sim-ulation model for evaluation of impacts ofatmospheric nitrogen deposition on dryheathlands. Ecological Modelling 68: 161-182.

Helling CS, Chesters G, Corey RB (1964)Contribution of organic matter and clay tosoil cation exchange capacity as affectedby the pH of the saturating solution. Soil

Sci. Soc. Am. J. 28: 517-520.

Hoekstra C, Poelman JNB (1982) Density ofsoils measured at the most common soiltypes in the Netherlands (in Dutch).Report 1582, Soil Survey Institute,Wageningen, The Netherlands, 47 pp.

Jacobsen C, Rademacher P, Meesenburg H,Meiwes KJ (2002) Element contents intree compartments – Literature study anddata collection (in German). Report,Niedersächsische ForstlicheVersuchsanstalt, Göttingen, Germany, 80pp.

Jenkins A, Larssen T, Moldan F, Posch M,Wright RF (2002) Dynamic modelling ofsurface waters: Impact of emission reduc-tion – possibilities and limitations. ICP-Waters Report 70/2002, NorwegianInstitute for Water Research (NIVA), Oslo,Norway, 42 pp.

Jenkins A, Cosby BJ, Ferrier RC, Larssen T,Posch M (2003) Assessing emissionreduction targets with dynamic models:deriving target load functions for use inintegrated assessment. Hydrology andEarth System Sciences 7(4): 609-617.

Johnson DW, Todd DE (1983) Relationshipsamong iron, aluminium, carbon, and sul-fate in a variety of forest soils. Soil Sci.Soc. Am. J. 47: 792-800.

Keller W, Gunn JM (1995) Lake water qualityimprovements and recovering aquaticcommunities. In: Gunn JM (ed),Restoration and Recovery of an IndustrialRegion, Springer Verlag, New York, pp.67-80.

Klap JM, Brus DJ, De Vries W, Reinds GJ(2004) Assessment of site-specific esti-mates of critical deposition levels fornitrogen and acidity in European forestecosystems using measured and interpo-lated soil chemistry. (in prep).

Kros J, Reinds GJ, De Vries W, Latour JB,Bollen M (1995) Modelling of soil acidityand nitrogen availability in natural ecosys-tems in response to changes in acid dep-osition and hydrology. Report 95, DLOWinand Staring Centre, Wageningen, TheNetherlands, 90 pp.

Kurz D, Alveteg M, Sverdrup H, (1998)Integrated assessment of soil chemicalstatus. 2. Application of a regionalizedmodel to 622 forested sites inSwitzerland. Water, Air and Soil Pollution105: 11-20.

Latour JB, Reiling R (1993) A multiple stressmodel for vegetation (MOVE): a tool forscenario studies and standard setting.Science of the Total EnvironmentSupplement 93: 1513-1526.

Martinson L, Alveteg M, Warfvinge P (2003)Parameterization and evaluation of sul-fate adsorption in a dynamic soil chem-istry model. Environmental Pollution124(1): 119-125.

Mills KH, Chalanchuk SM, Allan DJ (2000)Recovery of fish populations in Lake 223from experimental acidification. Can. J.Fish. Aquat. Sci. 57: 192-204.

Moldan F, Beier C, Holmberg M, Kronnäs V,Larssen T, Wright RF (2003) Dynamicmodelling of soil and water acidification:Display and presentation of results forpolicy purposes. Acid Rain ResearchReport 56/03, Norwegian Institute forWater Research (NIVA), Oslo, Norway, 62pp.

Oja T, Yin X, Arp PA (1995) The forest model-ling series ForM-S: applications to theSolling spruce site. Ecological Modelling83: 207-217.

Posch M, Reinds GJ, De Vries W (1993)SMART – A Simulation Model forAcidification’s Regional Trends: Modeldescription and user manual.Mimeograph Series of the National Boardof Waters and the Environment 477,Helsinki, Finland, 43 pp.

Posch M, Hettelingh J-P, Slootweg J (eds)(2003) Manual for dynamic modelling ofsoil response to atmospheric deposition.RIVM Report 259101012, Bilthoven, TheNetherlands, 69 pp. See alsowww.rivm.nl/cce

Posch M, Reinds GJ (2005) VSD - UserManual of the Very Simple Dynamic soilacidification model. Coordination Centerfor Effects, RIVM, Bilthoven, TheNetherlands (in preparation).

6 Dynamic Modelling


6 Dynamic Modelling


Raddum GG (1999) Large scale monitoringof invertebrates: Aims, possibilities andacidification indexes. In: GG Raddum, BORosseland, J Bowman (eds) Workshop onBiological Assessment and Monitoring;Evaluation and Models. ICP-WatersReport 50/99, Norwegian Institute forWater Research, Oslo, Norway, pp.7-16.

Reinds GJ, Posch M, De Vries W (2001) Asemi-empirical dynamic soil acidificationmodel for use in spatially explicit integrat-ed assessment models for Europe. AlterraReport 084, Alterra Green WorldResearch, Wageningen, The Netherlands,55 pp.

Reuss JO (1980) Simulation of soil nutrientlosses due to rainfall acidity. EcologicalModelling 11: 15-38.

Reuss JO (1983) Implications of the calcium-aluminum exchange system for the effectof acid precipitation on soils. Journal ofEnvironmental Quality 12(4): 591-595.

Reuss JO, Johnson DW (1986) AcidDeposition and the Acidification of Soilsand Waters. Ecological Studies 59,Springer, New York, 119 pp.

SAEFL (1998) Acidification of Swiss forestsoils - Development of a regional dynam-ic assessment. EnvironmentalDocumentation No.89, (Swiss Agancy forthe Environment, Forest and Landscape),Berne, Switzerland, 115 pp

Schöpp W, Posch M, Mylona S, JohanssonM (2003) Long-term development of aciddeposition (1880-2030) in sensitive fresh-water regions in Europe. Hydrology andEarth System Sciences 7(4): 436-446.

Schouwenberg EPAG, Houweling H, JansenMJW, Kros J, Mol-Dijkstra JP (2000)Uncertainty propagation in model chains:a case study in nature conservancy.Alterra Report 001, Alterra Green WorldResearch, Wageningen, The Netherlands,90 pp.

Singh BR, Johnson DW (1986) Sulfate con-tent and adsorption in soils of two forest-ed watersheds in southern Norway. Water,Air and Soil Pollution 31: 847-856.

Snucins S, Gunn JM, Keller W, Dixit S, HindarA, Henriksen A (2001) Effects of regional

reductions in sulphur deposition on thechemical and biological recovery of lakeswithin Killarney Park, Ontario, Canada.Journal of Environmental Monitoring andAssessment 67: 179-194.

Starr M (1999) WATBAL: A model for estimat-ing monthly water balance components,including soil water fluxes. In: S Kleemola,

M Forsius (eds) 8th Annual Report,UNECE ICP Integrated Monitoring,Finnish Environment Institute, Helsinki,Finland. The Finnish Environment 325: 31-35.

Tietema A, Duysings JJHM, Verstraten JM,Westerveld JW (1990) Estimation of actu-al nitrification rates in an acid forest soil.In: AF Harrison, P Ineson, OW Heal (eds)Nutrient cycling in terrestrial ecosystems;Field methods, application and interpre-tation. Elsevier Applied Science, Londonand New York, pp.190-197.

Tiktak A, Van Grinsven JJM (1995) Review ofsixteen forest-soil-atmosphere models.Ecological Modelling 83: 35-53.

Ulrich B (1981) Ökologische Gruppierungvon Böden nach ihrem chemischenBodenzustand. Z. Pflanzenernähr.Bodenk. 144: 289-305.

Vanmechelen L, Groenemans R, Van Ranst E(1997) Forest soil condition in Europe.Results of a large-scale soil survey. EC-UN/ECE, Brussels, Geneva, 261 pp.

Van Oene H (1992) Acid deposition and for-est nutrient imbalances: a modellingapproach. Water, Air and Soil Pollution 63:33-50.

Van Wallenburg C (1988) The density ofpeaty soils (in Dutch). Internal Report, SoilSurvey Institute, Wageningen, TheNetherlands, 5 pp.

Wamelink GWW, Ter Braak CJF, Van DobbenHF (2003) Changes in large-scale patternsof plant biodiversity predicted from envi-ronmental economic scenarios.Landscape Ecology 18: 513-527.

Warfvinge P, Sverdrup H (1992) Calculatingcritical loads of acid deposition with PRO-FILE - A steady-state soil chemistrymodel. Water, Air and Soil Pollution 63:119-143.

Warfvinge P, Holmberg M, Posch M, WrightRF (1992) The use of dynamic models toset target loads. Ambio 21: 369-376.

Warfvinge P, Falkengren-Grerup U, SverdrupH, Andersen B (1993) Modelling long-termcation supply in acidified forest stands.Environmental Pollution 80: 209-221.

Wright RF, Lie MC (eds) (2002) Workshop onmodels for biological recovery from acidi-fication in a changing climate, 9-11September 2002 in Grimstad, Norway.Acid Rain Research Report 55/02,Norwegian Institute for Water Research(NIVA), Oslo, Norway, 42 pp.

6 Dynamic Modelling


In this Chapter the calculation ofexceedances, i.e., the comparison of criticalload/levels with depositions/concentrations,is described. In Section 7.1 the basic defini-tion of an exceedance is given, includingsome historical remarks on the origin anduse of the word 'exceedance'. In Section 7.2the concept of a conditional critical load of S and N is introduced, which allows to treatthese two acidifying pollutants separately,and thus also straight-forward exceedancecalculations. In Section 7.3 the exceedanceof critical loads of acidity is defined, whichinvolves two pollutants simultaneously, inSection 7.4 the exceedances of surfacewater critical loads are considered, and inSection 7.5 the relationship between deposi-tions and target loads are briefly touched.Most of the material presented here is takenfrom Posch et al. (1997, 1999).

The word 'exceedance' is defined as "theamount by which something, especially apollutant, exceeds a standard or permissiblemeasurement" (The American HeritageDictionary of the English Language, FourthEdition 2000) and is a generally acceptedterm within the air pollution discipline.Nevertheless, the term 'critical load excess'is preferred by some (English) speakers dueto the apparent coining of 'exceedance' during the development of the critical loadconcept. Interestingly, the Oxford EnglishDictionary (OED) database has an exampleof 'exceedance' from 1836 (Quinion 2004) -36 years before Robert Angus Smith is credited with coining of the term 'acid rain'(Smith 1872). However, the term 'acid rain'(in French) had already been used in 1845 byDucros in a scientific journal article (Ducros 1845).

Critical loads and levels are derived to char-acterise the vulnerability of ecosystem(parts, components) in terms of a depositionor concentration. If the critical load of pollu-tant X at a given location is smaller than thedeposition of X at that location, it is said that

the critical load is exceeded and the differ-ence is called exceedance. In mathematicalterms, the exceedance Ex of the critical loadCL(X) is given as:

(7.1)

where Xdep is the deposition of pollutant X. Inthe case of the critical level, the comparisonis with the respective concentration quantity.If the critical load is greater than or equal tothe deposition, one says that it is notexceeded or there is non-exceedance of thecritical load.

An exceedance defined by eq. 7.1 canobtain positive, negative or zero value. Sinceit is in most cases sufficient to know thatthere is non-exceedance, without beinginterested in the magnitude of non-exceedance, the exceedance can be alsodefined as:

(7.2)

An example of the application of this basicequation is the exceedance of the criticalload of nutrient N (see Chapter 5.3.1), whichis given by:

(7.3)

It should be noted that exceedances differfundamentally from critical loads, as they aretime-dependent. One can speak of thecritical load of X for an ecosystem, but not ofthe exceedance of it. For exceedances thetime for which they have been calculatedhas to be reported, since - especially in inte-grated assessment - it is exceedances dueto (past or future) anthropogenic depositionsthat are of interest.

Of course, the time-invariance of criticalloads and levels has its limitations, certainly

7 Exceedance Calculations

7.1 Basic Definitions

Mapping Manual 2004 • Chapter VII Exceedance Calculations Page VII - 1


Mapping Manual 2004 • Chapter VII Exceedance CalculationsPage VII - 2

when considering a geological time frame.But also during shorter time periods, such asdecades or centuries, one can anticipatechanges in the magnitude of critical loadsdue to global (climate) change, which influ-ences the processes from which criticalloads are derived. An example of a study ofthe (first-order) influence of temperature andprecipitation changes on critical loads ofacidity and nutrient N in Europe can be foundin Posch (2002).

The exceedance of a critical load is oftenmisinterpreted as the amount of excessleaching, i.e., the amount leached above thecritical/acceptable leaching. This is in general not the case as exemplified by the exceedance of the critical load ofnutrient N. The excess leaching due to thedeposition Ndep, Exle, is given as:

(7.4)

Inserting the mass balance of N and the deposition-dependent denitrification oneobtains for the excess leaching (eqs.5.2-5.5):

(7.5)

which shows that a deposition reduction of 1 eq/ha/yr reduces the leaching of N by only1–fde eq/ha/yr. Only in the simplest case, inwhich all terms of the mass balances areindependent of depositions, equals the

change in leaching the change in deposition.

The non-uniqueness of the critical loads of Sand N acidity makes both their implementa-tion into integrated assessment models andtheir communication to decision makersmore difficult. However, if one is interestedin reductions of only one of the two pollutants, a unique critical load can bederived, and thus also a unique exceedanceaccording to eq. 7.1 can be calculated.

If emission reductions deal with nitrogenonly, a unique critical load of N for a fixedsulphur deposition Sdep can be derivedfrom the critical load function. We call it theconditional critical load of nitrogen,CL(N|Sdep), and it is computed as:

(7.6)

with

(7.7)

In Figure 7.1a the procedure for calculatingCL(N|Sdep) is depicted graphically.

7.2 Conditional Critical Loads of N and S

Ndep

Sde

p

S'

CL(N|S')

S"

CL(N|S")

(a)

Ndep

Sde

p

N'

CL(S|N')

N"

CL(S|N")

(b)

Figure 7.1: Examples of computing (a) conditional critical loads of N for different S deposition values S' and S",and (b) conditional critical loads of S for different N deposition values N' and N".

In an analogous manner a conditional criticalload of sulphur, CL(N|Sdep), for a fixed nitrogen deposition Ndep can be computedas:

(7.8)

where α is given by eq. 7.7. The procedurefor calculating CL(N|Sdep) is depicted graphi-cally in Figure 7.1b. Setting Ndep=CLnut(N),the resulting conditional critical load hasbeen termed minimum critical load of sulphur: CLmin(S)=CL(S|CLnut(N)).

Eq. 7.8 would have been the consistent procedure for calculating critical loads of Sused in the negotiations of the 1994 OsloProtocol; however, critical loads for nitrogen - CLmin(N) and CLmax(N) - were notavailable then.

When using conditional critical loads, thefollowing caveats should be kept in mind:

(a)A conditional critical load can be considered a true critical load only when the chosen deposition of the other pollutant is kept constant.

(b)If the conditional critical loads of both pollutants are considered simultane-ously, care has to be exercised. It is not necessary to reduce the exceedances of both, but only one of

them to reach non-exceedance for both pollutants; recalculating the conditional critical load of the other pollutant results (in general) in non-exceedance. However, if Sdep>CLmax(S) or Ndep>CLmax(N), depo-sitions have to be reduced at least totheir respective maximum critical load values, irrespective of the conditional critical loads.

As has been shown in Chapter 5.3, their is nounique critical load of S and N acidity, and alldeposition pairs (Ndep,Sdep) lying on the critical load function lead to the criticalleaching on ANC (see eq. 5.19 and Figure5.1). Whereas non-exceedance is easilydefined (as long as its amount is not impor-tant), there is no unique exceedance of acidity critical loads. This is illustrated inFigure 7.1a: Let the point E denote the (current) deposition of N and S. By reducingNdep substantially, one reaches the point Z1and thus non-exceedance without reducingSdep; on the other hand one can reach non-exceedance by only reducing Sdep (by asmaller amount) until reaching Z3; finally,with a reduction of both Ndep and Sdep, onecan reach non-exceedance as well (e.g.point Z2).



7.3 Two Pollutants

Ndep

Sde

p

CLmin(N) CLmax(N)

CLmax(S)Z1

Z3

E

Z2

(a)

Ndep

Sde

p

CLmin(N) CLmax(N)

CLmax(S)E

ExN

ExS

(b)

Figure 7.2: Critical load function for S and acidifying N (thick line; see Figure 5.1). The grey-shaded area belowthe critical load function defines deposition pairs (Ndep, Sdep) for which there is non-exceedance. (a) The points Eand Z1-Z3 show that there is no unique exceedance; (b) the quantities involved in the definition of an exceedance(see text for further explanations).



Intuitively, the reduction required in N and Sdeposition to reach point Z2 (see Figure7.2b), i.e., the shortest distance to the criticalload function, seems a good measure forexceedance. Thus we define theexceedance for a given pair of depositions(Ndep,Sdep) as the sum of the N and S deposi-tion reductions required to reach the criticalload function by the 'shortest' path. Figure7.3 depicts the five cases that can arise:

(a)the deposition falls on or below the critical load function (Region 0). In this case the exceedance is defined as zero (non-exceedance);

(b)the deposition falls into Region 1 (e.g. point E1). In this case the line perpen-dicular to the critical load function would yield a negative Sdep, and thus

every exceedance in this region is defined as the sum of N and S depo-sition reduction needed to reach point Z1;

(c)the deposition falls into Region 2 (e.g. point E2): this is the 'regular' case, the exceedance is given by the sum of Nand S deposition reduction, ExN+ExS, required to reach the point Z2, such that the line E2-Z2 is perpen-dicular to the critical load function;

(d)Region 3: every exceedance is defined as the sum of N and Sdeposition reduction needed to reach point Z3;

(e)Region 4: the exceedance is simply defined as Sdep–CLmax(S).

Ndep

Sde

p

Region 0

Region 1

Region 2Region 3Region 4

E1

Z1

E2

Z2

ExS

ExN

E3

Z3

E4

Z4

Figure 7.3: Illustration of the different cases for calculating the exceedance for a given critical load function.

The exceedance function can be describedby the following equation; note the point Z2on the critical load function obtained bydrawing a perpendicular line through a point

in Region 2 (see Figure 7.3) is denoted by(N0, S0):

(7.9)

The function thus defined fulfils the criteriaof a meaningful exceedance function: it iszero, if there is no exceedance of criticalloads, positive when there is exceedance,and increasing in value when the point(Ndep,Sdep) moves away from the critical loadfunction.

The computation of the exceedance functionrequires the estimation of the coordinates ofthe point Z2 on the critical load function. If(x1,y1) and (x2,y2) are two arbitrary points of astraight line g and (xe,ye) another point (noton that line), then the coordinates (x0,x0) ofthe point obtained by intersecting the linepassing through (xe,ye) and perpendicular tog (called the 'foot' or 'foot of the perpendicu-lar') are given by:

(7.10a)

with(7.10b)

and(7.10c)

Applying these equations to(x1,y1)=(CLmin(N),CLmax(S)), (x2,y2)=(CLmax(N),0) and (xe,ye)=(Ndep,Sdep) oneobtains the point (x0,y0)=(N0,S0) (Z2 in Figure7.3). The final difficulty in computing theEx(Ndep,Sdep) is to determine into which of theregions (Region 0 through Region 4 in Figure7.3) a given pair of deposition (Ndep,Sdep) falls.Without going into the details of the geomet-rical considerations, a FORTRAN subroutineis listed below, which returns the number ofthe region as well as ExN and ExS:





For a given critical load function anexceedance can be defined for every pair ofdepositions (Ndep,Sdep) as outlined above.Connecting points in the (Ndep,Sdep) planewhich have identical values of the

exceedance function, results in exceedanceisolines as illustrated Figure 7.4. The ‘kinks’in the isolines when passing from oneexceedance region to a neighbouring onecan be clearly discerned.

Ndep

Sde

p

0

50

100

150

200

250

Figure 7.4: Exceedance isolines for a given critical load function. The line labelled "0" corresponds to the criticalload function and delimits the grey-shaded area of zero exceedance.

Since exceedance calculations for the critical loads for surface waters require special considerations due to the peculiari-ties of (some of) the models, they are treatedhere separately. The three critical load mod-els mentioned are described in Chapter 5.4.

7.4.1 The SSWC model

In the SSWC model sulphate is assumed tobe a mobile anion (i.e. leaching equals deposition), whereas nitrogen is assumed toa large extent to be retained in the catch-ment by various processes. Therefore, onlythe so-called present-day exceedance canbe calculated from the leaching of N, Nle,which is determined from the sum of themeasured concentrations of nitrate andammonia in the runoff. This presentexceedance of the critical load of acidity isdefined as (Henriksen and Posch 2001):

(7.11)

where CL(A) is the critical load of acidity ascomputed with eq. 5.50. No N depositiondata are required for this exceedance calcu-lation; however, Ex(A) quantifies only theexceedance at present rates of retention ofN in the catchment. Only in the FAB model(see below) are nitrogen processes modelledexplicitly, and thus only that model can beused for comparing the effects of different Ndeposition scenarios. In the above derivationwe assumed that base cation deposition andnet uptake did not change over time. If thereis increased base cation deposition due tohuman activities or a change in the netuptake due to changes in management prac-tices, this has to be taken into account in theexceedance calculations by subtracting thatanthropogenic BC*

dep–Bcu from S*dep+Nle.

7.4.2 The empirical diatom model

For this empirical model the exceedance ofthe critical load of acidity is given by:

(7.12)

where fN is the fraction of the nitrogen depo-sition contributing to acidification (see eq.5.66).

7.4.3 The FAB model

Since in the FAB model a critical load func-tion for surface waters is derived from the inthe same way as in the SMB model for soils,the same considerations hold as given inSection 7.2. Again there is no uniqueexceedance for a given pair of depositions(Ndep,Sdep), but an exceedance can bedefined in an analogous manner as for thesoil critical load function (see also Henriksenand Posch 2001).

In the previous sections the excess of depo-sitions over critical loads has been definedas exceedance. In the case of target loads ortarget load functions the same quantities asdefined above can be calculated, but if theyare positive, we talk about the non-achieve-ment of the target load; if it is zero or nega-tive, we say the target (load) for a given yearhas been achieved.



7.4 Surface Waters

7.5 Target Loads



Ducros M (1845) Observation d'une pluieacid. Journale de la PharmacologieChimique 3: 273-277.

Henriksen A Posch M (2001) Steady-statemodels for calculating critical loads ofacidity for surface waters. Water, Air andSoil Pollution: Focus 1: 375-398.

Posch M, Hettelingh J-P, De Smet PAM,Downing RJ (eds) (1997) Calculation andmapping of critical thresholds in Europe.Status Report 1997, Coordination Centerfor Effects, RIVM Report 259101007,Bilthoven, Netherlands, iv+163 pp.www.rivm.nl/cce

Posch M De Smet PAM, Hettelingh J-P,Downing RJ (eds) (1999) Calculation andmapping of critical thresholds in Europe.Status Report 1999, Coordination Centerfor Effects, RIVM Report 259101009,Bilthoven, Netherlands, iv+165 pp.www.rivm.nl/cce

Posch M (2002) Impacts of climate changeon critical loads and their exceedances inEurope. Environmental Science andPolicy 5(4): 307-317.

Quinion M (2004) "The example I quoted [onwww.worldwidewords.org] is known tothe compilers of the OED (who will bewriting an entry for 'exceedance' at somepoint) from their in-house database ofunpublished citations. They didn't provideme with a full citation, only with the date"(personal communication to JulianAherne in July 2004).

Smith RA (1872) Air and Rain: TheBeginnings of a Chemical Climatology.Longmans, Green & Co., London, 600 pp.

References

In this Chapter procedures are discussedwhich can used for summarising and presenting results of critical load andexceedance calculations and on a regionalscale, in particular on the EMEP grid. Thematerial presented here is a summary of thematerial from CCE Status Reports (see Posch et al. 1995, 1997, 1999). In Section 8.1 the EMEP grid is defined, inSection 8.2 the methods for the calculationof percentiles (in one and two dimensions)are presented, and in Section 8.3 the (average) accumulated exceedance isdefined.

To make critical loads usable and useful forthe work under the LRTAP Convention, onehas to be able to compare them to deposi-tion estimates. Deposition of sulphur andnitrogen compounds have earlier beenreported by EMEP on a 150 x 150 km2 gridcovering (most of) Europe, but in recentyears depositions have also become available on a 50 x 50 km2 grid. Both are special cases of the so-called polar stereo-graphic projection, which is described in thefollowing.

8.1.1 The polar stereographic projection

In the polar stereographic projection eachpoint on the Earth’s sphere is projected fromthe South Pole onto a plane perpendicular tothe Earth’s axis and intersecting the Earth ata fixed latitude f0. (See Figure A-1 in the CCEStatus Report 2001, p. 182.) Consequently,the coordinates x and y are obtained fromthe geographical longitude l and latitude f(in radians) by the following equations:

(8.1)

and

(8.2)

where (xp, yp) are the coordinates of theNorth Pole; l0 is a rotation angle, i.e. the longitude parallel to the y-axis; and M is thescaling of the x-y coordinates. In the abovedefinition the x-values increase and the y-values decrease when moving towards theequator. For a given unit length (grid size) d in the x-y plane the scaling factor M is givenby

(8.3)

where R (= 6370 km) is the radius of theEarth. The inverse transformation, i.e. longi-tude and latitude as function of x and y, isgiven by

(8.4)

and(8.5)

The arctan in eq. 8.5 gives the correct longitude for quadrant 4 (x>xp and y<yp) andquadrant 3 (x<xp and y<yp); p (=180°) has tobe added for quadrant 1 (x>xp and y>yp) andsubtracted for quadrant 2 (x<xp and y>yp).Note that quadrant 4 is the one covering(most of) Europe.

Every stereographic projection is a so-calledconformal projection, i.e. an angle on thesphere remains the same in the projection

8 General Mapping Issues

Mapping Manual 2004 • Chapter VIII General Mapping Issues Page VIII - 1

8.1 The EMEP grid


Mapping Manual 2004 • Chapter VIII General Mapping IssuesPage VIII - 2

plane, and vice versa. However, the stereo-graphic projection distorts areas (even locally), i.e. it is not an equal-area projection(see below).

We define a grid cell (i,j) as a square in thex-y plane with side length d (see eq. 8.3) andcentre point as the integral part of x and y,i.e.

(8.6)

where ‘nint’ is the nearest integer (roundingfunction). Consequently, the four corners ofthe grid cell have coordinates (i±½, j±½).

8.1.2 The EMEP grid

The 50×50 km2 grid (EMEP50 grid):The eulerian dispersion model of EMEP/MSC-W produces concentration and depo-sition fields on a 50 x 50 km2 grid with theparameters (see also www.emep.int):

(8.7)

yielding M=237.7314...

The 150×150 km2 grid (EMEP150 grid):The coordinate system used by EMEP/MSC-W for the (old) lagrangian long-rangetransport model is defined by the followingparameters (Saltbones and Dovland 1986):

(8.8)

which yields M=79.2438... .

An EMEP150 grid cell (i,j) contains 3x3=9EMEP50 grid cells (m,n) with all combina-

tions of the indices m=3i–2, 3i–1, 3i andn=3j–2, 3j–1, 3j. The part of the two EMEPgrid systems covering Europe is shown inFigure 8.1.



5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41123456789101112131415161718192021222324252627282930313233343536373839404142

14 17 20 23 26 29 32 35 38 41 44 47 50 53 56 59 62 65 68 71 74 77 80 83 86 89 92 95 98 101 104 107 110 113 116 119 122

2

5

8

11

14

17

20

23

26

29

32

35

38

41

44

47

50

53

56

59

62

65

68

71

74

77

80

83

86

89

92

95

98

101

104

107

110

113

116

119

122

125

Figure 8.1: The EMEP150 grid (solid lines) and EMEP50 grid (dashed lines). The numbers at the bottom and to theright are EMEP150 grid indices; those at the top and to the left are EMEP50 grid indices (every third).



To convert a point (xlon,ylat), given in degrees of longitude and latitude, into EMEP coordinates(emepi,emepj) the following FORTRAN subroutine can be used:

EMEP50 coordinates are obtained by calling the above subroutine with par(1)=50, par(2)=8 andpar(3)=110; and EMEP150 coordinates are obtained with par(1)=150, par(2)=3 and par(3)=37.Conversely, the EMEP coordinates of a point can be converted into its longitude and latitudewith the following subroutine:

8.1.3 The area of an EMEP grid cell

As mentioned above, the stereographic projection does not preserve areas, e.g. a 50 x 50 km2 EMEP grid cell is 2,500 km2 onlyin the projection plane, but never on theglobe. The area A of an EMEP grid cell withlower-left corner (x1, y1) and upper-right corner (x2, y2) is given by:

(8.9)

where u1=(x1 – xp)/M, etc.; and I(u,v) is the

double integral (see Posch et al. 1997 for details):

(8.10)

These two equations allow the calculation ofthe area of the EMEP grid cell (i,j) by setting(x1,y1)=(i–½, j–½) and (x2, y2)=(i+½, j+½).



The following FORTRAN functions compute the area of an EMEP grid cell for arbitrary gridindices (i,j), for the EMEP50 or the EMEP150 grid, depending on the parameters in par() (seeabove):



The area distortion ratio a, i.e. the ratiobetween the area of a small rectangle in theEMEP grid and its corresponding area on theglobe, is obtained by the following limitoperation:

(8.11)

where R, M, d and r are defined in eqs. 8.1–8.5. Using eqs. 8.3 and 8.5 and the identities1/(1+tan2z) = cos2z and 2cos2(p/4 – z/2) = 1+sin z,one arrives at the following expression forthe area distortion ratio:

(8.12)

which shows that the distortion ratiodepends on the latitude f only, and (small)areas are undistorted, i.e. a=1, only atf=f0=60°.

In this section we first define and investigatedifferent methods for calculating percentilesof a cumulative distribution function (cdf)given by a finite number of values. Then wegeneralise the concept of a percentile to thecase in which the cdf is defined by a set offunctions (critical load functions), resulting inthe so-called percentile function (protectionisoline).

8.2.1 Cumulative distribution function

Let us assume we have critical load valuesfor n ecosystems. We sort these values inascending order, resulting in a sequence x1 £ x2 £ … £ xn. Each value is accompanied

by a weight (area) Ai (i=1,...,n), characterizingthe size (importance) of the respectiveecosystem. From these we compute normalized weights wi according to

(8.13)

resulting in:

(8.14)

The cumulative distribution function (cdf) ofthese n critical load values is then defined by

(8.15)

with

(8.16)

F(x) is the probability of a critical load beingsmaller than (or equal to) x, i.e. 1–F(x) is thefraction of ecosystems protected. With thisdefinition F(x) has the mathematical proper-ties of a cdf: F is a monotonously increasingright-continuous function with F(–¥)=0 andF(¥)=1. In Figure 8.2 an example of a cdf isshown; note that the function assumes onlya finite number of values.

8.2 Percentiles and ProtectionIsolines

8.2.2 Quantiles and percentiles

All ecosystems in a region (grid cell) are protected, if deposition stays below thesmallest critical load values. However, to discard outliers and to account for uncer-tainties in the critical load calculations, butalso to ensure that a sufficient percentage ofecosystems are protected, (low) percentilesof the cdf are compared to the deposition.

The q-th quantile (0£ q £1) of a cdf F, denoted by xq, is the value satisfying

(8.17)

which means that xq, viewed as a function ofq, is the inverse of the cdf, i.e. xq=F–1(q).

Percentiles are obtained by scaling quantiles to 100, i.e. the p-th percentile is the(p/100)-th quantile. Other terms used aremedian for the 50-th percentile, lower andupper quartile for the 25-th and 75-th percentile, respectively. We suggest the termpentile for the 5-th percentile (from the Greekword penta for five). Note that the p-th percentile critical load protects 100–p per-cent of the ecosystems.

Computing quantiles, i.e. the inverse of a cdfgiven by a finite number of points poses aproblem: due to the discrete nature of thecdf, a unique inverse simply does not exist.For many values of q no value xq exists at all so that eq. 8.17 holds; and for the n values xisuch a value exists (i.e. q=F(xi)), but the

resulting quantile is not unique – every valuebetween xi and xi+1 could be taken (seeFigure 8.2). Therefore, the cdf is approxi-mated (interpolated) by a function whichallows solving eq. 8.17 for every q. There isneither a unique approximation, nor is therea single accepted way for calculating percentiles; e.g., Posch et al. (1993) discusssix methods for calculating percentiles. Notethat commonly definitions are given for datawith identical weights (i.e. wi=1/n), but thegeneralization to arbitrary weights is mostlystraightforward. It should be also born inmind that the differences between differentapproximation methods vanish when thenumber of points becomes very large (andall weights small).

In the following we have a closer look at twotypes of quantile functions: (a) those derivedfrom linearly interpolating the cdf, and (b)those using the empirical cdf. After definingtheir equations for arbitrary weights we dis-cuss their advantages and disadvantages.

(a) Linear interpolation of the cdf:In this case the quantile function is theinverse of the linearly interpolated cdf givenby:

(8.18)

where the Wk are defined in eq. 8.16. Anexample is shown in Figure 8.3a.



x1 x2 x3 x4 x50

0.2

0.4

0.6

0.8

1.0

(a)

x1 x2 x3 x4 x50

0.2

0.4

0.6

0.8

1.0

(b)

Figure 8.2: (a) Example of a cumulative distribution function for n=5 data points (x1<x2<x3<x4<x5, with weightsw1=2/15, w2=4/15, w3=5/15, w4=1/15, w5=3/15). The filled (empty) circles indicate whether a point is part (not part) of thefunction. (b) The same cdf is drawn by connecting all points, the way a cdf is usually displayed.



The advantage of this quantile function isthat it is continuous, i.e. a small change in qleads to only a small change in the resultingquantile xq. However, it has the followingthree disadvantages:

(i) In case of two (or more) identical datapoints the definition of the quantile functionis not unique: for identical critical load values the shape of the interpolation

function depends on the order of the weights(see Figures 8.4a,a’). This could be resolvedby sorting the weights of identical datapoints according to size (smallest first, as inFigures 8.4a.b). This minimizes the difference to the empirical distribution function (see below), but requires fairly complicated (and time-consuming) routinesfor the actual computations.

x1 x2 x3 x4 x50

0.2

0.4

0.6

0.8

1.0

(a)

x1 x2 x3 x4 x50

0.2

0.4

0.6

0.8

1.0

(b)

Figure 8.3: Examples of the two quantile functions discussed in the text. Values and weights are the same as inFigure 8.2. The filled (empty) circles indicate whether a point is part (not part) of the function. The thin horizontal linesindicate the cumulative distribution function. Note that for almost all values of q (e.g. q=0.35) the resulting quantile issmaller in (a) than in (b).

x1 x2 x3,4 x50

0.2

0.4

0.6

0.8

1.0

(a)

x1 x2 x3,4 x50

0.2

0.4

0.6

0.8

1.0

(b)

x1 x2 x3,4 x50

0.2

0.4

0.6

0.8

1.0

(a')

x1 x2 x3,4 x50

0.2

0.4

0.6

0.8

1.0

(b')

Figure 8.4: Examples of the two quantile functions discussed in the text. Values and weights are the same as inFigure 8.2, except that x3=x4 (compare Figure 8.3). Note, that for the linearly interpolated quantile function (a,a’) itsshape depends on the order of the weights for the identical values.

(ii) As mentioned above, a critical load xq isselected to protect the (1–q)-th fraction ofthe ecosystems within a given region (gridcell). However, for the linear interpolatedquantile function certain choices of q resultin xq-values which are below the actual valueneeded to protect a fraction 1–q of theecosystems (see example in Figure 8.3). Thisis fine for the ecosystems, but may lead tohigher costs for abatement.

(iii) The computation of quantiles is notorder-preserving when using linear inter-polation. We say the order is preserved by aquantile function, if the following holds fortwo cdfs:

(8.19)

i.e. the smaller cdf leads to smaller quantiles.In Figure 8.5a an example is shown with twodata sets for the same n ecosystems, x1,...,xnand y1,...,yn with common weights w1,...,wnand the property xi < yi for i=1,...,n (e.g.CLmin’s and CLmax’s). But for certain values ofq it turns out that xq > yq when computed bylinear interpolation (Fig. 85.a).

(b) Empirical distribution function:In this case the quantile function assumesonly values defining the cdf:

(8.20)

An example of this quantile function isshown in Figure 8.3b. The disadvantage ofthis quantile function is that it is not continuous, i.e. a very small change in q maylead to a significant change in the quantile xq(jump from xi to xi’’1).

However, none of the disadvantages of thelinear interpolation holds for this function,but:

(i) identical values do not lead to ambiguities(see Figures 8.4b,b’),

(ii) the quantile xq protects (at least) a fractionq of the ecosystems (see Figure 8.3b), and

(iii) the computation of quantiles is order-preserving (see eq. 8.19 and Figure 8.5b).

It is especially property (iii) which makes theempirical distribution function the only viablechoice for computing percentiles. The following FORTRAN subroutine computesthe q-quantile of a given vector of data witha corresponding vector of weights. The datahave to be sorted in ascending order, but theweights don’t have to be normalised to one:



x1 y1 x2 y2x3 y30

0.2

0.4

0.6

0.8

1.0

(a)

x1 y1 x2 y2x3 y30

0.2

0.4

0.6

0.8

1.0

(b)

Figure 8.5: Example of two quantile functions for 3 values each (x1, x2, x3 and y1, y2, y3) and common weightsw1, w2, w3 and the property xi<yi for i=1,2,3. However, in case (a) the median x0.5 is greater than the median y0.5.



8.2.3 Percentile functions and protection isolines

In this section we generalize of the conceptof cumulative distribution function (cdf) andquantile (percentile) to the case when thedata (e.g. critical loads) are given as a functions (rather than as single values),which is the case when considering two pollutants (e.g. sulphur and nitrogen in thecase of acidification), leading to the so-called percentile function or (ecosystem)protection isoline.

In the following we assume that a (criticalload) function is defined by a set of pairs ofvalues (nodes) (xj,yj), (j=1,...,m), and the function is given by connecting (x1,y1) with(x2,y2) etc., in this way generating a polygonin the x-y plane. We denote this polygon by:

(8.22)

For the values xj and yj we assume that:

(8.23)

i.e. the nodes on the polygon are numberedfrom left to right, starting on the y-axis andending on the x-axis. Eq. 8.23 also ensuresthat the polygon is monotonically decreasing, when considered as a functionof x or y. (Alternatively, the numbering couldstart on the x-axis, etc.). With the notation(x,y)<¦ we mean that the point (x,y) lies belowthe polygon (i.e. critical loads are notexceeded).

Considering the critical load for S and Nacidity the critical load function for anecosystem is defined by 3 values, namelyCLmin(N), CLmax(N) and CLmax(S), and as apolygon with m=3 nodes it is written according to eq. 8.22 as:

(8.24)

where we assumed that the N-deposition isplotted along the x-axis and the S-depositionalong the y-axis.

Now we assume that we have n critical loadfunctions ¦1,...,¦n with respective weightsw1,...,wn (Swi=1). In general it will not be possible to sort these critical load functions,i.e. it is not possible to say that ¦i is larger orsmaller than ¦j, because CLmax(S) for ¦i couldbe larger and CLmax(N) smaller than the corresponding values for ¦j (see Figure 8.6for examples). Nevertheless, we can define acumulative distribution function F in the following way:

(8.25)

meaning that for a given point (x,y) we sumall weights wi for which (x,y)<¦i, i.e. for whichthere is no exceedance. Obviously 0 £ F(x,y) £ 1, and F has also otherwise allproperties of a (two-dimensional) cdf. A percentile p is now easily defined as theintersection of such a function with a

horizontal plane at height q=p/100. The result(projected onto the x-y plane) is a curve,more precisely a polygon which has theproperty defined in eq. 8.23. Let ¦q be thequantile (percentile) function for a given q,then every point (x,y), i.e. every pair of N andS deposition, with (x,y)<¦q protects (at least)a fraction of 1–q of the ecosystems; and ¦q isalso called a (ecosystem) protection isoline.Note that protection isolines for the same setof polygons (critical load functions) do notintersect (although they might partly coin-cide), and for r<s ¦r lies below ¦s.

Since an exact computation of a percentilefunction is hardly feasible (especially in caseof a large number of critical load functions),we have to use an approximate method (seeFigure 8.6): we draw rays through the originof the x-y plane (i.e. lines with a constant S:Ndeposition ratio) and compute the inter-sections of these rays with all critical loadfunctions (small circles in Figure 8.6a). Foreach ray the intersection points are sortedaccording to their distance from the originand the chosen quantiles of these distancesare calculated according to eq. 8.20. Finally,the resulting quantile values are connectedto obtain the percentile functions (protectionisolines). Obviously, the more rays are usedin this procedure the more accurate are theprotection isolines. As Figure 8.6b shows, aprotection isoline need not be convex.



250 750 1250

250

750

N

S (a)

250 750 1250

250

750

N

S (b)

Figure 8.6: Computation of protection isolines: (a) set of critical load functions and intersection of these CL-functions with rays from the origin (small circles); (b) computing the percentiles (q=0.25, 0.50 and 0.75 in this case)along each ray (small diamonds) and connecting them to obtain the protection isolines (thick [red] lines).



Above we showed how to summarise criticalloads data with the help of cumulative distribution functions and protection isolines. Here we want to accomplish thesame for exceedances computed for twopollutants, i.e. exceedances of critical loadfunctions. Let Exi(Ndep,Sdep) be theexceedance for ecosystem i with area Ai asdefined in Chapter 7, then we define theaccumulated exceedance (AE) of necosystems on a region (grid cell) as:

(8.26)

For a given deposition, AE is total amount ofacidity (in eq/yr) which is deposited inexcess over the critical loads in the region ina given year. This function is thus stronglydetermined by the total ecosystem area in agrid cell. In order to minimise this dependence and to obtain a quantity whichis directly comparable to depositions (in eq/ha/yr), we define the average accu-mulated exceedance (AAE) by dividing theAE function by the total ecosystem area:

(8.27)

Instead of the total ecosystem area, onecould also think of dividing by another area,e.g. the area exceeded for a given (fixed)deposition scenario. However, re-calculatingthe AAE with new areas when depositionshave changed can lead to inconsistencies:the new AAE could be larger, despite declining deposition – as can be shown withsimple examples. In analogy to protectionisolines, isolines of AAE can be calculatedfor a given region (grid cell).

Except for the earliest protocols, integratedassessment modellers have used uniformpercentage reductions of the excess deposition (so-called gap closures) to defineemission reduction scenarios. In the following we summarize the different gapclosure methods used and illustrate them forthe case of a single pollutant. This sectionfollows largely Posch et al. (2001).

In the 1994 Sulphur Protocol, only sulphurwas considered as acidifying pollutant (N deposition was fixed; it determined,together with N uptake and immobilization,the sulphur fraction). Furthermore, takinginto account the uncertainties in the CLcalculations, it was decided to use the 5-thpercentile of the critical load cdf in a grid cellas the only value representing the ecosystem sensitivity of that cell. And theexceedance was simply the differencebetween the (current) S deposition and that5-th percentile critical load. This is illustratedin Figure 8.7a): Critical loads and depositionare plotted along the horizontal axis and the(relative) ecosystem area along the verticalaxis. The thick solid and the thick brokenlines are two examples of critical load cdfs(which have the same 5-th percentile criticalload, indicated by ‘CL’). ‘D0’ indicates the(present) deposition, which is higher than theCLs for 85% of the ecosystem area. The difference between ‘D0’ and ‘CL’ is theexceedance in that grid cell. It was decidedto reduce the exceedance everywhere by afixed percentage, i.e. to ‘close the gap’between (present) deposition and (5-th percentile) critical load. In Figure 8.7a, adeposition gap closure of 60% is shown asan example. As can be seen, a fixed deposition gap closure can result in very different improvements in ecosystem protection percentages (55% vs. 22%),depending on the shape of the critical loadcdf.

8.3 The Average Accumulated Exceedance (AAE)

8.4 Critical Load Exceedance and Gap Closure Methods



100%

eq/ha/yr

ecos

yste

m a

rea

CL exceedance60% deposition gap closure

CL

5%

D0

85%

55%

22%

D1

(a)

100%

eq/ha/yr

ecos

yste

m a

rea

area

unp

rote

cted

60%

are

a ga

p cl

osur

e

D0

85%

D1 D2

(b)

100%

eq/ha/yr

ecos

yste

m a

rea

60% AAE gap closure

D0

85%

D1

67%

(c)

Figure 8.7: Cumulative distribution function (thick solid line) of critical loads and different methods of gap closure:(a) deposition gap closure, (b) ecosystem gap closure, and (c) accumulated exceedance (AE) gap closure. The thickdashed line in (a) and (b) depict another cdf, illustrating how different ecosystem protection follows from the samedeposition gap closure (a), or how different deposition reductions are required to achieve the same protection level(b).



To take into account all critical loads within agrid cell (and not only the 5-th percentile), itwas suggested to use an ecosystem areagap closure instead of the deposition gapclosure. This is illustrated in Figure 8.7b: fora given deposition ‘D0’ the ecosystem areaunprotected, i.e. with deposition exceedingthe critical loads can be read from the vertical axis. After agreeing to a certain (percent) reduction of the unprotected area(e.g. 60%), it is easy to compute for a givencdf the required deposition reduction (‘D1’and ‘D2’ in Figure 8.7b). Another importantreason to use the ecosystem area gap closure is that it can be easily generalized totwo (or more) pollutants, which is not thecase for a deposition-based exceedance.This generalization became necessary forthe negotiations of the 1999 GothenburgProtocol, as both N and S contribute to acidification. Critical load values have beenreplaced by critical load functions and percentiles replaced by ecosystem pro-tection isolines (see above). However, theuse of the area gap closure becomes problematic if only a few critical load valuesor functions are given for a grid cell. In sucha case the cdf becomes highly dis-continuous, and small changes in depositionmay result in either no increase in the protected area at all or large jumps in thearea protected.

To remedy the problem with the area gapclosure caused by discontinuous cdfs, theaccumulated exceedance (AE) concept hasbeen introduced (see above). In the case ofone pollutant, the AE is given as the areaunder the cdf of the critical loads (the entiregrey-shaded area in Figure 8.7c). Depositionreductions are now negotiated in terms of anAE (or AAE) gap closure, also illustrated inFigure 8.7c: a 60% AE gap closure isachieved by a deposition ‘D1’ which reducesthe total grey area by 60%, resulting in thedark grey area; also the corresponding protection percentage (67%) can be easilyderived. The greatest advantage of the AEand AAE is that it varies smoothly as deposition is varied, even for highly discontinuous cdfs, thus facilitating optimization calculations in integratedassessment. The advantages and disadvantages of the three gap closure methods described above are summarizedin the following table.

* It assumes a linear damage function. However, this feature could also be an advantage.

Posch M, Kämäri J, Johansson M, Forsius M(1993) Displaying inter- and intra-regionalvariability of large-scale survey results.Environmetrics 4: 341-352.

Posch M, De Smet PAM, Hettelingh J-P,Downing RJ (eds) (1995) Calculation andmapping of critical thresholds in Europe.Status Report 1995, Coordination Centerfor Effects, RIVM Report 259101004,Bilthoven, Netherlands, iv+198 pp.www.rivm.nl/cce

Posch M, Hettelingh J-P, De Smet PAM,Downing RJ (eds) (1997) Calculation andmapping of critical thresholds in Europe.Status Report 1997, Coordination Centerfor Effects, RIVM Report 259101007,Bilthoven, Netherlands, iv+163 pp.www.rivm.nl/cce

Posch M, De Smet PAM, Hettelingh J-P,Downing RJ (eds) (1999) Calculation andmapping of critical thresholds in Europe.Status Report 1999, Coordination Centerfor Effects, RIVM Report 259101009,Bilthoven, Netherlands, iv+165 pp.www.rivm.nl/cce

Posch M, Hettelingh J-P, De Smet PAM(2001) Characterization of critical loadexceedances in Europe. Water, Air andSoil Pollution 130: 1139-1144.

Saltbones J, Dovland H (1986) Emissions ofsulphur dioxide in Europe in 1980 and1983. EMEP/CCC Report 1/86, NorwegianInstitute for Air Research, Lillestrøm (nowin Kjeller), Norway.



References

Manual on methodologies and criteria for Modelling and Mapping ...

Documents

Transcript of Manual on methodologies and criteria for Modelling and Mapping ...