Review of: Provisions for Air Quality Measurement, Air ... 2 FAIRMODE Final pos… · Review of:...

Review of: Provisions for Air Quality

Measurement, Air Quality Modelling,

Management Framework,

Assessment, and Public

Information; and Stakeholder

Consultation Support

Assessment of the FAIRMODE

Recommendations for the revision of the

current Air Quality Legislation

Final report

Client: European Commission, DG Environment

28th June 2013

Review of: Provisions for Air

Quality Measurement, Air Quality

Modelling, Management

Framework, Assessment, and

Public Information; and

Stakeholder Consultation Support

Assessment of the FAIRMODE Recommendations for

the revision of the current Air Quality Legislation

Client: European Commission, DG Environment

Authors:

Jørgen Brandt, Helge Rørdam Olesen, Mohammed

Hussen Alemu, Louise Martinsen, Berit Hasler and

Ole Hertel,

DCE - Danish Centre for Environment and Energy,

Aarhus University (AU), Denmark

Enda Hayes, Jo Barnes and Tim Chatterton.

UWE - University of the West of England, Bristol, UK

28th

June 2013

Table of Contents

i

Review of: Provisions for Air Quality Measurement, Air Quality Modelling, Management Framework, Assessment,

and Public Information; and Stakeholder Consultation Support

List of Tables iii

List of Figures iv

1 Analysis of the FAIRMODE Recommendations 1

1.1 Introduction 1

1.2 Challenges in assessing the FAIRMODE recommendations 1

1.3 The European Air Quality Directive (Directive 2008/50/EC) 2

1.4 The FAIRMODE network 2

1.5 The FAIRMODE Recommendations 3

1.6 Material considered in the assessment 4

1.7 Analysis of each recommendation 4

1.7.1 On the use of models (F1) 4

1.7.2 Model quality objectives (F2) 4

1.7.3 Forum of EU AQ regulatory modelling (F3) 5

1.7.4 Quality assurance and consistency of emission inventories (F4) 5

2 Assessment of the FAIRMODE recommendations 7

2.1 The initial framing of the cost assessments 7

3 Defining the framework for the FAIRMODE cost assessment 12

3.1 Challenges in establishing a cost assessment of implementing the FAIRMODE

recommendations in the revised AQD 13

3.2 The bottom-up approach: Information Request for each option and sub-

recommendation 16

3.2.1 FAIRMODE Information Request for Each Option and Sub-Recommendations 16

3.2.2 Understanding the current baseline for modelling practise for your Member State.

17

3.2.3 Cost assessment of the individual FAIRMODE recommendations. 19

3.3 The top-down approach: expert estimates calibrated to each member state. 20

4 Cost Assessment 25

4.1 The bottom-up approach: Information request for each option and sub-

recommendation 25

4.1.1 Overview 25

4.1.2 Introduction to results 27

4.1.3 Discussion of selected results 28

4.2 The top-down approach: expert estimates calibrated to each member state. 41

4.3 FAIRMODE Consultation Survey of the top-down assessment 50

4.4 Combining the bottom-up with the top-down approach: estimating the cost for fulfilling

the minimum requirements 52

5 Summary and conclusions 54

References 57

ii



Appendix A: The FAIRMODE Recommendations 58

Appendix B: Case studies for three countries 62

Belgium 62

Summary: Situation within the country and findings of interview 62

BASELINE 1: COMPETENCE BUILDING – ATMOSPHERIC SCIENCE 62

BASELINE 2: COMPETENCE BUILDING – HIGH RESOLUTION EMISSION DATABASES

63

BASELINE 3: DEVELOPMENT OR ACQUISITION OF ATMOSPHERIC MODELS 63

BASELINE 4: COMPUTER FACILITIES FOR RUNNING HIGH RESOLUTION MODELS 63

BASELINE 5: OPERATIONAL / RUNNING COSTS 63

OPTION 1: ON THE USE OF MODELS FOR REGULATORY PURPOSE AND TO

SUPPORT AIR QUALITY POLICY 64

Croatia 65


Hungary68


Appendix C: Summary of open-ended responses – by issue 72

Appendix D: Bottom-up approach. Graphs for each of the 14 questions 84

Appendix E: FAIRMODE Consultation Survey of the top-down assessment 89

Appendix F: Table 2.1, enlarged 97

iii



List of Tables

Table 4.1: List of countries which responded. It is indicated whether any quantitative data are

supplied. 25

Table 4.2: Bottom-up approach. Total cost by baseline and recommendation. 28 Table 4.3: Degree of fulfillment of the minimum requirements (percent). 37

iv



List of Figures

Figure 4.1 Sum of current + required costs to fulfil the requirements indicated in the Baseline 1

question on Competence building – atmospheric science 29

Figure 4.2: Sum of current + required costs to fulfil the requirements indicated in the Baseline 2

question on Competence building – high resolution data bases. 31

Figure 4.3: Sum of current + required costs to fulfil recommendation 1a: FAIRMODE recommends

use of models for assessment of air quality levels to establish the extent of exceedances and

establish population exposure. 32

Figure 4.4: Sum of current + required costs to fulfil recommendation 2: Revise Model Quality

Objectives 33

Figure 4.5: Sum of current costs to fulfil the requirements for all 5 baseline questions 34

Figure 4.6: Sum of current + required costs to fulfil the requirements for all 5 baseline questions 34

Figure 4.7: Degree of fulfillment of the minimum requirements (percent). 38

Figure 4.8 Sum of current costs to fulfil all 9 recommendations 39

Figure 4.9 Sum of current + required costs to fulfil all 9 recommendations. 39

Figure 4.10: Top-down approach. The total estimated cost/year for the baseline (sum of baseline 1-

5) for each member state calibrated from expert estimates for Denmark with respect to the

minimum requirements for fulfilling the FAIRMODE recommendations. 43

Figure 4.11: The total estimated cost/year for baseline 1 (modelling) for each member state

calibrated from expert estimates for Denmark with respect to the minimum requirements for fulfilling

the FAIRMODE recommendations. 44

Figure 4.12: The total estimated cost/year for the baseline 2 (emissions) for each member state



Figure 4.13: The total estimated cost/year for the baseline 4 (computing) for each member state



Figure 4.14: The total estimated cost/year for recommendation 1a (assessment exceedances and

population exposure) for each member state calibrated from expert estimates for Denmark with

respect to the minimum requirements for fulfilling the FAIRMODE recommendations. 45

Figure 4.15: The total estimated cost/year for recommendation 1b (air pollution forecasting) for

each member state calibrated from expert estimates for Denmark with respect to the minimum

requirements for fulfilling the FAIRMODE recommendations. 46

Figure 4.16: The total estimated cost/year for recommendation 1c (source allocation) for each

member state calibrated from expert estimates for Denmark with respect to the minimum


Figure 4.17: The total estimated cost/year for recommendation 1d (plans and measures to control

AQ exceedances) for each member state calibrated from expert estimates for Denmark with


Figure 4.18: The total estimated cost/year for recommendation 1e (designing monitoring networks)

for each member state calibrated from expert estimates for Denmark with respect to the minimum


v



Figure 4.19: The total estimated cost/year for recommendation 1f (number of fixed monitoring

stations) for each member state calibrated from expert estimates for Denmark with respect to the

minimum requirements for fulfilling the FAIRMODE recommendations. 48

Figure 4.20: The total estimated cost/year for recommendation 2 (revision of the data quality

objective for modelling) for each member state calibrated from expert estimates for Denmark with


Figure 4.21: The total estimated cost/year for recommendation 3 (competent authorities for

modeling activities are nominated by the Member states) for each member state calibrated from

expert estimates for Denmark with respect to the minimum requirements for fulfilling the

FAIRMODE recommendations. 49

Figure 4.22: The total estimated cost/year for recommendation 4 (investigate and improve the

compilation, consistency and quality assurance of emissions data) for each member state



Figure 4.23: Estimated total additional cost/year for each member state for fulfilling the minimum

requirements, using the EU standard cost model. The total additional cost for all member states all

together is 1.4 mio. Euros/year. 53

1



1 Analysis of the FAIRMODE Recommendations

1.1 Introduction

The present report has been produced as a part of Special Agreement 4 under Framework Contract

ENV C3/2011/FRA/008 in response to the request from the European Commission, Directorate-

General Environment: Services to support the “Review of the Thematic Strategy on Air Pollution”, in

particular concerning the Review of: “Provision for Air Quality Measurement, Air Quality Modelling,

Management Framework, Assessment, and Public information; and stakeholder Consultation

Support.”

The present report provides an assessment of the recommendations from FAIRMODE (The Forum

for Air quality Modelling in Europe), which is a joint response action of the European Environment

Agency (EEA) and the European Commission Joint Research Centre (JRC) (see also:

http://fairmode.ew.eea.europa.eu/). FAIRMODE’s aim is to bring together air quality modellers and

users in order to promote and support the harmonised use of models by EU member countries, with

emphasis on their application within the context of the European Air Quality Directive.

The aim of the presented work is to provide guidance for the European Commission with respect to

possible adaptation of the FAIRMODE recommendations in the revised Air Quality Guidelines that

are to be launched in 2014. In the following we will briefly outline the background for the

FAIRMODE recommendations

1.2 Challenges in assessing the FAIRMODE recommendations

In subsequent sections we analyse and assess the recommendations from FAIRMODE. However,

it should be recognized that the recommendations of FAIRMODE are formulated at a level that is

not yet specific enough to be directly transposed into statutory requirements. Also as a

consequence of the ongoing level of debate in the air quality management community and of the

lack of legally binding requirements in EU legislation, they are of a general nature in the sense that

there is a wide spectrum of possible ways to implement the recommendations. Therefore, an

assessment of the consequences of the recommendations can lead to several possible results,

depending on how the recommendations are interpreted. As an example, one recommendation

states that the text of the Directive should be clarified, whereas the question of how the text should

be clarified is open to discussions.

We have pursued an approach where we have chosen a certain interpretation for a full

implementation of the recommendations, while we acknowledge that other interpretations of the

recommendations can be made. Given these challenges, quantitative assessment of the

consequences of the recommendations is difficult and can give a range of outcomes. Nevertheless,

this assessment is based on the results of a survey within the member states, where we have

asked questions about the baseline within atmospheric modelling in each member state and an

assessment of additional cost for fulfilling the FAIRMODE recommendations.

2



1.3 The European Air Quality Directive (Directive 2008/50/EC)

The European Air Quality Directive (Directive 2008/50/EC) was adopted by 21st May 2008

(http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2008:152:0001:0044:

EN:PDF). By this directive most of the existing legislation on air quality was merged into one single

directive (except for the fourth Daughter Directive) without changes to existing air quality directives:

Framework Directive 96/62/EC, 1-3 daughter Directives 1999/30/EC, 2000/69/EC, 2002/3/EC,

Decision on Exchange of Information 97/101/EC. The Directive 2008/50/EC included new air quality

objectives for PM2.5 including limit value and exposure related objectives in the form of exposure

concentration obligation and exposure reduction target. Furthermore it introduced the possibility to

discount natural sources of pollution when assessing compliance against limit values, and

possibility for time extensions for compliance with guidelines for PM2.5, NO2 and benzene.

1.4 The FAIRMODE network

The main aim of FAIRMODE is to promote the use of models in a harmonised way in the context of

the European Air Quality Directive. Emphasis is on the Air Quality Directive (AQD) requirements,

mainly on the promotion of good modelling practises and the interaction between authorities and

the modelling community at national and European levels.

The FAIRMODE focuses on 1) coordinating and gathering information from modellers and users

within Europe, 2) developing guidance and recommendations on air quality modelling for modellers,

users and the European Commission, 3) providing harmonised tools and methodologies for model

benchmarking and assessment, and 4) providing recommendations for scientific research in air

quality modelling.

Some important objectives of the Forum for Air Quality Modelling are

(http://fairmode.ew.eea.europa.eu/fol568175/objectives):

To establish tools and mechanisms for enhancing communication between modellers and

model users and provide a framework for exchange of experience at all levels of application.

This will include electronic interfaces, databases (such as MDS, COST728 Metadatabase, EEA

Data Centre) and tools as well as workshops, seminars and common activities.

To provide a centralised portal for information concerning the AQD, submission of compliance

data based on modelling, references and experiences of other users through case studies, and

will provide QA/QC methods for users and provide information support for these services.

To establish a common infrastructure based on best practice for reporting and storing the

information, results and maps in a standardised and harmonised manner to create an archive

for reference where tools, data and information will be readily available to authorities and

scientists of the member states.

To promote model validation and quality assurance of model results to identify limitations and

remove error factors, which implies the organisation of and participation in model validation and

intercomparison exercises at national or European level. Such exercises will be complementary

to other parallel activities. The JRC will take on a leading role in the coordination of such

actions, gaining from its experience in leading the "Eurodelta" and "CityDelta" intercomparison

exercises.

For this purpose the following Working Groups(WG) have been set up in FAIRMODE(see

http://fairmode.ew.eea.europa.eu/fol568175/work-groups) :

WG1 - Guidance on use of models(lead by EEA).

WG2 - Quality assurance of models (lead by JRC).

http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2008:152:0001:0044:%20EN:PDF

http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2008:152:0001:0044:%20EN:PDF

http://fairmode.ew.eea.europa.eu/fol568175/objectives

http://fairmode.ew.eea.europa.eu/fol568175/work-groups

3



Within the frame of WG1, a technical reference guide has been produced, which provides

suggestions and advice on best practices for the use of modelling for the purposes of the Air

Quality Directive (AQD).

The main aim of WG2 activities is to create a European Framework for Model Evaluation which will

include the development of widely accepted quality assurance procedures throughout Europe for

different models (regional, urban and local/hotspot) and for different purposes according to the

requirements of the AQD (air quality assessment scenario calculations and impact modelling,

forecast of exceedances and assessing contribution from natural sources and winter

sanding/salting).

For each of the main modelling purposes relating to air quality management mentioned in the AQD,

a Sub-Group (SG) was formed as follows:

SG1 - Combined use of monitoring and modelling

SG2 - Contribution of natural sources and source apportionment

SG3 - Urban emissions and projections

SG4 - Tools for benchmarking of air quality models

1.5 The FAIRMODE Recommendations

The FAIRMODE Recommendations have been created through a series of discussions in the

working groups of FAIRMODE. The full paper describing the recommendations is found in

Appendix A. In the latest version dated 12/10/2012 the list of recommendations have been reduced

to four overall recommendations, of which the first comprises 6 sub-recommendations. Previous

draft versions of the recommendations exist, and some comments from stakeholders refer to the

(different) numbering of recommendations used in a version from June (marked 6/8/2012).

The condensed set of the final recommendations is given in the following: The recommendations

are labelled F1 through F4.

F1. ON THE USE OF MODELS FOR REGULATORY PURPOSE AND TO SUPPORT AIR

QUALITY POLICY

Recommendation: FAIRMODE strongly recommends the use of models for the following

applications. The AQD text relating to these applications should be clarified:

Assessment of air quality levels to establish the extent of exceedances and establish population

exposure

Forecasting air quality levels for short term mitigation and public information and warnings

Source allocation to determine the origin of exceedances and to provide a knowledge basis for

planning strategies

Development and assessment of plans and measures to control AQ exceedances

In addition to these applications the use of models is strongly recommended for:

Designing monitoring networks when models are used in combination with monitoring

Determining the number of fixed monitoring sites that are required

F2. MODEL QUALITY OBJECTIVES

Recommendation: FAIRMODE recommends a revision of the data quality objective for modelling

F3. FORUM OF EU AQ REGULATORY MODELLING

Recommendation: FAIRMODE recommends that in parallel to what has already been established

for the monitoring of air quality, competent authorities for modeling activities are nominated by the

Member States(ref Article (3) and bullet d) quality assurance of modelling)

4



F4. QUALITY ASSURANCE AND CONSISTENCY OF EMISSION INVENTORIES

Recommendation: FAIRMODE recommends to investigate and improve the compilation,

consistency and quality assurance of emissions data suitable for AQ modeling under the directive.

1.6 Material considered in the assessment

The FAIRMODE recommendations are relatively short, and as noted in section 1.2 they

are open to several interpretations. In order to guide our interpretation, as a central

reference we have considered the Technical reference guide produced as an outcome of

FAIRMODE WG1 (EEA, 2011). The guide explains how the text of the Directive relates to

modelling.

1.7 Analysis of each recommendation

1.7.1 On the use of models (F1)

FAIRMODE ’strongly recommends’ the use of models for four major applications as well as in

relation to the design of monitoring networks. The background for the recommendation is that in the

current Directive text, the role of models is not clearly specified; in the case of the application

Development and assessment of plans and measures, they are not mentioned at all, even though

this activity can hardly take place without models.

However, there is a wide spectrum of possible ways to implement the recommendation of ‘strongly

recommending’ the use of models.

When interpreted in a narrow perspective the recommendations can be fulfilled by clarifying the text

of the Directive on specific points, and encouraging member states to make use of models.

However, the underlying agenda of FAIRMODE is much more far reaching. Models are a necessary

tool to identify appropriate measures for achieving policy objectives, and their use by competent

staff should be ensured and promoted. The FAIRMODE recommendations are interpreted in this

broad perspective as a proposal to initiate a multi-year process, which includes capacity building

and eventually leads to a situation where models are used in a qualified manner throughout the

member states.

1.7.2 Model quality objectives (F2)

FAIRMODE points to the fact that the quality requirements for modelling as specified in Annex I of

the Directive are ambiguous and open to interpretation. FAIRMODE does not have an immediate

remedy for the situation. Instead, it proposes a multi-year process: FAIRMODE is developing and

testing new data quality objectives for modelling for ambient air quality assessment, in collaboration

with the Member States. This is a technically complicated process, as many issues will have to be

considered. Different quality objectives will have to be defined for each substance. It should be

clearly recognized that model performance depends on many other factors than the quality of the

model itself. The performance is critically dependent on the challenge the model is exposed to.

Thus it depends on the complexity of the situation at hand in terms of chemistry, topography and

meteorology. Especially, model performance will vary with the quality of emission available

emission data. All such issues should be considered when defining model quality objectives.

The so-called DeltaTool is in focus as a possible basis for assessing model performance, but the

use of the tool cannot yet be considered mature. A commonly agreed set of parameters to be used

EU wide to assess model performance should be discussed and accepted amongst member states.

5



Within FAIRMODE it is recognized that the common used of the tool should be further defined, and

this is a process which goes on.

The roadmap envisaged by FAIRMODE as regards model quality objectives is to further test and

develop procedures for model performance evaluation for the task of assessing current air quality

for a limited number of substances, and later extend the work of performance evaluation to other

substances as well as to the task of modelling future scenarios. The latter task requires that the

model responds dynamically to changes in the environmental situation, which is not necessary for

the task of simulating the current situation.

It should be recognized that the quality of model results not only depends on the quality of a model

as such, but also is highly dependent on the competence of modelers. Also, various models are fit

for various purposes, and modelers should be able to recognize the limitations of their models. A

skilled modeler is able to provide good foundation for decision making if the modeler knows the

limitations of the model applied. On the other hand an un-skilled modeler using the best model in

the community can provide a bad foundation for decision making, if the model is not used correct or

if the modeler is making misinterpretation of the model results. Therefore, it is equally important to

include guidance for competence building as for model quality. This makes the case for capacity

building.

1.7.3 Forum of EU AQ regulatory modelling (F3)

FAIRMODE proposes to act as a coordination forum for modelling and to support competent

authorities in a way somewhat similar to AQUILA. However, the situation for AQUILA and

FAIRMODE are not completely parallel, as there are differences between quality assurance with

monitoring and modelling. The focus is on activities as competence building, preparation of

guidance and inter-comparison exercises. Furthermore, the nature of the networks, where AQUILA

is made of accredited National Laboratories that do not have a parallel in the modelling community,

can also play a possibly relevant role.

1.7.4 Quality assurance and consistency of emission inventories (F4)

High resolution and high quality emission databases are essential for modelling. FAIRMODE

recognizes that there are many on-going activities on emission inventories, and it is not the

intention to duplicate any of these. FAIRMODE’s recommendations on emission inventories are not

specific, but are put forward in order to ensure that the needs of modellers are addressed when

compiling emission inventories.

7



2 Assessment of the FAIRMODE recommendations

2.1 The initial framing of the cost assessments

The initial framing of the cost assessment was carried out at a partner workshop at DCE,

AU, on October 26th 2012. The overall result can be seen in the following table (see also

Appendix F for an enlarged version).

Table 2.1 The initial framing of the cost assessment carried out at a partner workshop assessing

positive and negative impacts, qualitatively.

Definitions of the cost, burdens and benefits are:

Admin Cost and Admin Burden– these are split between the European Commission (EC) and

the Member States (MS).

AQDirect – will the measure directly lead to an improvement/change in air quality (e.g. new

restrictions on emissions).

AQIndirect – is there likely to be an indirect improvement in AQ

Information – will the recommendation result in better (higher resolution, more uniform and

comparable) information on air quality

Management – will the measure improve the ability of Commission and MS to better manage air

pollution, not just by better information, but better reporting etc. The focus should be on the day

to day operation of European Air Quality Management

Regulation – Will it allow the Commission to better regulate and enforce Limit Values and other

aspects of the Directives (e.g. through more robust and standardised reporting by MS).

Equity – will it impact on whether there is a ‘level playing field’ across all MS – this may mean

either applying the same standards across all MS, or accounting for regional differences.

Coherence – “the extent to which options are coherent with the overarching objectives of EU

policy, and the extent to which they are likely to limit trade-offs across the economic, social, and

environmental domain.” (from the Impact Assessment Guidelines) The key principles are

subsidiarity, proportionality and sustainable development.

Code Recommendation Sub-Option-FULL ECAdminCost ECAdminBurden MSAdminCost MSAdminBurden AQDirect AQIndirect Information Management Regulation Equity Coherence

1. Assessment of air quality levels to

establish the extent of exceedances and

establish population exposure

neutral - -- -- neutral ++ ++ ++ ++ ++ +

2. Forecasting air quality levels for short term

mitigation and public information and

warnings

Neutral neutral -- - + + ++ ++ + ++ +

3. Source allocation to determine the origin

of exceedances and to provide a knowledge

basis for planning strategies

neutral - -- -- ++ ++ ´++ ++ ++ ++ +

4. Development and assessment of plans and

measures to control AQ exceedancesneutral neutral Neutral Neutral ++ ++ ++ ++ ++ ++ +

• Designing monitoring networks when

models are used in combination with

monitoring

neutral neutral ++ + neutral ++ + + Neutral ++ Neutral

• Determining the number of fixed

monitoring sites that are requiredneutral neutral ++ + neutral ++ + + Neutral ++ Neutral

F2 2. MODEL QUALITY OBJECTIVESFAIRMODE recommends a revision of the

data quality objective for modelling- - - - neutral + + + + ++ +

F33. FORUM OF EU AQ REGULATORY

MODELLING

FAIRMODE recommends that in parallel to

what has already been established for the

monitoring of air quality, competent

authorities for modeling activities are

nominated by the Member States(ref Article

(3) and bullet d) quality assurance of

modelling)

- - - - neutral + + ++ ++ ++ ++

F4

4. QUALITY ASSURANCE AND

CONSISTENCY OF EMISSION

INVENTORIES

FAIRMODE recommends to investigate and

improve the compilation, consistency and

quality assurance of emissions data suitable

for AQ modeling under the directive

- - -- -- neutral ++ ++ ++ ++ ++ ++

Costs and Burdens Benefits

F1

1. ON THE USE OF MODELS FOR

REGULATORY PURPOSE AND TO

SUPPORT AIR QUALITY POLICY

8



Positive and Negative impacts – Positive (+) indicates a beneficial change i.e. reduced costs or

burdens, and improved management, regulation AQ etc.

In the following, a summary of the discussions during the partner workshop is given for each of the

recommendations. The discussions relate to the results in the table above.

1. Assessment of air quality levels to establish the extent of exceedances and

establish population exposure

Admin Cost and Admin Burden: The use of models to the applications in 1, will give an increased

administrative burden to the EC, but a neutral administrative cost. The models will have the

capability to provide better information on exceedances and population exposure, but will also

increase the level of information, and therefore the reporting from the MS to the EC can increase,

increasing the burden. On the other hand, the better information will make it easier for the EC to

guide the MS and therefore the cost is estimated to be neutral. The administrative cost and burden

for the MS will increase, due to additional cost to modelling and reporting.

AQdirect and AQindirect: The recommendation will not directly lead to an improvement/change in

air quality, but it has a potential for a high indirect improvement in AQ if the information is used by

decision makers for management and regulation.

Information, management and regulation: Using models to provide mapping of the exceedances

and exposure will be a powerful tool for decision makers for management and regulation, and will

provide much better information.

Equity and Coherence: A clarification of the AQD text where it is mandatory to use models to

establish the extent of exceedances and establish population exposure will have a very positive

impact on equity giving the same standards across all MS, and it will give the possibility to account

for regional differences. It will also have a positive impact on coherence with respect to the

overarching objectives of EU policy and trade-offs. A full assessment of AQ using both model

results and measurements will provide a much better foundation for decision making across the

economic, social, and environmental domain.

2. Forecasting air quality levels for short term mitigation and public information

and warnings

Admin Cost and Admin Burden: The use of models to the applications in 2, will give an neutral

administrative burden and cost to the EC. For the MS, which have not already established an

operational forecast system, the recommendation can lead to a significantly increased

administrative cost in establishing the forecast systems, but a neutral to a small increased burden

after the forecast system has been established.

AQdirect and AQ indirect: Operational air quality forecasts will have the potential to result in both

a direct improvement in air quality, since the information can lead to a change in the populations

transport behaviour, and an indirect improvement if the information is used by decision makers for

day-to-day management and long-term regulation. It can also increase the population’s

comprehension of the air quality situation and increase the pressure on decision makers to act.

Information, management and regulation: Using models for forecasting will be a powerful tool for

decision makers for the day-to-day management and will also have the potential to have a positive

9



impact on the regulation of emission sectors, since forecasts can be used to explain the possible

reasons for any exceedances of limit values due. It will provide much better information both to the

public and to decision makers.

Equity and Coherence: A clarification of the AQD text where it is mandatory to use models for

forecasting will have a very positive impact on equity, since MS will have similar and comparable

information on a day-to-day basis. It will also have a positive impact on coherence with respect to

the overarching objectives of EU policy and trade-offs. Forecasting AQ using model will provide a

much better foundation for decision making across the economic, social, and environmental

domain.

3. Source allocation to determine the origin of exceedances and to provide a

knowledge basis for planning strategies

Admin Cost and Admin Burden: The use of models to source allocations, will give an increased

administrative burden to the EC, but a neutral administrative cost. The models will have the

capability to provide direct answers to decision makers, but also more information and therefore the

reporting from the MS to the EC can increase, increasing the burden. On the other hand, a better

knowledge basis for planning strategies will make it easier for the EC to guide the MS and therefore

the cost is estimated to be neutral. The administrative cost and burden for the MS will increase, due

to additional cost to modelling and reporting.

AQdirect and AQindirect: Source allocation to determine the origin of exceedances and to provide

a knowledge basis for planning strategies will have a potential for a strong both direct and indirect

positive improvement in AQ.

Information, management and regulation: Using models for source allocation will be a powerful

tool for decision makers for the day-to-day management and on the regulation of emission sectors.

It will provide much better information both to the public and to decision makers, and will make it

easier for decision makers to obtain acceptance in the population for regulation of specific emission

sectors.

Equity and Coherence: A clarification of the AQD text where it is mandatory to use models for

source allocation will have a very positive impact on equity, since MS will have similar and

comparable foundation for management and regulation. It will also have a strong positive impact on

coherence with respect to the overarching objectives of EU policy and trade-offs. Source allocation

using model will provide a much better foundation for decision making across the economic, social,

and environmental domain.

4. Development and assessment of plans and measures to control AQ

exceedances

Admin Cost and Admin Burden: The use of models to the development and assessment of plans

and measures to control AQ exceedances, will give a neutral administrative burden and cost – both

to the EC and MS. The same work has to be carried out as before, but on a much better foundation

and on a much higher degree of information.

AQdirect and AQindirect: The use of models will have a strong positive impact on both the direct

and indirect improvement of air quality.

10



Information, management and regulation: Using models to the development and assessment of

plans and measures will have a strong positive impact on taking the right decisions – both on short-

and long-term. It will provide much better information both to the public and to decision makers, and

will make it easier for decision makers to obtain acceptance in the population for regulation of

specific emission sectors.

Equity and Coherence: A clarification of the AQD text where it is mandatory to use models for the

development and assessment of plans and measures to control AQ exceedances,will have a very

positive impact on equity, since MS will have similar and comparable foundation for management

and regulation. It will also have a strong positive impact on coherence with respect to the

overarching objectives of EU policy and trade-offs. Using model will provide a much better

foundation for decision making across the economic, social, and environmental domain.



Admin Cost and Admin Burden: The use of models to design monitoring networks when models

are used in combination with monitoring and use models for determining the number of fixed

monitoring sites that are required has a neutral impact on the EC cost and burden. However, it has

a potential for significant positive impacts on the MS administrative cost and burden, since there is

a great potential for reducing the number of measurement stations and at the same time increasing

the information level to the MS’s when models are used in combination with monitoring. However,

there is also a risk, that more fixed monitoring sites are needed and therefore increasing the cost.

AQdirect and AQindirect: The use of models will have a neutral direct impact on AQ direct but a

strong positive impact on indirect improvement of air quality, due to more and more precise

combined monitoring data, which will give a better basis for decision making.

Information, management and regulation: Using models and monitoring combined for integrated

monitoring, will have a strong positive impact on taking the right decisions – both on short- and

long-term. It will provide much better information both to the public and to decision makers of the

state of the environment based on combined monitoring data, and will make it easier for decision

makers to obtain acceptance in the population for initiating plans and measures. However, if

combined data will only be used for designing the network and determine the number of fixed

stations, it will have a neutral impact on regulation.

Equity and Coherence: Designing networks using models will have a strong positive impact on

equity, since every MS will have the same tools for addressing monitoring and AQ challenges. It will

also have a neutral impact on coherence with respect to the overarching objectives of EU policy

and trade-offs.

FAIRMODE recommends a revision of the data quality objective for modelling

Admin Cost and Admin Burden: The revision of the data quality objective for modelling will

induce an increased cost and burden for both the EC and MS.

AQdirect and AQindirect: A revision of the da quality objective for modelling will have a neutral

impact on AQ directly, but has a potential to have a strong positive indirect impact on AQ, due to

improved models and therefore improved foundation for decision making.

11



Information, management and regulation: A revision of the data quality objective for

modellingwill have the potential for improved models and therefore also improved information,

management and regulation.

Equity and Coherence: A strong positive impact on equity is foreseen, since equal data quality

objectives applicable for all MS’s will increase equity. It will also have a positive impact on

coherence. Using models with known quality objectives will provide a much better foundation for

decision making across the economic, social, and environmental domain.

3. FORUM OF EU AQ REGULATORY MODELLING




4. QUALITY ASSURANCE AND CONSISTENCY OF EMISSION INVENTORIES


consistency and quality assurance of emissions data suitable for AQ modeling under the directive

General conclusions from the discussions

Admin Cost and Admin Burden: The administrative cost and burden for the EC is believed to be

small or neutral. The administrative cost and burden will in general be significant for the MS,

depending on whether the MS already are using models or not. There is a considerable difference

between the cost and burden for MS’s that already are using models or have national model groups

that are available and the MS’s which have no history in using models. However, once the

modelling groups are established, the burden and costs are equal throughout the MS’s. The main

challenge is not in developing the models, since there are many high quality and available models

in Europe that can be achieved. The main cost and challenge lies within the competence building in

the MS’s that not already have national expertise in the field. On the other hand, there is a potential

for saving costs on the monitoring networks, when models are used together with measurements in

integrated monitoring.

Equity: there can be a negative impact on equity due to differences in models used in the different

MS. The differences in models must be dealt with in the quality assurance or minimum standards

that the models should fulfil. However, just the fact that models are used in all MS for better

assessments, forecasting, management and regulation of AQ has a very strong positive impact on

equity.

12



3 Defining the framework for the FAIRMODE cost assessment

In the following, the framework for the cost assessment is defined. The methodology for the cost

assessment is given in Hasler et al., (2013).

First of all, the cost assessment is based on the assignment in which it is assumed that the

FAIRMODE recommendations and the use of models are made mandatory in the directives. It is

important to recognize that this is a working assumption for analytical purposes and not equivalent

to the background of the FAIRMODE recommendations as they are formulated at present, but this

is a necessary condition and assumption in order to carry out a cost assessment. If there is not a

mandatory requirement for the use of models, this is similar to the present situation in the present

AQD. The question to be answered is: what would be the resource implications if the MS enacted

the recommendations emerging from the FAIRMODE forum?

The overall purpose of the cost assessment is that requirements within the member states are

assessed with respect to cost, resources and burden implications for each of the recommendations.

However, all the FAIRMODE recommendations require first of all the establishment of modelling

expertise, models and emission databases. It can be very difficult to distinguish the cost of one

recommendation from another and from the baseline costs, since they are all interconnected and all

require the same basic expertise and databases. Therefore, an assessment of cost and burden is

primarily made of the basic requirement (the baseline) for a member state authority, which does not

already use models as a part of decision making support. On top of that, the cost assessment is

made for the individual recommendations, when possible.

When issuing the member state information request (described below), it was not anticipated that

all member states would actually respond. Firstly, the survey itself was rather complicated, requiring

experts and managing staff in the field to respond. Secondly, the response period was rather short,

but necessary due to the tight time schedule. Therefore, an additional approach was also taken. A

top-down approach was initiated, where the baseline was established for Denmark as the reference

country and the results for Denmark was calibrated to all the member states with respect to salary

levels in each member states.

The cost analysis is therefore based on both the bottom-up approach represented by the output of

the information request from the member states and the top-down approach where expert

estimates for Denmark are calibrated to each member state.

In order to fulfil the FAIRMODE recommendations, the following activities and resources are

needed to establish the necessary capabilities:

1. Competence building in the area of atmospheric science with focus on atmospheric modelling

and air quality modelling at regional, urban background and urban street scales.

2. Competence building in the area of high resolution emission databases

3. The atmospheric models have to be acquired/developed – both on regional scale and on urban

background and urban street scales.

4. Computer facilities for running high-resolution models

5. Operational cost, including overhead

In overall terms, the key questions for the cost assessment are as follows:

13



Is there a need for additional staff, or staff time?

Is there a need for staff capacity development?

Is there a need for additional equipment?

Is there a need for new data generation/compilation?

Is there a requirement for additional reporting?

3.1 Challenges in establishing a cost assessment of implementing the FAIRMODE

recommendations in the revised AQD

The fundamental challenge of implementing the FAIRMODE recommendation into the revised

directive is that the recommendations are not presently in a form where they can be included

directly in the directive. They are recommendations and not precisely defined mandatory obligations

or measures. The recommendations could never point out what is needed in every member state

and could only generically address the issue. Even with a more specific approach it would be left to

member states to interpret what is needed and Interpretations of the recommendations could not be

eliminated. The listing of the fundamental needs to start from scratch a modelling activity is a good

approach on which every other state can build on depending on what is there and what is not. It

cannot be demanded to recommendations specifications which are out of their scopes and of the

legislation. Recommendations are not implementation rules. It should be emphasized that the focus

of this assessment is not a critical analysis of the recommendation but rather the resources

required to implement them.

Therefore any cost assessment will be based on individual interpretations of what is needed to fulfill

the recommendations in each member state. The assignment in this work was to ask the member

states for the cost related to implementing the full FAIRMODE recommendations.

The main uncertainties related to a cost estimate of the present formulation of the

recommendations are:

It is not mentioned in the recommendations at what scales models should be applied. E.g., are

the member stated required to run both regional models, urban background models, and urban

street models, in order to take into account the contributions from all these scales or can less do

The minimum requirements for chemical species included in the models are not mentioned –

e.g. is it required to model the full range of secondary organic aerosols, some harmful VOCs,

ultrafine particles and even particle number concentrations in the cities, or is it sufficient to

model ozone, nitrogen-dioxides, and some secondary inorganic aerosols?

The required quality, resolution, number of species and SNAP category sub-level of the

emission data are not addressed

The required quality of the model results at different scales is not mentioned.

In this work, we have assumed that state-of-the-art atmospheric modelling is a requirement. This

means that all scales (regional, urban background, urban street) should be covered and that the

chemical species presently included in state-of-the-art modelling are required, but not more than

that. We have also assumed that emission databases with a 1 km x 1 km resolution are needed.

They should at least be divided into the ten overall SNAP categories, and preferable some sub-

level categories as well (e.g. sub-division of SNAP category 8 – other mobile sources, or 2, non-

industrial combustion), while others are more well defined in the main SNAP (e.g. SNAP category

1, the major power plants).

We also assumed that atmospheric modelling at all scales (regional, urban background and urban

street) is a requirement, since it is not possible to take into account the effect of non-linear

14



atmospheric chemistry in e.g. cities, if the boundary conditions are not well described. We assumed

that member states need to run all scales to be able to e.g. produce assessments at all scales

within the countries – both in rural and urban areas. Furthermore, it is not obvious whether the

assessment of point sources should also be included in the requirements when producing e.g.

assessments and source allocations.

It is, of course, mentioned in the FAIRMODE recommendations that quality objectives should be

formulated for both models and emissions, but the resources needed to develop the models or

emission databases are very dependent on the quality required. E.g., is it required that all modeled

concentrations should compare with measurements at a +/- 30% level? And is that a requirement

for annual values or hourly values and at all scales, or should the requirements for the urban street

scale be different from the others, e.g. +/- 10%?

Likewise, in order to set up a precise baseline for how large a modeling group is needed in each

member state, it is necessary to make a interpretation as to the minimum-level scope of modelling

activities that would meet the recommendations. Are only a few persons needed to run regional

models to cover the basic chemical species and run them at a resolution, which can be decided by

the individual member state? – or is a very large modeling group needed, in order to be able to

model full mass closure of particle mass, where e.g. the speciation of the particles are fully

understood?

These questions call for an interpretation. Therefore, the minimum requirements must be defined

for building up a modeling group in countries, which presently do not apply modeling.

Discussions at the partner workshop in Denmark estimated the “baseline” to include 2 senior

scientists to cover regional modeling, 2 senior scientists to cover urban background and urban

street modeling and 1 junior staff member to run operational forecasts and take care of equipment,

e.g. Linux computers etc. Furthermore, it was estimated that one additional senior staff member is

needed to make gridded emissions at high resolution. All the member states already have the

obligation to report emissions at the national level and gridded emissions at a 50 km x 50 km

resolution every 5 years, according to the Conventions. Therefore, it was decided that just one

additional staff member is needed to produce high-resolution gridded emission data bases.

In this basic modeling group, we have not included a person to do point source calculations, since

this matter has not previously been a part of the EU directives. On the other hand, one can argue,

that the FAIRMODE recommendations are of such general nature that this kind of models and

calculations should also be included.

The number of senior scientists is also estimated from the fact that several persons are needed in

order to further develop the models or to ensure a living research environment. Additionally, it can

be argued that in order to ensure a living research environment, a PhD program is also needed with

at least one active PhD student at all time. The task of air pollution assessment or source allocation

is tightly linked with research, since there are still many open questions in this research area, e.g.

obtaining mass closure for particles for assessing concentration levels of PM2.5. On the other hand,

it is impossible to require in an AQD that member states all have a living research environment in

order to ensure high quality decision support.

If a member state already has a larger modeling group including all skills and scales, one could

argue, that the added cost would include one person dedicated at fulfilling the FAIRMODE

recommendations and one person dedicated at making high resolution gridded emissions, based

on what is already reported according to present obligations

15



In other member states which do not have a modeling community already, the main challenge is the

competence building with respect to atmospheric modeling, and the challenge of producing

emission inventories with a sufficiently high resolution and quality.

After many discussions we concluded, however, that the necessary minimum costs are actually the

same for all member states no matter their current level of development within atmospheric

modelling. Hence, in MS where the required modeling group is already in place, implementing the

recommendations will still be associated with costs due to the fact that the resources demanded for

implementation of FAIRMODE will be drawn from other uses. It is only a question of timing – i.e. the

time when the modelling groups in each country are ready to fulfill the recommendations. The main

challenge in all member states is an issue of competences. For all countries, the minimum

requirements are the competences of scientists at the senior level within the following four areas:

1. Regional scale modelling

2. Urban background modelling

3. Urban street modelling

4. High-resolution emissions

As the competences of one scientist seldom cover all aspects at the same time, the minimum

number of staff members needed is the same as the number of overall competences needed,

namely four. Furthermore, as the FAIRMODE recommendations do not specify e.g. the number of

assessments on an annual basis or the number of reports needed to make source allocations, the

required number of staff members has been set to this minimum. Therefore, we assume in the

following that the baseline cost for all member states are the cost of the above competences, as

well as equipment (in this case computers), operational expenses (as travelling) and overhead cost

to cover housing, management and secretaries, etc.

All the modelling groups need models. Well established modelling groups already have state-of-the-

art models, used on an operational basis as well as a part of the research goals in the groups. It is

of vital importance that the scientists have participated in the development of the models they use,

since only in this way, the scientists are aware of the models’ limitations and strengths. Therefore, it

does not make sense to introduce a community model that all modelling groups are required to use.

The key question is not the quality of the model but of the modeler using the model for obtaining

high quality decision support.

For countries not already including modelling groups, the key question is the competence building.

We have not included a cost assessment of acquiring the models, since it is possible to acquire

state-of-the-art models for free. For example, establishing a new modelling group would require

engagement of PhD students in all three modelling scales. These PhD students would be able to

work abroad as a part of their study visiting one of the well-established modelling groups in Europe

and they will be able to bring home the models they have been working on as part of their PhD

study. Models can also be acquired for free on the internet and therefore, we have set the cost for

models to zero.

When running atmospheric models, meteorological data are needed. In the following we assume

that meteorological data are freely available from the national weather services in each member

state. This is not always the case.

In the following the methodology and assumptions for the cost analysis based on both the bottom-

up approach and the top-down approach are discussed. Chapter 4 presents results.

16



3.2 The bottom-up approach: Information Request for each option and sub-

recommendation

In the period December 2012 - January 2013 an information request was established and sent to

the member states as a basis for the bottom-up approach. In Hasler et al., (2013), the common

setup of the member state information requests for FAIRMODE, AQUILA and SEG is described.

The specific information request formulation for FAIRMODE is reproduced in the subsequent

section.

The following pages until Section 3.3 is a reproduction of the main material in the information

request: the introduction presenting the background, and the main content of the questions posed.

The detailed questions on cost categories etc. are not reproduced here.

The information request required a slight reduction in the volume of the background text provided in

the FAIRMODE recommendations. However, a link to the full document was included, so those not

acquainted with the recommendations could read the full document. The information request had

the following content:

3.2.1 FAIRMODE Information Request for Each Option and Sub-Recommendations

This Information Request concerns a cost assessment of the recommendations from FAIRMODE

(The Forum for Air quality Modelling in Europe), which is a joint response action of the the

European Environment Agency (EEA) and the European Commission Joint Research Centre (JRC)

(see also: http://fairmode.ew.eea.europa.eu/). Its aim is to bring together air quality modellers and

users in order to promote and support the harmonised use of air quality (AQ) models by EU

member countries, with emphasis on their application to the European Air Quality Directive.

FAIRMODE has formulated a set of recommendations, which we ask you to assess in the following.

In overall terms, FAIRMODE strongly recommends the use of models for:

1. assessment of air quality levels to establish the extent of exceedances and establish population

exposure,

2. forecasting air quality levels for short term mitigation and public information and warnings,

3. source allocation to determine the origin of exceedances and to provide a knowledge basis for

planning strategies,

4. development and assessment of plans and measures to control AQ exceedances,

5. designing monitoring networks when models are used in combination with monitoring, and

6. determining the number of fixed monitoring sites that are required

Besides the above items, FAIRMODE recommends

1. a revision of the data quality objective for modelling.

2. that competent authorities for modelling activities are nominated by the Member States.

FAIRMODE will act as coordination forum for modelling, and

3. to investigate and improve the compilation, consistency and quality assurance of emissions

data suitable for AQ modelling under the directive.

For each Member State (MS) to meet the recommendations made by FAIRMODE, the MS needs to

fulfil a minimum standard within atmospheric modelling activities at all scales (regional, urban

background and urban street) as well as the possibility to assess strong point sources.

17



Furthermore, appropriate emission databases have to be developed and maintained and the

competent authority set up by the MS within modelling and emissions have to be operational in the

sense that they can support decision making within a sufficiently short time scale. In that sense the

competent authority for modelling and emissions need resources both for competence building,

model development, and building emission databases and for being able to respond operationally

within a short time frame to requests from decision makers within the MS.

Since, some MS already have competent authorities for modelling and emissions and some does

not, there will be large differences between the MS with respect to how much is required and the

need for building up the appropriate competences. We envisage that for many MS there will be two

phases in the process:

1. a build-up phase, where the competences within modelling and emissions are educated (e.g.

via PhD degrees in modelling or emissions including mobility exchange to already existing

modelling groups in Europe) as well as building up the necessary infrastructure (office and

computer facilities as well as collaboration agreements with universities and international

collaboration with respect to further education and research. As some MS already have the

appropriate competences, the build-up phase is not necessarily relevant for them.

2. an operational phase, where the competent authorities within modelling and emissions are

responsive with respect to the decision support e.g. for the MS EPA’s and operational in fulfilling

the FAIRMODE recommendations.

The following questionnaire is divided into the two phases listed above.

Firstly an assessment of the build-up phase for understanding the current baseline for your

member state, and

Secondly an assessments of the operational phase for compliance with the individual

FAIRMODE recommendations.

3.2.2 Understanding the current baseline for modelling practise for your Member State.

The first part of the questionnaire concerns an assessment of the build-up phase for understanding

the current baseline for your member state

As a working hypothesis for the minimum set of standards to be met by the MS e.g. concerning

quality and resolution of models, quality and resolution of emission data, as well as response times

for supporting decision making, we are in the following suggesting a set of minimum requirements

for a competent authority for modelling and emissions. This set of minimum requirements should be

interpreted as the minimum resources needed for developing and maintaining AQ models and

emission databases. We envisage that the following is needed to build up to meet the FAIRMODE

recommendations for building up competences, and modelling/emission groups within each MS

Description of issue: Competence building in the area of atmospheric science with focus on

atmospheric modelling and air quality modelling. Further development of models at regional,

urban background and street scale as well as strong point sources for national use. The

estimated minimum resources needed:

- Additional staff or staff time: 3-5 full time senior staff within atmospheric modelling

- Staff capacity development: PhD education of 3-5 persons within atmospheric modelling,

including mobility exchange to established modelling groups in Europe

- Additional equipment: 3-5 normal PCs

Does your Member State current conform to the standards or practices specified in the

minimum requirements listed above?

18



Description of issue: Competence building in the area of building high resolution emission

databases within the ten major emission categories (SNAP categories). The minimum

resources needed:

- Additional staff or staff time: 3-5 senior + 2 junior staff (including one database expert and

one GIS expert) full time staff

- Staff capacity development: education in e.g. geography, databases, GIS

- Additional equipment: 5-7 state-of-the-art PCs. Database software, GIS software.

- New data generation/compilation: Generation of national databases for the ten major SNAP

categories, e.g. traffic, power sector, agriculture, industry, etc.

- Additional reporting: Annual reporting of emissions

Does your Member State current conform to the standards or practices specified in the minimum

requirements listed above?

Description of issue: Further development or acquisition of atmospheric models – both on

regional scale, urban background and urban street scales as well as a model capable of

handling strong point sources.. The minimum resources needed:

- Additional software: Models can be downloaded freely, or acquired for a smaller amount.

Alternatively, existing models have to be further developed in 1) to comply with the


- Additional staff or staff time: 1 junior staff for technical implementation and maintenance of

models and computer facility.

- Additional data needed: high-resolution meteorological data with 1-hour time resolution as

input to AQ models e.g. provided by the national met office.

- Additional reporting: Annual reporting including annual evaluation of models



Description of issue: Computer facilities for running high-resolution models. The resources

needed:

- Additional equipment: 2-4 powerful 8-48 core computers, depending on required model

resolution.

- Running costs: housing, power and cooling for the computers



Description of issue: Operational/running costs: The resources needed:

- Additional administration: 1 half time economic officer, 1 half time staff leader

- Accommodation: Office space for 10-15 personnel

- Other: Travelling, mobility for scientists for longer periods (up to 6 months), attendance to

conferences.

- Operational costs for maintaining readiness for the staff’s ability for decision making support

- Running costs for maintaining an operational forecast system.



19



3.2.3 Cost assessment of the individual FAIRMODE recommendations.

The second part of the questionnaire concerns a cost assessment of the individual FAIRMODE

recommendations.

We here assume that you already went through the build-up phase explained in the previous

section, in the sense that a competent authority for modelling and emissions exists in your member

state. Therefore, the cost assessment here should be carried out on the basis that your member

state already comply with the minimum requirements for building up the modelling and emission

capacities in phase one.

We would like to know according to your judgement, to which extent your member state already

comply with the specific recommendations and what additional staff, competence building,

equipment, facilities, operating/maintenance cost, new data generation, reporting and

administration are needed to conform to the standards or practices specified in the specific

recommendations.

Option 1: ON THE USE OF MODELS FOR REGULATORY PURPOSE AND TO SUPPORT AIR

QUALITY POLICY

Description of issue: Models are widely used for the investigation and assessment of ambient air

quality at various spatial and temporal scales. In situ monitoring data is by its nature only a small

sample of the spatial distribution of air pollutant concentrations. Air pollution models can be used to

provide an assessment in areas where monitoring data are not available and thus, in combination

with monitoring data, provide a more complete assessment of the current air quality situation.

Models can also be essential for the development and assessment of the effectiveness of air quality

plans including measures to improve air quality. One of the major advantages of models is their

potential to provide detailed spatial distributions of air pollutant concentrations. This means they

can be applied for area-wide exposure assessments for human health and for the environment. An

application of models allows a much broader assessment of the extent of exceedances of air quality

environmental objectives and can also provide information required for improved measurement

network design. Furthermore, models can be used for short term forecasting of air quality as well as

emission scenarios for management and regulation of sources.

Recommendation 1(a): Use AQ models for assessment of air quality levels to establish the extent

of exceedances and establish population exposure

Recommendation 1(b): Use AQ models for forecasting air quality levels for short term mitigation

and public information and warnings

Recommendation 1(c): Use AQ models for source allocation to determine the origin of

exceedances and to provide a knowledge basis for planning strategies

Recommendation 1(d): Use AQ models for development and assessment of plans and measures

to control AQ exceedances

Recommendation 1(e): Use AQ models for designing monitoring networks when models are used

in combination with monitoring

Recommendation 1(f): Use AQ models for determining the number of fixed monitoring sites that

are required

20



Option 2: MODEL QUALITY OBJECTIVES (MQO)

Description of issue: Data quality objective for modelling (MQO) are mentioned in the 2008 Air

Quality Directive, Annex I, but the wording of the text remains ambiguous and open to

interpretation. FAIRMODE is proposing to develop new data quality objectives for modelling for

ambient air quality assessment, in collaboration with the Member States. This type of objectives is

expected to be useful as basis to investigate MQO for the other model applications. Furthermore,

FAIRMODE proposes that subsequent the work of FAIRMODE the European Commission initiates

a process for the preparation of a Guidance document on the revision of model quality objectives

for assessment.

Recommendation 2(a): FAIRMODE recommends a revision of the data quality objective for

modelling.

Option 3: FORUM OF EU AQ REGULATORY MODELLING

Description of issue: FAIRMODE recommends that in parallel to what has already been

established for the monitoring of air quality, competent authorities for modelling activities are

nominated by the Member States. FAIRMODE will act as coordination forum for modelling and

support the competent authorities in activities that are recognised as very relevant for the for model

applications listed in Recommendation #1 and within an extended list of supporting motivations,

including model evaluation, combine use of model and monitoring data, source apportionment

modelling, and monitoring station characterisation.

Recommendation 3(a): FAIRMODE recommends that in parallel to what has already been

established for the monitoring of air quality, competent authorities for modelling activities are

nominated by the Member States (ref Article (3) and bullet d) quality assurance of modelling)

Option 4: QUALITY ASSURANCE AND CONSISTENCY OF EMISSION INVENTORIES

Description of issue: Air quality emissions inventories (EIs) have been compiled at European level

for regulatory purposes and also to support air quality modelling applications for the assessment

and improvement of air quality. Emission information is essential in support for AQ planning under

the AQ directive because it provides the link between responsible emission sources, their relative

shares and abatement potentials. FAIRMODE proposes that 1) Emissions are not mentioned in the

AQD and the need to work to increase the quality of emission inputs needs to be introduced in the

revised text. 2) Promote guidance initiatives for the compilations of emission data for AQ models

under the directive, and 3) Support competence building initiatives to secure the consistency of

detailed bottom-up emission inventories with those compiled for regulatory purposes at local,

national and European scale.

Recommendation 4(a): FAIRMODE recommends to investigate and improve the compilation,

consistency and quality assurance of emissions data suitable for AQ modelling under the directive.

3.3 The top-down approach: expert estimates calibrated to each member state.

As argued in section 3.1, the basic requirement for all member states to fulfil the FAIRMODE

recommendations is the competences of four different staff member. Furthermore, we have

concluded that by virtue of this basic group, the staff is able to fulfil the recommendations to a

certain extent, depending on the amount of assessment or reporting needed. For countries already

having well-established modelling groups, it is our experience in Denmark that the four persons can

easily be occupied by the work suggested in the FAIRMODE recommendations and therefore the

additional cost equals the cost of the baseline. For member states, which do not already have a

21



modelling group, but have to build it up from scratch, the requirements are the same. If the

expertise is not available already in the member state, at least a four year period is needed for

starting up the modelling group, engaging and finishing at least three PhD students covering the

three different modelling scales. After this four year period, it is assumed that these modelling

groups are operational and able to fulfil the FAIRMODE recommendations to the same extent as

the well-established modelling groups.

The top-down approach is based on experiences in Denmark, which includes a well-established

modelling group with a long history. The modelling group includes around 10 senior scientists,

working with research, development and application of models at regional, urban background and

urban street scales as well as point sources for regulation. Furthermore, the group includes typical

up to around 5 PhD students working with special challenges and scientific questions related to air

pollution modelling at all scales. The modelling group has a long expertise in decision support to the

Danish authorities, including ministries, the Danish EPA and municipalities, as well as private

industries and international authorities. Besides that, Denmark has an emission group of around 10

staff members, working primarily with fulfilling the obligations of conventions with respect to climate

and air pollution. Some of this staff is working on high-resolution emission data bases at a

resolution of 1 km x 1 km in Denmark, which is used as input to the models and supporting the

decision support. These two groups are both a part of Department of Environmental Science

(ENVS), Aarhus University (AU). Furthermore, ENVS are responsible for the Danish monitoring

programme related to air pollution, which generates a living research environment where

monitoring, atmospheric modelling, research, education and decision support to authorities and

private industries go hand in hand.

In the following we define the cost assessment for the baseline for Denmark, which is used to

calibrate all other member states in the top-down approach, while applying different tariffs within the

individual countries. As mentioned, we assume that at least four people with four different

competences are needed to fulfil the FAIRMODE recommendations. The cost estimates are given

in Table 3.1.

22



Table 3.1 The annual baseline cost in Denmark for fulfilling the FAIRMODE recommendations, assuming

three senior scientists in atmospheric modelling and one senior staff working on high-resolution

emissions.

Cost category Modelling part

[Euros]

Emission part

[Euros]

Comment

Senior staff 239200 79733 4 persons in total1

Junior staff Support is included in OH

Equipment, PC's 4500 1500 4 normal PC’s, lifetime 3 years2

Licenses 2000 1000 Databases, compilers, GIS and

visualisation software3

Computer servers 30000 3 servers, life time 3 years4

Operational expenses, travelling 4500 1500 4 travels annually5

Training, competence building 2000 667 Basic training, courses6

Sum 282200 84400

OH, administration 70550 21100 25% Overhead (OH)7

Total Euros (annually) 458250

Typical tariffs for staff members at research institutes in Denmark are assumed in this baseline. An

overhead of 25% is assumed according to the EU standard cost model. In Denmark, however, an

overhead rate of 116% is used. All figures in Euros. Extended comments to the individual costs are

given below.

1 The general assumption is that in order to fulfill the FAIRMODE recommendations, 3 additional modelling senior scientists and

1 emission senior staff member are needed. It is assumed that all atmospheric models can be obtained for free - if they are

not already available. The cost is given without overhead, which is added in the bottom of the table. 2 One normal PC per staff member. One PC has the cost of around 1500 Euros and has a life time of around three years

3 Licenses are needed for software for the PCs and computer servers - GIS software, database software, Fortran and C

compilers, visualization tools for visualizing model results 4 At least 3 fast computer servers for model calculations are needed. One server is needed for making operational forecasts and

2 servers are needed for assessment and source allocation modelling, etc. The price for one server has been estimated to

10000 euros, for which a powerful server can be acquired, including at least 48 cores. 5 Travelling is assumed for participating in meetings related to FAIRMODE. The cost is estimated to 1500 Euros per travel, 4

travels in total per year. This cost does not include participation in conferences, etc. but does only cover participation in

FAIRMODE meeting for e.g. quality objectives and the modelling forum. 6 Annual cost for training of staff member - at ENVS we have a budget of around 667 euros per staff member per year on

average. 7 An overhead of 25% is assumed according to the EU cost model. The overhead is assumed to cover housing, management,

secretary, etc. The assumed overhead of 25% is considerably lower than the overhead of 116% normally used in

Denmark.

23



Table 3.2: The baseline cost in Table 3.1 distributed across the 5 baseline categories (B1-B5) as defined

in the information request, see section 3.3

Modelling Emissions Models Computing OH

B1 B2 B3 B4 B5

Equipment 4500 1500

30000

Facilities

45825

Operation 4500 1500

Maintenance

Data Generation

Data Compilation 2000 1000

Reporting

Administration

45825

Additional Staff Costs 239200 79733 0

Staff Capacity Development 2000 667

SUM 252200 84400 0 30000 91650

Total of the above 458250

In Table 3.3 the figures in Table 3.1 and 3.2 are attempted distributed over the individual

FAIRMODE recommendations (options). In this context it is extremely important to note that the

costs listed in Tables 3.1, 3.2 and 3.3 cannot be added; hence, the tables basically display three

different ways of distributing the same total costs across more specific activities. The distribution

within recommendations 1a-1f is subject to uncertainty, since it is very difficult to distinguish the

cost in the individual recommendations from each other. All the options require the above

mentioned competences, and computers. The distribution is chosen according to experiences in

Denmark for solving the different tasks within the recommendations 1a-1f. The working hours in

recommendation 2 are estimated time used for participating in defining data quality objective for

modelling (MQO). The recommendation 3 and 4 includes travelling to meetings with respect to the

FAIRMODE forum and for quality assurance and consistency of emission inventories. Computers

and software licenses are distributed evenly over recommendation 1, except for recommendation

1b, which requires one computer server allocated full time for producing operational air pollution

forecasts. All staff cost are given in hours - all other numbers in Euros. The number of hours

includes 4 people full time in total, assuming 1300 working hours per year. Overhead cost is not

included in this table.

24



Table 3.3 The baseline cost in Tables 3.1 and 3.2 distributed in the FAIRMODE recommendations

(options). There is no overhead cost in this table - all figures are without OH, which should be added

Option 1a 1b 1c 1d 1e 1f 2 3 4

Equipment (Euros) 5200 10000 5200 5200 5200 5200

Facilities

Operation (Euros) 4500 1500

Maintenance

Data Generation

Data Compila-tion

(Euros)

500 500 500 500 500 500

Reporting

Administration

Additional Staff Costs

(hours)

600 500 1300 1300 500 300 200 300 200

Staff Capacity

Development (Euros)

444 444 444 444 444 444

Man-years 0.46 0.38 1.00 1.00 0.38 0.23 0.15 0.23 0.15

In all tables 3.1-3.3 above, the numbers add up to the same totals. We assume that the baseline

cost equals the additional cost for implementing FAIRMODE recommendations in all countries as

explained in the previous sections.

We assumed that facilities, maintenance and administration is a part of the overhead cost, which is

not added in table 3.3, since the calibration with other member states are made without overhead.

We assume that reporting is a part of the overall staff cost. There are no clear requirements given

by FAIRMODE on reporting.

No specific cost is assigned to data generation, since generation of data is covered by computer

cost and staff cost.

It is also assumed that meteorological forecast data and analysed data can be obtained freely e.g.

from national weather services.

All the cost for recommendations 1a-1f should be added as one option, since it is very difficult to

distinguish the cost under the individual sub-options - they are all connected.

We assume that the total requirements are equal to all countries - no matter whether they already

have an operational modelling expertise or not. It is a matter of timing - if the recommendations are

made mandatory, the countries, which do not already have operational modelling expertise, will

have to use the same resources to build up the expertise - e.g. over a 4 year period, but with the

same cost per year. The cost for training PhD student equals the cost of a senior scientist, when

the administrative cost to the university and cost to supervising, travelling, etc. are included.

25



4 Cost Assessment

In this chapter, results from the cost assessment are described, based on both the bottom-up and

top-down approaches. The bottom-up approach is based on data from the information request to

the member states carried out in December 2012 - January 2013. Results from the top-down

approach are based on expert estimates of minimum requirements for implementing the capability

and competences for fulfilling the FAIRMODE recommendations. This expert estimate is adjusted

according to the EU standard cost model and subsequently calibrated to all member states with

respect to salary levels in the individual member states.

4.1 The bottom-up approach: Information request for each option and sub-

recommendation

4.1.1 Overview

There were responses to the information request concerning FAIRMODE from 17 countries. The

questions concerning quantitative cost estimates were filled in to some extent by 12 countries. No

countries except Denmark provided costs for all of the items. Table 4.1 provides a brief overview,

while Table 4.2 provides some further detail.

Table 4.1: List of countries which responded. It is indicated whether any quantitative data are supplied.

Country Quantitative data

Austria y

Belgium y

Croatia y

Czech Republic y

Denmark y

Finland

France

Germany

Hungary y

Ireland y

Italy y

Lithuania y

Netherlands y

Poland

Romania y

Sweden

UK

26



Table 4.2 Overview of responses to the information request

27



Overall, the results of the information request supply the following information

Some indications of costs for fulfilling the baseline requirements and for the recommendations.

However, the raw numbers which can be extracted from the questionnaires cannot be taken at

face value. This is discussed in more detail in subsequent sections.

Much useful qualitative, textual information.

Information as to which extent the countries fulfil the stipulated minimum requirements.

As a supplement to the information request case studies were conducted for three countries. For

these countries phone interviews were carried out in order to substantiate the information given in

the information request and to learn in greater detail what has motivated the responses. The

countries selected for the case study were Belgium, Croatia and Hungary. The information derived

from the case studies is used in the subsequent discussion. Summaries of the information obtained

in each case study are reproduced in Appendix B.

4.1.2 Introduction to results

The structure of the questions in the information request was as follows:

5 questions concerning the baseline requirements. The general question was: Does your

Member State currently conform to the standards or practices specified in the minimum

requirements listed above? Respondents were asked to clarify their response (thus providing

qualitative information) and to supply quantitative information on

- Current costs

- Change in costs in order to fulfil minimum requirements.

9 questions concerning the FAIRMODE recommendations (recommendations 1a-1f, 2, 3 and 4).

The general question was: Does your Member State currently conform to the standards or

practices specified in the recommendation listed above? Respondents were asked to clarify

their response (providing qualitative information) and to supply quantitative information on

- Current costs

- Change in costs in order to fulfill the recommendation.

In the process of analysing the incoming data, three items were generated for each question:

A graph representing the costs that can be derived from the responses.

A compilation of notes to interpret the values

A summary of the textual response to the question (“Please clarify your response”).

The graph is produced in accordance with the framework set up in Hasler et al. (2013).

Specifically, this implies that when the respondent has indicated labour costs in terms of staff time

the costs are calculated according to tariffs, whereas when a quantification is provided in monetary

terms (e.g. payment to a company for an outsourced service), the cost enters directly into the

graph.

The compilation of interpretative notes is necessary in order to understand what the graphs include.

E.g., if a respondent leaves a field intended for a cost blank, this will appear in the graph the same

way as a cost of zero. The notes serve to indicate whether zero should be interpreted as zero or as

missing. Also, a quantitative estimate may be partially filled in, so the apparent cost is an

incomplete representation of the total costs. Such interpretative notes have been assigned to each

question and each country, and were used in the analyses. The overview given in the next section

is a simplified version of the interpretative notes.

Examples of the items described above are presented in the following discussion, while a complete

collection of graphs and summary of textual responses can be found in appendices.

28



Table 4.3 below shows a summary of all costs that were reported in the responses to the

information requests.

Table 4.3: Bottom-up approach. Total cost by baseline and recommendation.

4.1.3 Discussion of selected results

Subsequently, results are presented for some selected questions. The presentation has the form of

graphs with accompanying notes. In appendices there is similar information for all questions.

Appendix D presents graphs for all questions, while Appendix C presents summaries of textual

responses to each question.

The questions selected for discussion here are:

Baseline 1: Competence building – atmospheric science

Baseline 2: Competence building – high resolution emission data bases

Recommendation 1a: FAIRMODE recommends use of models for assessment of air quality

levels to establish the extent of exceedances and establish population exposure

Recommendation 2: Revise Model Quality Objectives

In addition, the results of groups of questions are presented in graphical form and accompanied by

notes. The following groups are considered:

Current costs pertaining to the sum of all 5 baseline questions

Total costs (current + required) pertaining to the sum of all 5 baseline questions

Current costs pertaining to the sum of all 9 recommendations

Total costs (current + required) pertaining to the sum of all 9 recommendations

Baseline 1: Competence building – atmospheric science

Baseline Cost Costs for the different options

MS Baseline 1 Baseline 2 Baseline 3 Baseline 4 Baseline 5 option1a option1b option1c option1d option1e option1f option2 option3 option 4 total cost

Denmark 188476 63159 0 11016 0 30288 27478 62295 62295 25715 16571 9145 18217 10645 525300

Belgium 815000 1325000 1285,22998 74000 1035000 0 0 0 0 3250285

Bulgaria 0

Czech Republic 518959 241773 95155 95284 83769 7638 27756 15821 989 0 353 0 0 1087498

Germany 0 0 0 0 0 0 0 0 0 0 0

Estonia 0

Ireland 254422 9460 263883

Greece 0

Spain 0

France 0 0 0 0 0 0 0 0 0 0 0 0

Italy 5080000 1640000 1315000 1000000 2480000 220000 165000 210000 300000 255000 140000 30000 115000 12950000

Cyprus 0

Latvia 0

Lithuania 11636 7130 0 0 2200 0 0 0 0 0 0 0 0 20966

Luxembourg 0

Hungary 66422 138336 31957 14284 16421 53854 7486 5816 48615 0 0 11111 11111 0 405415

Malta 0

Netherlands 0 0 0 0 0 2810000 130000 0 0 0 0 1025000 3965000

Austria 441908 325527 3856 40352 0 16947 0 110161 0 0 938750

Poland 0

Portugal 0

Romania 75011 13064 25520 0 113595

Slovenia 0

Slovakia 0

Finland 0 0 0 0 0

Sw eden 0 0 0 0 0 0 0 0 0 0 57117 0 57117

United Kingdom 0

Croatia 364291 632043 30106 1007478 216728 179951 133495 18952 57029 9000 9000 9000 0 2667073EU+1 7741113 4447980 1486566 2231438 3872270 3303930 352622 267885 489089 334715 280924 169256 116445 1150645 26244879

29



Figure 4.1 Sum of current + required costs to fulfil the requirements indicated in the Baseline 1

question on Competence building – atmospheric science

For Baseline question 1 the quantitative information from Denmark, Czech Republic, Ireland, Italy

and Croatia can be considered complete, although it should be recognized that any such

information cannot be more than uncertain estimates.

For other countries which have supplied quantitative information some costs elements are missing.

For instance Belgium has an administrative structure with an interregional environmental agency

plus 3 regional agencies. Costs are estimated for current costs, but concerning costs to reach the

required baseline it is indicate that the information is incomplete: “Some of the regions indicated a

change in costs to conform to the baseline 1 standard but no further details of the costs could be

provided at the moment.”

The cost for Italy is strikingly high (5 million Euros). These are current costs, and Italy has no need

for additional costs in order to comply with the baseline requirements. The figure is accompanied by

the following clarification: “ENEA (National Agency) supports Ministry of the Environment in

developing and maintaining the Integrated Atmospheric Pollution National Model. ENEA has team

on atmospheric research with about 15 staff researchers, research fellowship and temporary

researcher. They regularly attend to international group (including Fairmode) and have links with

foreign research groups.”

Several countries state that they fulfil the requirements, but do not provide cost estimates for this

question. This is the case for Finland, France, Netherlands, Germany, Romania and UK.

The textual clarifications to the baseline 1 question are quite informative concerning the situation in

the countries. They are reproduced below. A complete collection of responses for all questions can

be found in Appendix C.

BASELINE 1: COMPETENCE BUILDING – ATMOSPHERIC SCIENCE

Does your Member State currently conform to the standards or practices specified in the minimum

requirements listed above? Please clarify your response

Austria No. At the Austrian EPA there is currently no staff dedicated to modelling issues as there are no legal

requirements.

0

1000000

2000000

3000000

4000000

5000000

6000000

Total cost of baseline 1 - bottom-up approach

Total baseline costs

Capacity development

Additional staff

Reporting and administration

Data generation and compilation

Operation and maintenance

Equipment and facilities

30



Belgium This competence building is available at the different regional environment administrations and the Belgian

interregional Environment Agency. Subcontractors (like e.g. the Flemish Institute for Technological

Research) are also playing an important role in air quality modelling in Belgium. Some of the Regions

indicate that competence in atmospheric modelling and air quality modelling will be further developed, but

additional efforts have to be made to achieve the specified minimum requirements. The following

calculation is an estimation of the existing annual costs of baseline 1. Some of the regions indicated a

change in costs to conform to the baseline 1 standard but no further details of the costs could be provided

at the moment.

Croatia Our member state conform to the specified standards when whole country is taken into the consideration

and expert employed in different institutions are accounted. However, a team of at least 5 scientists

working in the field of atmospheric modelling should be developed in one institute that will be able to work

together and develop competence in atmospheric modelling to provide data for reporting on national level,

planning and other MS obligations toward EU.

Finland Finland meets the requirements. There is a Air Quality modelling group that develops the atmospheric and

air quality models for different scales (local, urban, regional, global..) (~20 Research Scientists and PhD

students) in FMI. In addition, there is Air Quality expert service group that offers dispersion modeling

services for various customers (Cities, industry etc..) in Finland and abroad.

France yes definitively. At the national level a team of 12 people (more than half of them being senior scientists) is

entirely devoted to air pollution modelling research and expertise. This team belongs to INERIS and

participates to the national reference laboratory for air quality (LCSQA). It works for the Ministry in charge

of the Environment. At the local level, associations in charge of air quality monitoring develop skill and

competences in air pollution modelling as well. A staff of about 30 local experts works can be mobilized.

Note that INERIS develops research projects in the fields of AQ modelling and hosts a number of PHD

students. On the material point of view, it benefits from high performance computing resources.

Germany Practice in Germany is contracting consultants in most of the cases. Answers rely on assuming that

presumption made for this questionnaire is proper.

Hungary Not yet, there is only 1 full-time senior staff within atmospheric modelling, working with only one 10-year

old PC. Because of serious restrictions in the budget, there is no possibility to employ any more staff

persons and to purchase new computers at the moment.

Ireland No. Currently there is no capacity available for modelling. 2013 will see a full time fellowship devoted to

developing the specified criteria in CAFÉ with regard to modelling components of the Directive

Italy Yes, it does. ENEA (National Agency) support Ministry of the Environment in developing and maintaining

the Integrated Atmospheric Pollution National Model. ENEA has team on atmospheric research with about

15 staff researchers, research fellowship and temporary researcher. They regularly attend to international

group (including Fairmode) and have links with foreign research group

Netherlands Yes, the NL spends more than the 3-5 full time persons for atmospheric modelling under 1). Dutch

research institutes and Universities together employ a number of PhD in atmospheric modelling but not

specifically for national use. Since NL already uses modelling for reporting AQ data, we do not fill out the

baseline sheets for building capacity but only the sheets on maintenance of the modelling in sheets under

Options.

Poland It is not known on what basis was estimated the above minimum requirements. Each Member State has its

own approach to the problem of human resources. For example in CIEP (Chief Inspectorate of

Environmental Protection) modelling issues are carried out on behalf of the CIEP by external institutions.

One full-time employee of the doctoral title (responsible for modeling) is currently employed in CIEP.

Romania YES

Sweden Yes, the Swedish Meteorological & Hydrological Institute (SMHI) have significant competence with regard

to atmospheric & air quality modelling. This far exceeds the estimated minimum resources above. Other

institutes and consultants also have a good degree of competence in air quality modelling.

UK Modelling is already used for compliance reporting (alongside measurement data for assessment of extent

of exceedance), short term forecasting, source allocation, development of plans and measures, network

design and establishment of monitoring requirements. Its not clear how this data gathering exercise is

relevant to the FAIRMODE recommendations or the review of the Directive or where the estimates have

been produced from.

31



Baseline 2: Competence building – high resolution emission data bases

Figure 4.2: Sum of current + required costs to fulfil the requirements indicated in the Baseline 2

question on Competence building – high resolution data bases.

The responses to the Baseline 2 question should be interpreted with care, because the competence

addressed in the questionnaire concerns high resolution emission data bases for modelling,

whereas countries already have an obligation to produce national inventories (low resolution data).

It can be difficult to distinguish between the two types of costs. Thus, only two countries – Denmark

and Austria – have delivered numbers which specifically represent the costs of producing high

resolution databases.

For the countries with the highest costs the situation is as follows:

Belgium has included some costs pertaining to producing national databases, but on the other hand

the indicated costs are incomplete, because some regions do not have the ability to produce high

resolution databases and could not provide a cost estimate.

For Italy, it is not quite clear to which extent high-resolution activities are included. The clarification

reads: “Though the National inventory is provided by ISPRA, ENEA performs many activities in

order to prepare the national inventory (annual on a county base) for feeding the atmospheric

pollution national model (hourly grid base inventory, speciation, etc)”

For Croatia, much work is required in order to comply with the requirement, because much

statistical information will have to be collected more or less from scratch. This can be contrasted

with the situation in a country like Denmark, where plenty of statistical information is readily

available and can be used for preparation of high resolution emission inventories.

The following countries have indicated that they currently possess capability to produce high

resolution emission databases: Denmark, France, Germany, Italy, Netherlands, Romania, Sweden

(and presumably Finland and UK).

0

200000

400000

600000

800000

1000000

1200000

1400000

1600000

1800000Total cost of baseline 2 - bottom-up approach



Additional staff





32



Recommendation 1a: FAIRMODE recommends use of models for assessment of air quality levels

to establish the extent of exceedances and establish population exposure

Figure 4.3: Sum of current + required costs to fulfil recommendation 1a: FAIRMODE recommends use

of models for assessment of air quality levels to establish the extent of exceedances and establish

population exposure.

As mentioned elsewhere in the report, a basic difficulty is that the FAIRMODE recommendations

have not been sufficiently developed within a policy context to determine how they can be

implemented. For this reason, quantitative data on the recommendations are scarce, and the

estimates depend on the respondent’s interpretation of them.

In the case of recommendation 1a, The Netherlands indicate a current cost of almost 3 million

Euros. The accompanying clarification reads: “The Netherlands reports AQ data on the bases of

measurements and modelling. The Netherlands uses AQ models for scenarios studies and

effectiveness of measures. The Netherlands has an operational (short term) AQ forecasting

system. Considering costs for the Assessment and planning of AQ levels: this is a rather complex

system with many partners. We have the feeling that the cost items asked here would require a

major operation to get costs of every organization for these items. Instead we tried to make

estimates and sort of fitted it into your items. So this is a very crude estimation.”

Thus, the costs reported by the Netherlands refer to the full costs of a very well developed system –

whereas the costs for a country like Denmark refer to maintenance of the capability to produce a

limited number of assessments.

0

500000

1000000

1500000

2000000

2500000

3000000Total cost of option 1a - bottom-up approach



Additional staff





33



Recommendation 2: Revise Model Quality Objectives

Figure 4.4: Sum of current + required costs to fulfil recommendation 2: Revise Model Quality

Objectives

FAIRMODE’s recommendation of revising the Model Quality Objectives is not meant as a request

for an immediate revision of the Directive, but requests a process where Fairmode works on

developing new data quality objectives, whereupon the Commission is requested to initiate a

process for developing a guidance document.

Only four countries indicate a cost associated with the recommendation of revising the Model

Quality Objectives, but many have comments (which can be found in Appendix C). The UK states:

“This work is currently in progress and is not sufficiently developed for an assessment to be made

of the possible cost associated with new data quality objectives.” Italy indicates a cost of 140000

Euro, and clarifies: “At the present fully involved in Fairmode activities. MQO are tested

continuously.”

There are several positive comments on developing new MQO (“We support and are involved with

the work that FAIRMODE is undertaking...” – UK and others), while Austria is very sceptical: “We

do not recommend to change the MQO. The implementation of new, stricter MQO (DELTA-tool in

its current version) will certainly increase time and effort in respect to modelling in Austria due to the

complex topography a lot. Additional costs can't be estimated, as it is not clear whether state-of-

the-art models are even able to provide results meeting the new MQO.”

0

20000

40000

60000

80000

100000

120000

140000

160000Total cost of option 2 - bottom-up approach



Additional staff





34



Costs pertaining to the sum of all 5 baseline questions

Figure 4.5: Sum of current costs to fulfil the requirements for all 5 baseline questions

Figure 4.6: Sum of current + required costs to fulfil the requirements for all 5 baseline

questions

Figure 4.5 and Figure 4.6 above are two related graphs. They both refer to the sum of baseline costs:

One displays current costs, and the other displays current plus required costs.

35



The graphs give to some extent an overview of the degree, to which the various countries fulfil the

baseline requirements. The two graphs resemble each other, but they differ somewhat, as the latter

includes required costs in order to fulfil the baseline requirements.

The following notes give – country by country – an overview of the situation in the country and of

how the cost estimates should be interpreted. Only countries which provide cost estimates are dealt

with here.

In summary, baseline costs for the following 5 countries can be regarded completely represented:

Croatia, Austria, Italy, Czech Republic and Denmark.

Croatia

The graph gives representation of all costs. Current activity on modelling is limited. Build-

up of high resolution emission database is estimated to be very resource demanding

because it starts from scratch.

Romania

Costs are only filled in for a few baseline activities, so the graph is very incomplete.

Austria

Graph can be taken at face value. Austria has currently no modelling activity. The

respondent has filled costs guided by the stipulated requirements, but without details

such as travelling.

Hungary

Very limited current activity (one person). Level of activity has been reduced due to the

crisis.

The graph represents current costs, whereas it does not include necessary costs to

conform to minimum requirements.

Lithuania

Has some modelling capacity, but does not fulfil the requirements. The main need is extra

staff, especially for emission inventories.

Italy

Italy fulfills the minimum requirements. ENEA has a staff on 15 working with modelling,

which is reflected in the graph. Italy has the largest indicated cost of almost 12 million

Euro.

Ireland

Has no modelling activity at present. Costs are only estimated for Baseline 1, so the

graph does not give a complete view of costs.

Czech Republic

In the Czech Republic CHMI has modelling competences, but does not quite meet the

stipulated minimum requirements. The graph represents all costs associated to meet the

minimum requirements, both current and necessary additions.

Belgium

36



There is an interregional environmental agency and three regional agencies. The

modelling capacity varies between regions. The baseline costs reflect modelling activities

within the agencies and an outsourced contract for one region.

Denmark

The graph reflects all costs associated to the minimum requirements.

The concept: Degree of fulfilment

Figure 4.5 and Figure 4.6 resemble each other, but they differ somewhat, as the latter includes

required costs in order to fulfil the baseline requirements.

For each country, the ratio between current costs and total costs should in theory be an indication

of the “degree of fulfilment” – in other words, it expresses to what extent the country possesses the

required minimum competences as specified in the baseline questions. However, the ratio can only

be interpreted in this way if costs are properly filled in for all questions. This is not always the case,

so the computed ratio should in general not be taken at face value.

Nevertheless, the concept of “degree of fulfilment” can be useful. Table 4.4 indicates the degree of

fulfilment assessed by three methods.

Firstly, the degree of fulfilment can be computed directly as a raw value based on the numbers

underlying the graphs.

Secondly, the estimate can be refined by taking into account all of the additional qualitative

information which is represented in the responses to the information request.

Thirdly, it is possible to produce an expert estimate of the potential for fulfilling the requirements.

This can be done even for countries which have not responded to the information request. Such an

estimate must be based on various available information (participation in Fairmode, participation in

conferences, scientific production within the field, web searches for relevant terms etc.). It is the

best option available for filling the data gaps in a study like the present.

We have produced such expert estimates for the countries which have not responded to the

information request. In doing this it has been necessary to make the simplifying assumption that

administrative barriers are disregarded. The method considers the estimated capacity in a country,

regardless of whether the capacity lies within a central unit or is present locally.

Thus, if a country has an active modelling group, which has the capability to perform modelling at

all scales, we have assigned a high degree of fulfillment potential to the country, while disregarding

that there may be administrative barriers to apply this expertise within all regions of the country.

In order to obtain comparable values for countries which responded to the information request and

those which did not, we have reconsidered the “Refined estimates” that were produced for

responding countries. We have revised the estimate, using the principle of disregarding

administrative barriers which was applied in the third method (expert estimate). This makes a

difference for a country like Belgium, where the required competences are present within the

country, but not within all regions of the country.

In summary, this yields a maximum of three estimates for the “degree of fulfillment” for each

country:

IR Raw (based on raw values from the information request)

37



IR Refined (based on both values and qualitative information from the information request)

Expert estimate (includes countries which did not deliver information)

The three estimates are presented in Table 4.4 and Figure 4.7.

Table 4.4: Degree of fulfilment of the minimum requirements (percent)

Country Quantitative data IR raw IR refined Expert estimate

Austria Yes 0 10 30

Belgium Yes 100 80 100

Bulgaria No response

30

Croatia Yes 57 50 50

Cyprus No response

10

Czech Rep Yes 76 60 60

Denmark Yes 100 100 100

Estonia No response

40

Finland

100 100

France

100 100

Germany

100 100

Greece No response

100

Hungary Yes 37 20 20

Ireland Yes 0 0 0

Italy Yes 100 100 100

Latvia No response

40

Lithuania Yes 12 30 30

Luxembourg No response

20

Malta No response

10

Netherlands Yes 0 100 100

Poland

60

Portugal No response

70

Romania Yes 0 70 70

Slovakia No response

30

Slovenia No response

50

Spain No response

100

Sweden

0 100 100

UK

100 100

38



Figure 4.7: Degree of fulfillment of the minimum requirements (percent).

The graph gives an indication of the various countries’ capacity to fulfil the baseline requirements. It

should be understood that the graph shows minimum requirements. Fulfilment of these

requirements does not ensure that there is capacity to perform modelling assessments for every

corner of a country.

Costs pertaining to the sum of all 9 recommendations

0 20 40 60 80 100

Austria

Belgium

Bulgaria

Croatia

Cyprus

Czech Rep

Denmark

Estonia

Finland

France

Germany

Greece

Hungary

Ireland

Italy

Latvia

Lithuania

Luxembo…

Malta

Netherlands

Poland

Portugal

Romania

Slovakia

Slovenia

Spain

Sweden

UK

IR raw

IR refined

Expert estimate

39



Figure 4.8 Sum of current costs to fulfil all 9 recommendations

Total costs (current + required) pertaining to the sum of all 9 recommendations

Figure 4.9 Sum of current + required costs to fulfil all 9 recommendations.

Figure 4.8 and Figure 4.9 above are related. They both refer to the sum of costs for

recommendations: One displays current costs, and the other displays current plus required costs.

Many countries consider the implementation of one or more of the recommendations to be included

in the baseline costs. Other countries find that it is not possible to assess the costs of

recommendations. For these reasons a graphical representation of costs of recommendations is

missing for most countries.

40



The following notes give – country by country – an overview of the situation concerning the

recommendations in the country and of how the cost estimates should be interpreted. Only

countries which provide cost estimates are dealt with here.

In summary, no country except Denmark has indicated costs for all recommendations. The costs of

the recommendations appear very diverse, reflecting that there is not a unified interpretation of how

the recommendations could be implemented.

Croatia

Current activity on modelling is limited. Croatia has indicated costs for some recommendations,

while costs for others are indicated as unknown.

Sweden

The only cost specified is for Recommendation 3 concerning administration of National Reference

Laboratory

Austria

Under the assumption that the baseline is present (which it is currently not), Austria considers many

recommendations cost-neutral. Additional costs are specified for forecasting (1b) and assessment

of plans (1d).

For Recommendation 2 (revised MQO) it is noted that costs are unknown and may be high.

Netherlands

The Netherlands fulfill basic requirements. They carry out reporting to the Commission based on

measurements and modelling. Reporting is a complex system with many partners; an estimate of

the current costs is made (recommendation 1a). An estimate for forecasting (recommendation 1b)

and emission inventory quality assurance (recommendation 4) is made, whereas costs for

recommendations 1c, 1d, 1e and 1f are considered to be part of recommendation 1a. Estimates for

recommendations 2 and 3 (MQO and FAIRMODE cannot be made).

Hungary

Very limited current activity (one person). Level of activity has been reduced due to the crisis.

The graph represents current costs for a few recommendations where work is ongoing. It does not

include necessary costs to fulfil the Fairmode recommendations.

Lithuania

Has indicated that all recommendations are fulfilled, but without any specifications. Neither costs

nor additional costs are indicated. The responses should probably not be taken at face value.

Italy

Italy has specified costs for all recommendations except for forecasting (1b) and some expected

additional costs for quality assurance of emission inventories (recommendation 4).

Ireland

Has no modelling activity at present. Costs are only estimated for forecasts (recommendation 1b).

Czech Republic

41



In the Czech Republic CHMI has modelling competences, but does not quite meet the stipulated

minimum requirements. The graph does not give a complete picture of costs, as it includes only

some costs for some recommendations.

Belgium

Has various current activity which is specified in wording. However, it is indicated that an estimation

of costs is not feasible at the moment.

Denmark

The graph reflects all costs for all recommendations, assuming that the required resources for

fulfilling the recommendations correspond to the minimum requirements.

4.2 The top-down approach: expert estimates calibrated to each member state.

Results from the top-down approach are based on expert estimates, based on Danish experiences

of the minimum requirements for implementing the capability and competences for fulfilling the

FAIRMODE recommendations. The baseline was established for Denmark as the reference country

and the results for Denmark were calibrated to all the member states with respect to salary levels in

each member states.

It should be emphasized that the cost estimates do not include the cost of actually carrying out all

kind of work (e.g. reports including assessments and source allocation) in individual member states,

which can be attributed to the FAIRMODE recommendations, since the amount of assessment,

decision support or management reporting is not clearly defined in the recommendations.

Therefore, the cost assessment only includes an investigation of the cost associated with the

capability of fulfilling the recommendations.

We concluded in chapter 3, that the basic requirement for all member states to fulfil the FAIRMODE

recommendations, are the competences of four different staff members. Furthermore, we

concluded that having this basic group, the staff is able to fulfil the recommendations to a certain

extent, depending on the amount of assessment or reporting needed.

For countries already having well-established modelling groups, it is our experience in Denmark

that the four persons can easily be occupied by the work suggested in the FAIRMODE

recommendations and therefore the additional cost equals the cost of the baseline. For member

states, which do not already have a modelling group, but have to build it up from scratch, the same

requirements are needed. If the expertise is not available already in the member state, at least a

four year period is needed for starting up the modelling group, engaging and finishing at least three

PhD students covering the three scales. After this four year period, it is assumed that these

modelling groups are operational to fulfil the FAIRMODE recommendations to the same extent as

the well-established modelling groups.

A point that deserves to be kept in mind is raised in a comment from the UK provided in the

consultation survey carried out in March 2013 (see section 4.3): “Costs have been assumed to be

the same in all MS (in terms of staff time). This is unlikely to be the case. Costs are likely to be

higher in MS with more complex air quality situations or more exceedances. The costs will also be

crucially dependent on the organisational structure of the MS. In some MS air quality assessment

and management is carried out by regional authorities and in such cases the total cost for the MS

may be many times higher.”

42



The calibration could be further scaled e.g. with respect to population in each country, since a large

country with numerous big cities, probably needs to carry out more assessments or source

allocation studies, than smaller countries. However, since FAIRMODE is generally formulated,

there would be an arbitrary interpretation concerning the number of assessment and projects, and

therefore we have not presently calibrated with respect to population. Furthermore, it is also our

experience that smaller member states with well-established modelling communities, can perform a

much larger number of e.g. decision support reports with respect to air quality, compared to larger

member states, where the tradition in using model results for decision making is less pronounced.

In Figure 4.10 - Figure 4.22, the results from the top-down approach are provided. As already

explained the expert estimate for Denmark is adjusted according to the EU standard cost model

and then calibrated to all member states with respect to salary levels in the individual member

states. An overhead of 25% on salaries is assumed according to the EU standard cost model. The

cost is given for the total cost for baseline, which in this approach equals the total cost of the

options, for each baseline case (1-5) as well as for the individual FAIRMODE recommendations.

The cost for baseline 3, acquiring models, is set to zero, since it is assumed that models can be

acquired for free. Likewise, the cost for baseline 5, the total overhead cost, is zero, since it is

included in the salaries in baseline, 1 and 2.

The total estimated cost in Figure 4.10 - Figure 4.22 includes staff and staff capacity development

costs, equipment, operational cost for travelling and software, and 25% of overhead on the salaries.

In Figure 4.10, the total estimated cost for the baseline (sum of baseline 1-5) for each member state


the FAIRMODE recommendations is given. The differences in the total cost between the member

states are due to differences in levels of salaries according to the EU standard cost model. The

total cost for all countries for implementing the minimum requirements is 4.5 mio. Euros/year

In Figure 4.11 and Figure 4.12, the total estimated cost for baseline 1 (modelling) and baseline 2

(emissions) for each member state is shown. The total estimated cost of modelling includes the

cost of three modellers covering the different scales (long-range, urban background and urban

street) and staff capacity development costs, equipment, operational cost for travelling and

software, and 25% of overhead on the salaries. The total cost for all countries for implementing the

minimum requirements for modelling is 3.1 mio. Euros/year and for the emissions 1.0 mio.

Euros/year.

The total estimated cost for the baseline 4 (computing) for each member state is shown in Figure

4.13. The total estimated cost includes equipment, only and is equal for all countries. The total cost

for computing in all countries is 1.3 mio. Euros/year.

In Figure 4.14 - Figure 4.22, similar figures are displayed for the individual FAIRMODE

recommendations. The basis for the distribution of cost between the recommendations is given in

Table 3.3, assuming the minimum requirement approach.

Figure 4.14 - Figure 4.19 provides the results for recommendation 1, including the use of models for

assessment exceedances and population exposure, forecasting, souirce allocation, plans and

measures to control AQ exceedances, Designing monitoring networks, and using models to

determine the number of fixed monitoring. The total cost for all countries for implementing the

FAIRMODE recommendation 1 is estimated to 3.8 mio. Euros/year.

43



In Figure 4.20, the estimated cost for revision of the data quality objective for modelling

(recommendation 2) is shown. The estimated cost includes staff and staff capacity development

costs, and 25% of overhead on the salaries. The total cost for all countries for implementing the

FAIRMODE recommendation 2 is estimated to 0.15 mio. Euros/year. The assumption for this

recommendation is the staff time of 200 hours per year.

Figure 4.21, shows the total estimated cost for recommendation 3, competent authorities for

modelling activities are nominated by the Member states, for each member state calibrated from

expert estimates for Denmark with respect to the minimum requirements for fulfilling the

FAIRMODE recommendations. The total estimated cost includes staff and staff capacity

development costs, 25% of overhead on the salaries, and travelling. The total cost for all countries

for implementing the FAIRMODE recommendation 3 is estimated to 0.35 mio. Euros/year. The

assumption for this recommendation is the staff time of 300 hours and 3 travels per year – one for

each of the modelling staff members.

Finally Figure 4.22 shows the total estimated cost for recommendation 4; investigate and improve

the compilation, consistency and quality assurance of emissions data, for each member state. The

total estimated cost includes staff and staff capacity development costs, 25% of overhead on the

salaries, and travelling. The total cost for all countries for implementing the FAIRMODE

recommendation 4 is estimated to 0.2 mio. Euros/year. The assumption for this recommendation is

the staff time of 200 hours and 1 travel per year for the emission staff member.

Adding the cost for all the recommendations, we reach a total cost of 4.5 mio. Euros/year as was

the case for the baseline. The distribution of the baseline costs between the recommendations and

sub-recommendations has been made by an expert assessment. However, it can be very difficult to

distinguish the cost of the individual recommendations from each other, since they are all

interconnected, and supplement each other.

Figure 4.10: Top-down approach. The total estimated cost/year for the baseline (sum of baseline 1-5) for


requirements for fulfilling the FAIRMODE recommendations.

44



The total estimated cost includes staff and staff capacity development costs, equipment,

operational cost for travelling and software, and 25% of overhead on the salaries.

Figure 4.11: The total estimated cost/year for baseline 1 (modelling) for each member state calibrated

from expert estimates for Denmark with respect to the minimum requirements for fulfilling the




Figure 4.12: The total estimated cost/year for the baseline 2 (emissions) for each member state

calibrated from expert estimates for Denmark with respect to the minimum requirements for fulfilling the


45





Figure 4.13: The total estimated cost/year for the baseline 4 (computing) for each member state

calibrated from expert estimates for Denmark with respect to the minimum requirements for fulfilling the


The total estimated cost includes equipment, only.

Figure 4.14: The total estimated cost/year for recommendation 1a (assessment exceedances and

population exposure) for each member state calibrated from expert estimates for Denmark with respect

to the minimum requirements for fulfilling the FAIRMODE recommendations.

46





Figure 4.15: The total estimated cost/year for recommendation 1b (air pollution forecasting) for each

member state calibrated from expert estimates for Denmark with respect to the minimum requirements

for fulfilling the FAIRMODE recommendations.



Figure 4.16: The total estimated cost/year for recommendation 1c (source allocation) for each member

state calibrated from expert estimates for Denmark with respect to the minimum requirements for

fulfilling the FAIRMODE recommendations.

47





Figure 4.17: The total estimated cost/year for recommendation 1d (plans and measures to control AQ

exceedances) for each member state calibrated from expert estimates for Denmark with respect to the

minimum requirements for fulfilling the FAIRMODE recommendations.



48



Figure 4.18: The total estimated cost/year for recommendation 1e (designing monitoring networks) for





Figure 4.19: The total estimated cost/year for recommendation 1f (number of fixed monitoring stations)

for each member state calibrated from expert estimates for Denmark with respect to the minimum




49



Figure 4.20: The total estimated cost/year for recommendation 2 (revision of the data quality objective

for modelling) for each member state calibrated from expert estimates for Denmark with respect to the

minimum requirements for fulfilling the FAIRMODE recommendations.

The total estimated cost includes staff and staff capacity development costs, and 25% of overhead

on the salaries.

Figure 4.21: The total estimated cost/year for recommendation 3 (competent authorities for modeling

activities are nominated by the Member states) for each member state calibrated from expert estimates

for Denmark with respect to the minimum requirements for fulfilling the FAIRMODE recommendations.

50



The total estimated cost includes staff and staff capacity development costs, 25% of overhead on

the salaries, and travelling.

Figure 4.22: The total estimated cost/year for recommendation 4 (investigate and improve the

compilation, consistency and quality assurance of emissions data) for each member state calibrated

from expert estimates for Denmark with respect to the minimum requirements for fulfilling the


The total estimated cost includes staff and staff capacity development costs, 25% of overhead on

the salaries, and travelling.

4.3 FAIRMODE Consultation Survey of the top-down assessment

In appendix E, the questionnaire and results from the consultation survey on the top-down

approach, conducted in March 2013 are provided. The idea of the consultation survey was to verify

the assumptions of the minimum requirements for building up the capacity and competences

needed for being able to fulfill the FAIRMODE recommendations. For all questions, the member

states were asked to assess the estimated level of cost on a relative scale and to make comment.

The countries, which replied to the consultation survey, were: Belgium, Croatia, The Czech

Republic, Ireland, Latvia and the UK. The following includes a summary of the answers and

comments from the survey. The full answers can be seen in appendix E.

Question 1: Do you think that the estimate for minimum requirements for the personal resources

and competences needed for fulfilling the FAIRMODE recommendations in your country is

adequate?

The answers range from 3 times too low to 25% too high.

51



Belgium answered 100 % too low. They estimated that a staff size of 22.5 + supplementary staff will

be needed to comply with the recommendations.

Croatia answered 50% too low due to high resolution emissions and the development and

application of models for urban scale still need to be built up. Croatia furthermore commented that

the proposed number of scientists is acceptable if such modelling system is already at operative

level. However in countries where this is a task to go at least two more scientists are needed.

The Czech Republic answered 25% too low, partly due to the assumption that meteorological

modelling has not been included, partly because forecast operation also requires separate time.

Ireland estimated the minimum requirements to be 25% too high. Latvia estimated 100% too low,

but states that 3 staff member is necessary.

The UK states that the estimate is too low by at least a factor of 3. The UK is currently using air

quality models to support compliance assessment, source apportionment, baseline projections,

development if air quality plans and review of air monitoring networks, so they have a good

understanding of the costs.

Question 2: Do you think that the estimate given for minimum PC requirements for the direct

operational costs (excluding overhead) needed for fulfilling the FAIRMODE recommendations in

your country is adequate?

The answers range from 100% too low to acceptable, depending of the number of staff

members needed. All agree, not surprisingly, that 1 PC per person is adequate.

Question 3: Do you think that the estimate given for minimum server requirements for the direct



The answers range from a factor 4 too low (Belgium) to acceptable (all other countries).

Question 4: Do you think that the estimate given for minimum license requirements for the direct



The answers range from far too low, by a factor of ten (UK), to acceptable.

The answers indicated quite big differences between the countries, which replied in the consultation

survey, concerning the total cost for licenses, especially depending on what kind of GIS software is

used (from freeware to costly software). Furthermore, it is mentioned that models are not always

free as well as meteorological data.

Question 5: Do you think that the estimate given for minimum travel requirements for the direct



The answers range from acceptable to 50% too high.

52



Question 6: Do you think that the estimate given for minimum training requirements for the direct



The answers range from 100% too low (Ireland) to acceptable. Ireland replied that the training costs

associated with models are high. They also tend to be in one location in Europe (eg calpuff, aermod

training), so traveling costs and subsistence have to be considered. Others consider the cost of

training one person for a 4-days training course. In the assumption in the top-down approach, the

number we have applied is an average over the years, meaning that not all staff members will have

training courses every year.

Additional comments from the member states:

The Czech Republic: The competence building phase will be important because I perceive there

to be a current lack of availability of such staff with the required AQ modelling-specific experience.

In other words, the right people might (do) currently exist but not all in the one institute and not

having a role dedicated to AQ modelling for European regulatory purposes.

UK: Costs have been assumed to be the same in all MS (in terms of staff time). This is unlikely to

be the case. Costs are likely to be higher in MS with more complex air quality situations or more

exceedances. The costs will also be crucially dependent on the organisational structure of the MS.

In some MS air quality assessment and management is carried out by regional authorities and in

such cases the total cost for the MS may be many times higher.

Conclusions from the consultation survey

All in all, the results from the consultation survey clearly display the very different interpretation in

each member state of the minimum requirements for fulfilling the FAIRMODE recommendations.

There is a clear tendency for member states which already operate large modelling capacities to

assess the necessary cost as corresponding to the present size of the modelling groups. On the

other hand, the member states actually assessed the present demand and needs for modelling

capacities in the individual countries, including the present level of assessments, reporting and

decision support, which are all activities that are related to the FIARMODE recommendations.

As an overall conclusion, the member states which indicate much larger costs already operate large

modelling groups that are used for the tasks described in the FAIRMODE recommendations.

Therefore, the larger costs are not additional costs. The countries that do not already have

operational modelling capacities, indicated that the minimum requirement of 4 staff member and the

related equipment and travelling was acceptable for starting up modelling capacities and

competences in their respective countries.

4.4 Combining the bottom-up with the top-down approach: estimating the cost for

fulfilling the minimum requirements

In section 4.3, the total estimated cost for fulfilling the minimum requirements for each member

state was given. The total cost for all member states were estimated to 4.5 mio. Euros.

Furthermore, we have on basis of the bottom-up approach and our own expert assessment

estimated the present level of fulfillment of the minimum requirements for all the member states –

e.g. Denmark already fulfills the minimum requirement for competences available, so the additional

cost in Denmark is zero.

53



From this we can calculate the additional cost for each member state for fulfilling the FAIRMODE

recommendations by:

Additional cost in country = 1 - (present level of fulfillment [0:1]) x total cost of the baseline (1-5) for

each country,

The result is the graph below. The total additional total cost in is estimated to 1.4 mio. Euros. The

relatively low number compared to the total baseline cost for all member states of 4.5 mio Euros, is

due to the fact that within many of the relatively high salary countries in the north western part of

Europe (e.g. Denmark, Belgium, The Netherlands, Germany, Finland, Sweden, UK), larger

modelling groups are already established, and for these countries there is no additional cost. For

many of the countries, where full modelling capabilities are not present, the salaries are relatively

low, and therefore the total additional cost for implementing the FAIRMODE recommendations is

also relatively low (1.4 mio. Euros, assuming the EU standard cost model)..

Figure 4.23: Estimated total additional cost/year for each member state for fulfilling the minimum

requirements, using the EU standard cost model. The total additional cost for all member states all

together is 1.4 mio. Euros/year.

54



5 Summary and conclusions

The work described in the present report had as its aim to produce a cost assessment of the

FAIRMODE recommendations. It has been a fundamental challenge throughout the work that the

FAIRMODE recommendation are not presently in a form where they can be included in a directive.

They are “soft” recommendations, which are open to a wide range of interpretations. Therefore, this

entire work represents an attempt to quantify costs despite this basic difficulty. The cost estimates

have only been possible by defining sets of assumptions concerning implementation of the

recommendations, as specified in the text of the report. This should be kept in mind in any use of

the cost estimates.

The cost assessment of the FAIRMODE recommendations has been carried out by using two

methodologies: One takes a bottom-up approach, and the other a top-down approach. It should be

emphasised that the baseline for the bottom-up approach did not equal the baseline for the top-

down approach. The first baseline applied to the information request in the bottom-up approach, a

typical research environment was defined with 3-5 staff members both in the field of atmospheric

modelling as well as within the development of emission databases. This lead to a total staff of 6-10

people, which is typical for an active and living research group. Furthermore, in the information

request the replies from the member states typically reflected their current situation. The second

baseline was used in the top-down approach and consisted of minimum requirements of 4 staff

members. The baseline was reduced in the top-down approach, due to considerations of the very

general nature of the FAIRMODE recommendations, where it is difficult to argue whether a member

state should have a certain number of staff member to fulfil the recommendations, since the amount

of e.g. assessments and reporting as well as scales, chemical species and quality of model results

was not defined in FAIRMODE. We argued that due to the general nature of the recommendations,

we could only assess the cost of the minimum requirements for building up modelling capacity and

competences in each member state. Therefore the total cost assessments of the bottom-up

approach and the top-down approach cannot be directly compared.

The bottom-up approach is based on results from an information request to the member states

carried out in December 2012- January 2013, combined with in depth case studies carried out for

Belgium, Croatia and Hungary. 17 countries replied to the information request. However, for each

question a much smaller number of member states provided quantitative data, which could be used

for the cost assessment. On the other hand, the information request has provided much valuable

qualitative information.

The top-down approach was based on expert estimates for Denmark calibrated/scaled to all other

member states + Croatia based on the EU standard cost model. As base for the cost assessment a

set of minimum requirements were defined, concerning establishment of modelling and emission

capacities and competences as well as computing facilities in the member states. The overall cost

of the minimum requirements with respect to staff time and computing, etc. was assumed to be the

same in all member states, no matter whether the member states already had well-established

modelling communities or not. The fulfilment of the FAIRMODE recommendations is a question of

timing, since the member states with well-established modelling communities already fulfil the

FAIRMODE recommendations, while the member states, which do not possess modelling

capacities, can build up a modelling group of limited size over a period of four years, by educating

PhDs within long-range transport modelling, urban background modelling, urban street modelling as

well as building up competences with emissions.

55



A main outcome of the top-down approach is the costs for each country to fulfil a minimum set of

modelling competences, irrespective of whether the state already possesses these competences.

With this approach, the variation between countries is entirely due to varying tariffs for wages.

Furthermore, results from the top-down approach can be combined with results derived from the

bottom-up approach. The bottom-up approach provides information for an expert estimate as to

which degree each member state already fulfils the FAIRMODE recommendations. When such an

estimate is combined with cost estimates from the top-down approach, it results in estimates of

additional cost (compared to now) for implementing the recommendations in the EU.

From the top-down approach, it was found that the estimated total cost of implementing the

FAIRMODE recommendation from scratch in the whole EU is an annual cost of 4.5 mio. Euros.

However, since several member states already to some extent fulfil the set of recommendations,

we further found that the additional cost compared to the present situation is 1.4 mio. Euros. Very

importantly, these estimates rely on the specified setup of minimum requirements, and they

represent the cost only to achieve the minimum requirements. The bottom-up results from the

information request show that some countries have a much a larger modelling activity than

prescribed by the minimum requirements, and consequently they spend a much larger sum on their

total modelling activity.

It should also be noted that the figures above assume the EU standard cost model. The salary level

and the overhead rate in this cost model are relatively low.

The basic challenge of assessing the FAIRMODE recommendations with respect to cost, is that

they are very general in nature and softly formulated. They are open to many different

interpretations, as it is apparent from the responses from the different member states. Some

indications of costs in various countries can be obtained from the questionnaires, but the numbers

cannot always be compared directly.

The cost assessment in this report provides an estimate of present cost in some member states

from the bottom-up approach as well as the total and additional cost for each member state based

on the top-down approach. A more valid cost assessment can only be made by translating the

softly formulated FAIRMODE recommendations into specific requirements, specifying amount of

reporting, modelling scales, model types, species to be modelled, required accuracy of results, etc.

A main challenge in implementing the FAIRMODE recommendations is the establishment of high-

resolution and high-quality emission data. All countries already report emissions with a 50 km x 50

km grid resolution every 5 years and national emissions every year, which can be further detailed.

However, there are large differences between countries as to which level of detail is available in

current registrations of sources/emission sectors.

A difficulty in assessing cost is due to the fact that the administrative organization within a country

can take various forms – one model is to have a central unit with expertise, another is to have

regional units, where each unit is assumed to have access to expertise. The cost in a specific

country will depend on the way this is organized.

The present study is focussing on costs only, but benefits should not be left out of sight. It is very

difficult to carry out an assessment of the benefits resulting from taking the right decisions regarding

air quality in Europe. From measurements in monitoring programme, the magnitude of the problem

can be assessed. However, the strengths of using models are their ability to provide understanding

of measurements, and to provide the best foundation for decision making. The potential for

56



improving health and welfare in Europe is considerable and the potential cost, found in this work,

for making the use of models mandatory is relatively very small.

57



References

EEA 2011. The application of models under the European Union’s Aior Quality Directive: A

technical reference guide. EEA Technical report No. 10/2011. 72 pp.

Hasler, B., L. Martinsen, M. Hussen Alemu, J. Brandt, H. R. Olesen, O. Hertel, A. Massling, C.

Nordström, E. Hayes, J. Barnes and T. Chatterton, Review of: Provision for Air Quality

Measurement, Air Quality Modelling, Management Framework, Assessment, and Public

Information; and Stakeholder Consultation Support – Cost Assessment Methodology of the

AQUILA, FAIRMODE and SEG Recommendations from, pp. 22. Client: DG Environment, March,

2013

58



Appendix A: The FAIRMODE Recommendations

Ispra 12/10/2012

1. ON THE USE OF MODELS FOR REGULATORY PURPOSE AND TO SUPPORT AIR QUALITY POLICY Issue:

Despite models being a widely used and essential tool for air quality assessment, very few member

states report modelled data as part of the compliance assessment under the Air Quality Directive

(AQD). There are many advantages to using and reporting modelled data which are not being

realised. The AQD text on the application of models is not sufficiently clear and requires revision.

Background Information

Models are widely used for the investigation and assessment of ambient air quality at various

spatial and temporal scales. In situ monitoring data is by its nature only a small sample of the

spatial distribution of air pollutant concentrations. Models can be used to provide an assessment in

areas where monitoring data are not available and thus, in combination with monitoring data,

provide a more complete assessment of the current air quality situation. Models are also essential

for the development and assessment of the effectiveness of air quality plans including measures to

improve air quality. One of the major advantages of models is their potential to provide detailed

spatial distributions of air pollutant concentrations. This means they can be applied for area-wide

exposure assessments for human health and for the environment. An application of models allows

a much broader assessment of the extent of exceedences air quality environmental objectives and

can also provide information required for improved measurement network design.

Currently the text of the AQD indicates that models may be used as ‘supplementary data‘ in assessment and

that models may be used to assess the level of exposure, but their role further to this is poorly defined. Text

such as ‘The results of modelling and/or indicative measurement shall be taken into account for the assessment

of air quality with respect to the limit values’ does not clarify what role the models will play in the assessment.

Indeed, models are not named at all as the major tool for developing and assessing plans and measures to

mitigate air pollution, nor are they referred to in regard to short term forecasting. Their use in source

apportionment is also not indicated within the AQD.

Recommendation

FAIRMODE strongly recommends the use of models for the following applications. The AQD text

relating to these applications should be clarified:

Assessment of air quality levels to establish the extent of exceedances and establish population

exposure

Forecasting air quality levels for short term mitigation and public information and warnings

Source allocation to determine the origin of exceedances and to provide a knowledge basis for

planning strategies

Development and assessment of plans and measures to control AQ exceedances

59



In addition to these applications the use of models is strongly recommended for:



Proposed action: In all articles of the Air Quality Directives (AQD) where the four above-mentioned

applications are mentioned the use of models should be strongly recommended. FAIRMODE will

provide a list of respective AQD and Commission Implementation Decision (CID) articles.

2. MODEL QUALITY OBJECTIVES

Motivation: Data quality objective for modelling (MQO) are mentioned in the 2008 Air Quality

Directive, Annex I, but the wording of the text remains ambiguous and open to interpretation. The

FAIRMODE report ‘The application of models under the European Union’s Air Quality Directive – A

Technical Reference Guide’ reviewed different interpretations and recommended the use of the

relative directive error (RDE) indicator to provide a quantitative estimate of the model uncertainty.

Despite these recommendations, the current AQD MQO retains some limitations which are inherent

to their formulation. In addition the quality objectives are only valid for assessment applications, not

for planning.

Recommendation: FAIRMODE recommends a revision of the data quality objective for modelling

Proposed action:

Fairmode is developing new data quality objectives for modelling for ambient air quality

assessment, in collaboration with the Member States . This type of objectives are expected to

be useful as basis to investigate MQO for the other model applications

We propose that subsequent the work of Fairmode the European Commission initiates a

process for the preparation of a Guidance document on the revision of model quality objectives

for assessment

3. FORUM OF EU AQ REGULATORY MODELLING

Motivation: Air quality modeling in support to air quality policies in the EU context requires a

constant level of communication and competence building among the various competent

authorities. The organization of periodic model evaluation activities are required in order to assure

harmonized practices and to guarantee comparable quality levels across Member States.




Proposed action: Fairmode will act as coordination forum for modelling and support the competent

authorities in activities that are recognised as very relevant for the for model applications listed in

Recommendation #1 and that are presented here with an extended list of supporting motivations.

Namely:

Model evaluation: model evaluation has to be a continuing activity. In the case of a community

of models and model users joint model evaluation activities have demonstrated to be very

important in speeding up the harmonization of the practices, identification and fixing of

problems, rapid improvement, transition to operational activities, competence building and

60



sharing. Fairmode can offer the right framework for testing the newly developed MQO and

promoting joint model evaluation activities on common case studies.

Coordinated action: model evaluation activities, competence building and sharing.

Combine use of model and monitoring data: Methodologies and techniques have been

developed and tested to combine model results and monitoring data in order to provide

improved assessment and predictions skills. that take in to account the experimental evidence

and complement it with model results. Those developments have shown very promising results

at various spatial and temporal scales. For the AQD application addressing assessments

(Application 1) the most powerful tool for providing complete spatial coverage of the air quality

situation, whilst still retaining the quality of fixed monitoring data, is the combined use of models

and monitoring. The preamble text of the AQD refers to this: ‘Information’ from fixed

measurements may be supplemented by modelling techniques and/or indicative measurements

to enable point data to be interpreted in terms of geographical distribution of concentrations’.

Coordinated action: inventory of ongoing activities, inter-comparison of methods, competence

building, preparation of guides.

Source apportionment modeling: There is an increasing need to demonstrate whether and to

what extent exceedances of limit values can be attributed to natural sources, human practices

(road salting and sanding), and transboundary pollution. In the context of the preparation and

implementation of air quality plans and short-term action plans, there is also a need to identify

and quantify the contribution of the main pollution sources in order to efficiently design

abatement measures and assess their effectiveness.

Coordinated action: inventory of ongoing activities, inter-comparison and evaluation of methods,

competence building, preparation of guides.

Monitoring station characterisation and meta data description for model applications and

support to optimisation of monitoring and network design: The location, characterization and

representativeness of a measurement station in a monitoring network is of fundamental

relevance for the evaluation of model results (see point 3.) and when using network output as

model input data. It is well recognized that current station classification and characterization is

not harmonized across the EU with some consequences for data interpretation and use.

Coordinated action: inventory of ongoing activities, competence building, preparation of guides.

4. QUALITY ASSURANCE AND CONSISTENCY OF EMISSION INVENTORIES

Motivation: Air quality emissions inventories (EIs) have been compiled at European level for

regulatory purposes and also to support air quality modelling applications for the assessment and

improvement of air quality. Emission information is essential in support for AQ planning under the

AQ directive because it provides the link between responsible emission sources, their relative

shares and abatement potentials. Current emission inventories are constructed at different scales

(regional, national, city/urban) but these EIs are often not consistent and may cause discrepancies

in impact assessments at the different scales. The present compilation methods do not always

allow relating emission sources with their abatement potential. More detailed emission inventory

compilation methods and better systems for QA/QC of emission information need to be

implemented to support AQ planning and account for the identified discrepancies in the different

scales.

61




consistency and quality assurance of emissions data suitable for AQ modeling under the directive

Proposed action:

Emissions are not mentioned in the AQD and the need to work to increase the quality of

emission inputs needs to be introduced in the revised text.

Promote guidance initiatives for the compilations of emission data for AQ models under the

directive

Support competence building initiatives to secure the consistency of detailed bottom-up

emission inventories with those compiled for regulatory purposes at local, national and

European scale

62



Appendix B: Case studies for three countries

This appendix presents information from three countries, where in addition to the information

request phone interviews were carried out with the person(s) who had filled in the questionnaire.

The interviews were based on the questionnaire. They were carried out in order to substantiate the

information given in the information request and to learn in greater detail what motivated the

responses.

For each country there is a short section summarising the situation within the country and findings

from the interview, followed by sections with details, combining the information from the information

request and the interview.

The countries chosen for the case study were

Belgium

Croatia

Hungary.

These were among the countries which had provided the most detailed quantitative information to

the information request, and thus it was very relevant to learn about the background for the

responses and possible pitfalls in interpretation of the indicated values.

Belgium

Summary: Situation within the country and findings of interview

There are three regions in Belgium, so there are several regional environmental agencies in

addition to the interregional agency IRCEL (respondent). The modelling capacity varies between

the regions, and is best developed in Flanders, where much of the expertise lies within one

contractor (VITO). The Baseline costs of over 3 milllion Euro reflect the activities within the

agencies and the contract sum for Flanders. Some regions indicate that additional efforts are

required in order to reach the minimum requirements, but have not made specifications of this.

Concerning the recommendations, some are partly fulfilled at present. Thus, many air quality

assessments regarding exceedances are available, and forecasting (recommendation 1b) is

applied. The response in the questionnaire indicates that these recommendations are cost-neutral.

The reasoning is that based on current work it is possible to report model based assessments,

although it is not done. For many other recommendations an often repeated comment is: “An

estimation of the costs to conform to this standard is not feasible at the moment.”

Several matters lie behind this statement: the requirements are not clearly specified; a baseline

capacity is not established in all regions; an estimate requires estimates from several regions who

have not delivered them.


This competence building is available at the different regional environment administrations and the

Belgian interregional Environment Agency. Subcontractors (like e.g. the Flemish Institute for

Technological Research) are also playing an important role in air quality modelling in Belgium.

Some of the Regions indicate that competence in atmospheric modelling and air quality modelling

63



will be further developed, but additional efforts have to be made to achieve the specified minimum

requirements. The following calculation is an estimation of the existing annual costs of baseline 1.

Some of the regions indicated a change in costs to conform to the baseline 1 standard but no

further details of the costs could be provided at the moment.

Supplementary information from interview:

The interregional agency IRCEL coordinates and performs national tasks. The costs reflect that

there are 3 regions: Wallonia, Brussels and Flanders, each with a regional environmental agency.

The Flemish Institute VITO is contractor (around 2 persons financed by subcontract). There is an

estimate of 6 additional persons working with modelling: 2 within IRCEL and the remaining within

the administrative environmental agencies in Belgium. Modelling capacity best developed in

Flanders.


This competence building is available at the different regional environment administrations and the


Technological Research) are also playing an important role in air quality modelling in Belgium.

Some of the Regions indicate that competence in geographic resolution emission databases is

developed, but will be further developed, but additional efforts have to be made to achieve the

specified minimum requirements. The following calculation is an estimation of the existing annual

costs of baseline 2. Some of the regions indicated a change in costs to conform to the baseline 2

standard but no further details of the costs could be provided at the moment.


There is a well-equipped team in Flanders. Costs are based on 10 people in Flanders, 2 in Wallonia

region, 1 in Brussels, ¼ in IRCEL. These emission inventories are not made for modelling, but for

reporting. Achieving the baseline is not cost-neutral (checkmark in questionnaire is set incorrectly),

as some of the regions indicated a change in costs.

BASELINE 3: DEVELOPMENT OR ACQUISITION OF ATMOSPHERIC MODELS

Belgium does not have a junior staff for technical implementation and maintenance of models and

computer facilities to make the models FAIRMODE compliant. Some of the regions developed

appropriate atmospheric models operating with meteorological data (from the national

meteorological institute) with a spatial resolution of 1x1 km and a temporal resolution of 3 hours.


One additional person presumably needed at IRCEL to run models.

BASELINE 4: COMPUTER FACILITIES FOR RUNNING HIGH RESOLUTION MODELS

Yes, considered that these facilities are not only present within the Belgian interregional

Environment Agency but also in the Regions and subcontractors.


The sum for equipment is based on 13 PC’s (1000 Euros each) and 15 servers (7000 Euros each).

BASELINE 5: OPERATIONAL / RUNNING COSTS

Yes, we can provide the indicative cost made by the Belgian interregional Environment Agency and

not the operational/running cost for the regions or consultants. We cannot provide the different cost

categories but only one figure (1.235.000 euro).

64




Operational costs are estimated as 30% of the IRCEL budget. IRCEL has modelling as one of its

responsibilities. Actually this method gives some double counting, as 2 persons at IRCEL were also

counted under Baseline 1. Thus, 200000 Euro may be subtracted from the figure indicated.

OPTION 1: ON THE USE OF MODELS FOR REGULATORY PURPOSE AND TO SUPPORT AIR

QUALITY POLICY

Recommendation 1(a): Use AQ models for assessment of air quality lvels to establish the extent of

exceedances and establish population exposure

In Belgium deterministic models and/or intelligent interpolation techniques are used to calculate

population exposure and/or to evaluate measures. For the annual questionnaire concerning the

reporting of the air quality in the framework of the directive 2008/50/EC mainly measurements are

used. Some of the regions have developed models and algorithms in order to meet these

recommendations: modelling of pollutant concentrations and calculation of population exposure

with a spatial resolution of 1 km x 1km, modelling of emissions scenarios to manage source

legislation or to optimise air quality monitoring etc.


A lot of air quality assessments regarding exceedances are available at present, so the

recommendation can be said to be fulfilled. However, for reporting only measurements are used.

Belgium intends to use modelling for supplementary information.

The response in the questionnaire indicates that the option is cost-neutral. The reasoning is that

based on current work it is possible to report modelbased assesments, although it is not done.



In BE a number of air quality forecast models are used to inform the public and to trigger short term

emission reduction measures (e.g. speed limits on high ways during smog episodes).


Indicated as cost-neutral because some forecasting is practised. Costs of model maintenance is

contained in the Baseline costs.



Source allocation is mainly performed via consultancy assignments e.g. in the framework of time

extension demands, plans and programmes when limit values are exceeded. In some Regions

local monitoring campaigns (from 6 weeks to 6 months) are aimed to determine exposure and

origin of emissions for regulation strategies purposes. These programs are supported by

mineralogic/ chemical analysis of PM (scanning electron microscopy coupled to a spectrometer

energy dispersive analysis (EDX), and the overall analysis using the X-ray diffraction (XRD)).

Additional investments will have to be done. However an estimation of the costs to conform to these

standards is not feasible at the moment.

Recommendation 1(d): Use AQ models for development and assessment of plans and measures to

control AQ exceedances

In BE deterministic models are used to evaluate emission reduction measures and to develop plans

and programmes. Not every Region uses deterministic models to evaluate measures but

65



developments are ongoing. Additional investments will have to be done. However an estimation of

the costs to conform to this standard is not feasible at the moment.


It is mainly the Flemish region, which has the capability to develop plans and measures. Therefore

the option is not considered cost-neutral.

Recommendation 1(e): Use AQ models for designing monitoring networks when models are used in

combination with monitoring

"The use of AQ models in combination with monitoring to design monitoring networks is not a

common practice in BE yet. The use of these techniques will generate a supplementary cost for BE,

however an estimation of these costs is not feasible at the moment.


IRCEL is involved in a related activity, the LIFE project ATMOSYS, which aims at classification of

monitoring stations.

Recommendation 1(f): Use AQ models for determining the number of fixed monitoring sites that are

required

The use of AQ models to determine the number of fixed monitoring sites is not yet a common

practice in BE. The use of modelling for this purpose will generate a supplementary cost for BE,

however an estimation of these costs is not feasible at the moment.

OPTION 2: MODEL QUALITY OBJECTIVES (MQO)

"BE currently uses data quality objectives (validation statistics) and will continue to actively follow

up the activities within the FAIRMODE working group concerning the model quality objectives. In

some regions model uncertainty is calculated for interpolation and/or dispersion models and model

validation is done with supplementary measurements from mobile measurement campaigns.


The option has been indicated as cost-neutral because there is an ongoing Fairmode activity, which

is covered by the baseline costs.

OPTION 3: FORUM OF EU AQ REGULATORY MODELLING

Yes, BE has nominated a competent authority for the FAIRMODE activities.


The option has been indicated as cost-neutral because there is an ongoing Fairmode activity, which

is covered by the baseline costs.

OPTION 4: QUALITY ASSURANCE AND CONSISTENCY OF EMISSION INVENTORIES

In BE work concerning the improvement of the emission inventories is ongoing. If supplementary

quality objectives have to be met in the framework of FAIRMODE this will generate an investment,

however an estimation of these costs is not feasible.

Croatia


The existing core modelling group consists of only 2 persons, although there is more modelling

expertise in various institutions. However, a team of at least 5 scientists working in the field of

66



atmospheric modelling should be developed in one institute that will be able to work together and

develop competence in atmospheric modelling to provide data for reporting on national level,

planning and other MS obligations toward EU.

The resolution of the models currently applied in Croatia is 10 x 10 km. Emissions are not at that

level of detail, and local scale modelling leaves much to be desired. The respondent assumes that

high resolution databases will be entail much work, because data will have to be collected more or

less from scratch (whereas in some other countries there is much existing statistical material to

build on).

The figures for computer capacity in Croatia are large compared to the stipulated minimum

requirements. The investment of 370000 Euro for increased computer capacity mentioned for the

baseline 4 question should be distributed over 3-5 years. The very large number appears because

a multicore machine with certain parallel functionality is required in order to run existing code. The

number corresponds to the cost of existing (SGI) computers.


Our member state conforms to the specified standards when whole country is taken into the

consideration and expert employed in different institutions are accounted. However, a team of at

least 5 scientists working in the field of atmospheric modelling should be developed in one institute

that will be able to work together and develop competence in atmospheric modelling to provide data

for reporting on national level, planning and other MS obligations toward EU.


Establishing a baseline: It is confusing that the leading text says “Additional staff cost”. For this

reason costs of current staff – which is two persons was omitted. Add 100000 to 112000 (50000 per

person).

Cost category 5 and 6: The respondent wished to indicate that 5 FTE staff members were required,

and they would have to divide their time between work and competence building. Thus, there

should not be 1100 days + 1100 days. Instead, divide them into 770 days + 330 days.


Emission data on 50x50 km horizontal resolution are being produced regularly. Emission inventory

on 10x10 km horizontal resolution is developed. Staff development, database software, GIS

software, new emission data generation is foreseen.


The emission inventory is currently outsourced. The numbers under “Establishing.” is meant to

reflect this. It is necessary with a staff of 10 (but presumably outsourcing could be replaced by in-

house activity in that case) to build up things.

The respondent assumes that high resolution databases will be entail much work, because data will

have to be collected more or less from scratch (whereas in some other countries there is much

existing statistical material to build on).

The 5 + 3 persons in questions 5 and 6 should only appear only once in these.

67




Air quality models for regional and urban scale are/can being used for assessment and reporting.

Our meteorological service provides appropriate meteorological data. Operational annual reporting

including model validation based on modelling is foreseen.


Currently we are using SGI computer with 54 core for all our model runs at Meteorological Service.

However their capacity is limited and the computer is used by different groups (weather forecasts,

climate etc.). Therefore in other to be able to produce modelled air quality data on operative basis

an upgrade or the purchase of new computers is needed.


740000 for computers corresponds to the price of existing computers.

The investment 370000 for increased computer capacity should be distributed over 3-5 years. The

very large number appears because a multicore machine with certain parallel functionality is

required in order to run existing code.


Administration costs are satisfying. Accommodation requests are not fulfilled while traveling and

mobility of scientist currently working in atmospheric modelling is covered additional resources for

new staff that needs to be employed to conform with the minimum number (5 scientists) need to be

assured. Operational costs for maintaining readiness for the staff's ability for decision making

support is not fulfilled. Costs for the operational forecast system are not covered.


QUALITY POLICY



Models are used for the air quality assessment to provide information on air quality levels in zones

that not covered with fixed measurements. The application of models in agglomerations is

envisaged and further development of emission inventories is needed. Models will be used for

annual air quality reporting regarding the assessment of exceedances and exposure as a

supplement to measurements.


A 50 core processor is required for two months per year.

Category 4 for reporting: It is indicated that 1 person is required, but only 30 days. This should be

corrected, so that existing staff delivers 30 days, but additional staff has to be employed and use

130 days on reporting.



No, we are not providing air quality forecasts.


In order to produce forecasts not only 10 days would be needed, but also resources from a new

person. Thus Cost category 2 should be 110 + 110 days, in total 220.

68



Recommendation 1(c): Use AQ models for souce allocation to determine the origin of exceedances

and to provide a knowledge basis for planning strategies

Yes. The source allocation techniques (trajectories, models) have been used. However permanent

scientific development and application of new methods is needed. The increase in reporting

obligations is foreseen.

Recommendation 1(d): Use AQ models for development and assessment of plans and measures to

control AQ exceedances

Yes, we have used models for development of national plans eg. for ozone. Models need to be

constantly improved and further application of models if foreseen.



Yes, we have used model results for assessment and design of compliance network. Further

applications are foreseen.


required

Yes, we have used models for development of national plans e.g. for ozone. Models need to be



We are actively involved in the work of Fairmode regarding MQO and the revision of existing

objective in the Directive is needed.


Costs are too difficult to quantify as long as it the necessary activities are undefined.


We do not have nominated competent authority for modelling and this Fairmode activity is fully

supported.


Too difficult to quantify as long as it the necessary activities are undefined.


(Questionnaire not filled in)


The recommendation will possibly not induce any additional costs, once a baseline capacity

concerning emission inventories is established.

Any further quantification is difficult.

Hungary


Hungary uses very few resources on modelling, and in recent years there has been a reduction in

staff. Nevertheless work on modelling is ongoing, which fulfils part of the stipulated baseline

69



activities and part of the recommendations. The Chimere model is used for forecasting. Such a

system has been set up so that it runs automatically.

The respondent would like to make use recommendations of FAIRMODE and participate in

Fairmode work. But costs such as registration fees at conferences in practice act as a barrier to

international cooperation. It would be appreciated if Fairmode would arrange training courses where

there is no registration fee, like it is the case for EMEP.


Baseline requirements are not yet fulfilled, there is only 1 full-time senior staff within atmospheric

modelling, working with only one 10-year old PC. Because of serious restrictions in the budget,

there is no possibility to employ any more staff persons and to purchase new computers at the

moment.


The numbers concerning current activities are essentially the respondent’s salary. There used to be

more staff involved in modelling, but it has been cut down. The respondent hopes for improvement

in the situation, but the current economic situation has forced the level of activity to be very low.

Forecasts are produced for Budapest with 2 km resolution. Work is ongoing to extend it to the entire

country. Available emission data are not sufficient for good results.

The forecast system is set up so that it runs automatically (Chimere).


Currently, there is a staff of 3 for both GHG and CLRTP reporting in the Hungarian Met. Service.

However, transport emission modeling has been outsourced, and the compilation of the forestry

inventory is done by the relevant governmental institute. We have had no education in databases

and GIS, and generally self-education is the dominant form of capacity building. We're at the very

beginning of the process of building national databases.


The sum indicated is for outsourced activities and in-house activities.


Existing models under maintenance (which were downloaded freely) are: AERMOD, CHIMERE.

High resolution meteorological data are available (AQ modelling is done by the national met. office).


Also the Flextra trajectory model is used.


There is only one 5-year old 2-core computer maintained (housing, power and cooling is OK).


Chimere is run on Linux PC. Supercomputer available at met office. It is desirable with more

computer capacity.


No administration, no economic officer, office space for 2 personnel, serious restrictions in

travelling, there are limited opportunities to attend conferences. An operational AQ forecast system

runs for Budapest in additional to the met. office forecasting tasks and costs.

70




It would be desirable with economic officer.


QUALITY POLICY



We do not use models for the investigation and assessment of ambient air quality.


The key question is to ensure the baseline capacity. The activity recommended by FAIRMODE has

to be put on top of an adequate baseline activity, but as long as the baseline activity isn’t there it is

hardly possible to specify the required additional resources.



We use CHIMERE chemical transport model to forecast air quality levels for Budapest. In the future

we plan to use air quality model to evaluate the air quality levels in Hungary.


The costs for current activities for Rec. 1b also constitute part of the activity indicated as Baseline 1

(part of the respondent’s time is used for forecasting).

Concerning scales, the forecasting is on a 2 x 2 km scale, not hotspot modelling.



We use AERMOD model for regulatory purposes to assess the future effect of a new investment.


The AERMOD activity possibly more properly could be classified as related to Planning (1d). In an

EMEP context there is some work on source allocation. The Flextra model is used.

Recommendation 1(d): UseAQ models for development and assessment of plans and measures to

control AQ exceedences

We use CHIMERE chemical transport modell to forecast air quality levels for Budapest.


There are plans to use Chimere also for scenario studies. However, there are not resources to do

this presently. There may come a PhD student who could assist in such work.



We do not have this type of modelling activity. We plan to do this.


There has not been time to do such work until now, but the responded finds it very relevant.

Respondent suspects that stations in the monitoring network are not properly placed, and she has

plans to look into the matter.

71




required

We have no activity in this field.


We have no activities on this field.


Would like to use recommendations of FAIRMODE. Would appreciate that Fairmode would

arrange training courses where there is no registration fee, like it is the case for EMEP.


We are ready to take part.


The meteorological institute is the obvious candidate to become national competent authority.


This can be fulfilled by the costs mentioned in Option 3.


Takes part in ongoing international activities on emission inventories, but would like to take part in

Fairmode activities also.

72



Appendix C: Summary of open-ended responses – by issue

Member State BASELINE 1: COMPETENCE BUILDING – ATMOSPHERIC SCIENCE



Austria No. At the Austrian EPA there is currently no staff dedicated to modelling issues as there are no legal

requirements.

Belgium This competence building is available at the different regional environment administrations and the


Technological Research) are also playing an important role in air quality modelling in Belgium. Some of

the Regions indicate that competence in atmospheric modelling and air quality modelling will be further

developed, but additional efforts have to be made to achieve the specified minimum requirements. The

following calculation is an estimation of the existing annual costs of baseline 1. Some of the regions

indicated a change in costs to conform to the baseline 1 standard but no further details of the costs could

be provided at the moment.

Croatia Our member state conform to the specified standards when whole country is taken into the consideration

and expert employed in different institutions are accounted. However a team of at least 5 scientists

working in the field of atmospheric modelling should be developed in one institute that will be able to work

together and develop competence in atmospheric modelling to provide data for reporting on national

level, planning and other MS obligations toward EU.

Finland Finland meets the requirements. There is a Air Quality modelling group that develops the atmospheric

and air quality models for different scales (local, urban, regional, global..) (~20 Research Scientists and

PhD students) in FMI. In addition, there is Air Quality expert service group that offers dispersion modeling

services for various customers (Cities, industry etc..) in Finland and abroad.

France yes definitively. At the national level a team of 12 people (more than half of them being senior scientists)

is entirely devoted to air pollution modelling research and expertise. This team belongs to INERIS and

participates to the national reference laboratory for air quality (LCSQA). It works for the Ministry in charge

of the Environment. At the local level, associations in charge of air quality monitorig develop skill and

competences in air pollution modelling as well. A staff of about 30 local experts works can be mobilised.

Note that INERIS develops research projects in the fields of AQ modelling and hosts a number of PHD

students. On the material point of view, it benefits from high performance computing resources.

Germany Practice in Germany is contracting consultants in most iof the cases. Answers rely on assuming that

presumption made for this questionnaire are proper.

Hungary Not yet, there is only 1 full-time senior staff within atmospheric modelling, working with only one 10-year

old PC. Because of serious restrictions in the budget, there is no possibility to employ any more staff

persons and to purchase new computers at the moment.

Ireland No. Currently there is no capacity available for modelling. 2013 will see a full time fellowship devoted to

developing the specified criteria in CAFÉ with regard to modelling components of the Directive

Italy Yes, it does. ENEA (National Agency) support Ministry of the Environment in developing and maintaining

the Integrated Atmospheric Pollution National Model. ENEA has team on atmospheric research with

about 15 staff researchers, research fellowship and temporary researcher. They regularly attend to

international group (including Fairmode) and have links with foreign research group

Netherlands Yes, the NL spends more than the 3-5 full time persons for atmospheric modelling under 1). Dutch

research institutes and Universities together employ a number of PhD in atmospheric modelling but not

specifically for national use. Since NL already uses modelling for reporting AQ data, we do not fill out the

baseline sheets for building capacity but only the sheets on maintenance of the modelling in sheets under

Options.

Poland It is not known on what basis was estimated the above. minimum requirements. Each Member State has

its own approach to the problem of human resources. For example in CIEP modelling issues are carried

out on behalf of the CIEP by external institutions. One full-time employee of the doctoral title (responsible

for modelling) is currently employed in CIEP.

Romania YES

Sweden Yes, the Swedish Meteorological & Hydrological Institute (SMHI) have significant competence with regard

to atmospheric & air quality modelling. This far exceeds the estimated minimum resources above. Other

institutes and consultants also have a good degree of competence in air quality modelling.

73



Member State BASELINE 1: COMPETENCE BUILDING – ATMOSPHERIC SCIENCE

UK Modelling is already used for compliance reporting (alongside measurement data for assessment of

extent of exceedance), short term forecasting, source allocation, development of plans and measures,

network design and establishment of monitoring requirements . Its not clear how this data gathering

exercise is relevant to the FAIRMODE recommendations or the review of the Directive or where the

estimates have been produced from.

Member State BASELINE 2: COMPETENCE BUILDING – HIGH RESOLUTION EMISSION DATABASES



Austria No. There is currently great expertise at the EPA in compiling emission trends for the whole of Austria (one

figure for each emission category and year) but there doesn't exist a high resolution emission data base

available as input for atmospheric modelling. Some of Austrian's provinces maintain such emission

inventories, but these do not cover the whole country and are not harmonised yet.

Belgium This competence building is available at the different regional environment administrations and the Belgian

interregional Environment Agency. Subcontractors (like e.g. the Flemish Institute for Technological

Research) are also playing an important role in air quality modelling in Belgium. Some of the Regions

indicate that competence in geographic resolution emission databases is developed, but will be further

developed, but additional efforts have to be made to achieve the specified minimum requirements. The

following calculation is an estimation of the existing annual costs of baseline 2. Some of the regions

indicated a change in costs to conform to the baseline 2 standard but no further details of the costs could

be provided at the moment.

Croatia Emission data on 50x50 km horizontal resolution are being produced regularly. Emission inventory on

10x10 km horizontal resolution is developed. Staff development, database software, GIS software, new

emission data generation is foreseen.

Finland NA (Not FMI'sd Responsibility) SYKE, TIlastokeskus

France France is committed in the implementation of a National Emission Inventory for air pollutants which

compiles emissions of more than 50 pollutants over the French territory with high temporal and spatial

resolutions (until 1 hour and 1km). This national inventory (INS) will be operational in 2013. The 26 local

organisations in charge of air quality monitoring developed local inventories for each French region with

very high spatio-temporal resolutions. Necessary human and material resources (similar or better than the

minimum requirements set in the baseline description) to maintain and develop these tools at both levels

are currently implemented. Annual reporting according to the CLRTAP requirements is under the

responsibility of the CITEPA, an organisation gathering more than 20 experts in the field of emission

inventories.



Hungary Currently, there is a staff of 3 for both GHG and CLRTP reporting in the Hungarian Met. Service. However,

transport emission modelling has been outsourced, and the compilation of the forestry inventory is done by

the relevant governmental institute. We have had no education in databases and GIS, and generally self-

education is the dominant form of capacity building. We're at the very beginning of the process of building

national databases.

Ireland We do not model ambient air quality currently. We have not built either a high resolution or low resolution

emission inventory for this specific purpose. Data regarding emissions are collated by our organisation, but

in my own opinion, it would be a volume of work to adapt this into a spatial emission inventory. We

currently do not have the resources available to us to complete such a task.

Italy Though the National inventory is provided by ISPRA, ENEA performs many activities in order to prepare

the national inventory (annual on a county base) for feeding the atmospheric pollution national model

(hourly grid base inventory, speciation, etc)

Netherlands Yes, the Netherlands has an integrated system including all the above resources. Option 4 shows all the

costs involved in (1) yearly generating the annual emission data on SNAP level and (2) allocating this data

on a 1*1 km grid.

Poland National reference centre of emission inventories is KASHUE/KOBIZE (The National Administration of the

Emissions Trading Scheme). Their data are partly used for the purposes of the national modelling. In

addition, the emission bases are carried out by provincial inspectorates of environmental protection -

regional level.

74



Member State BASELINE 2: COMPETENCE BUILDING – HIGH RESOLUTION EMISSION DATABASES

Romania YES

Sweden Yes, SMHI have developed high resolution emission databases (although the quality & resolution differs

between emission categories). Some local authorities have also developed high resolution emission

databases.



design and establishment of monitoring requirements . Its not clear how this data gathering exercise is

relevant to the FAIRMODE recommendations or the review of the Directive.

Member State BASELINE 3: DEVELOPMENT OR ACQUISITION OF ATMOSPHERIC MODELS



Austria No. There is currently no dispersion model operated at the EPA.

Belgium Belgium does not have a junior staff for technical implementation and maintenance of models and

computer facilities to make the models FAIRMODE compliant. Some of the regions developed appropriate

atmospheric models operating with meteorological data (from the national meteorological institute) with a

spatial resolution of 1x1 km and a temporal resolution of 3 hours.

Croatia Air quality models for regional and urban scale are/can being used for assessment and reporting. Our

meteorological service provide appropriate meteorological data. Operational annual reporting including

model validation based on modelling is foreseen.

Finland Finland meets the requirements. There is a Air Quality modelling group that actively develops the

atmospheric and air quality models for different scales (local, urban, regional, global..) (~20 Research

Scientists and PhD students) in FMI.

France 1- the basic French model for regional air quality modelling, CHIMERE, developed by INERIS and the

national research centre is freely downloadable on the internet. Some local models used by the local

organisations can be acquired for a limited cost through partnership with research laboratories. Urban

models (like ADMS) are available through contracts with private companies. Cost for development and

maintenance is well framed for both operation and further development.

2- OK

3- OK. partnership with Meteo France is established. Meteo France recently developed a high resolution

model (2,5 km) which is now used for air quality

4- Reporting on the evaluation of air quality model: operational at the national level; at the local scale a

framing process is on-going under the coordination of the reference laboratory LCSQA to assess the

quality of models used by local organisations in charge of air quality monitoring



Hungary Existing models under maintenance (which were downloaded freely) are: AERMOD, CHIMERE. High

resolution meteorological data are available (AQ modelling is done by the national met. office).

Ireland I have already discussed this under baseline 1. Again I make the point regarding reporting of modelled

data to EEA. Will there be experts at that end to assess the models used by member states? I say this as it

was mentioned at the FAIRMODE meeting in Oslo a few years ago that modelled data will be dealt with

the same gravitas as monitored. Have there been any modelled exceedances of the limit values by

member states to date that have lead to cases been taken by the commission? I included a general rough

cost for license purchase, renewal and the purchase of model ready met data if required

Italy Yes it does. ENEA's team deals with all the requirements listed. High resolution met simulation with 1 hour

time resolution are produced directly starting from ECMWF analysis.

Netherlands Yes, The Netherlands already reports AQ data based on measurements and modelling.

Poland Modelling is carried out on behalf of the CIEP by external institutions.

Romania YES

Sweden Yes, there is a good range of atmospheric models available in Sweden, which are constantly under further

development.

75



Member State BASELINE 3: DEVELOPMENT OR ACQUISITION OF ATMOSPHERIC MODELS



design and establishment of monitoring requirements . It is not the case that models can be downloaded

freely or acquired for a small amount to undertake regulatory reporting, this is misleading. Its not clear how

this data gathering exercise is relevant to the FAIRMODE recommendations or the review of the Directive.

Member State BASELINE 4: COMPUTER FACILITIES FOR RUNNING HIGH RESOLUTION MODELS



Austria No such Power-PCs are available at the EPA for the moment.

Belgium Yes, considered that this facilities are not only present within the Belgian interregional Environment Agency

but also in the Regions and subcontractors.

Croatia Curently we are using SGI computer with 54 core for all our model runs at Meteorological Service.

However their capacity is limited and the computer is used by different groups (weather forecasts, climate

etc.). Therefore in orther to be able to produce modelled air quality data on operative basis an upgrade or

the purchase of new computers is needed

Finland Finland meets the requirements. FMI has enough computing power inhouse (super computer capacities).

France INERIS benefits from high performance computing systems for its studies for the ministry of Ecology:

- the national air quality forecasting and mapping system, PREV'AIR, www.prevair.org, is hosted by a 256

core computer run in a fully operational 24h/7days mode

- an access to a national high performance system, the CCRT, which hosts one of the most powerful

computer in Europe

At the local scale the local organisations benefit from appropriate computational resources.

Germany Practice in Germany is contracting consultants in most iof the cases. Answers rely on assumming that

presumtion made for this questionnaiere are proper.

Hungary There is only one 5-year old 2-core computer maintained (housing, power and cooling is OK).

Ireland more or less answered this in baseline 1 (staffing numbers in terms of running models). Would require 1 8

- 48 core processor. Staff requirements outline in tab 1

Italy High resolution models run on theENEA- CRESCO high performance computing infrastructurei, a cluster

based on x86_64 architecture; the main system is the Portici Cluster (HPL test 17.1 TFlops). Estimated

costs only for atmospheric modelling activities

Netherlands Yes, The Netherlands already reports AQ data based on measurements and modelling.


Romania YES

Sweden Yes, SMHI have adequate resources in this regard.



design and establishment of monitoring requirements . Its not clear how this data gathering exercise is

relevant to the FAIRMODE recommendations or the review of the Directive.

Member State BASELINE 5: OPERATIONAL / RUNNING COSTS



Austria No not at all.

76



Member State BASELINE 5: OPERATIONAL / RUNNING COSTS

Belgium Yes, we can provide the indicative cost made by the Belgian interregional Environment Agency and not

the operational/running cost for the regions or consultants. We cannot provide the different cost

categories but only one figure (1.235.000 euro).

Croatia Administration costs are satisfying. Accommodation requests are not fulfilled while traveling and mobility

of scientist currently working in atmospheric modelling is covered additional resources for new staff that

needs to be employed to conform with the minimum number (5 scientists) need to be assured.

Operational costs for maintaining readiness for the staff's ability for decision making support is not

fulfilled. Costs for the operational forecast system are not covered.

Finland Finland meets the requirements. FMI has 650 employees and about 20employees in administration.

France Ok for all issues

Germany Practice in Germany is contracting consultants in most of the cases. Answers rely on assuming that


Hungary No administration, no economic officer, office space for 2 personnel, serious restrictions in travelling,

there are limited opportunities to attend conferences. An operational AQ forecast system runs for

Budapest in additional to the met. office forecasting tasks and costs.

Ireland Again outlined in baseline 1

Italy Yes, see previous sheet

Netherlands Yes, The Netherlands already reports AQ data based on measurements and modelling and has an

operational forecast system.


Romania YES

Sweden Yes, SMHI have adequate resources in this regard.

UK Modelling is already used for compliance reporting (alongside measurement data for assessment of

extent of exceedance), short term forecasting, source allocation, development of plans and measures,

network design and establishment of monitoring requirements . Its not clear how this data gathering

exercise is relevant to the FAIRMODE recommendations or the review of the Directive. Many of these

aspects are arbitrary and irrelevant, especially if modelling is not done in house.

Member State Recommendation 1(a): Use AQ models for assessment of air quality levels to establish the extent


Does your Member State currently conform to the standards or practices specified in the

recommendations listed above? Please clarify your response

Austria Supposing that an atmospheric modelling group exists at the EPA according to the minimum

requirements described in PART 1, we assume that such a group would handle the above mentioned

application.

Belgium In Belgium deterministic models and/or intelligent interpolation techniques are used to calculate

population exposure and/or to evaluate measures. For the annual questionnaire concerning the

reporting of the air quality in the framework of the directive 2008/50/EC mainly measurements are used.

Some of the regions have developed models and algorithms in order to meet these recommendations :

modelling of pollutant concentrations and calculation of population exposure with a spatial resolution of

1x1km, modelling of emissions scenarios to manage source legislation or to optimise air quality

monitoring etc.

Croatia Models are used for the air quality assessment to provide information on air quality levels in zones that

not covered with fixed measurements. The application of models in agglomerations is envisaged and

further development of emission inventories are needed. Models will be used for annual air quality

reporting regarding the assessment of exceedances and exposure as a supplement to measurements.

Finland Finland meets the requirements. Cities and municipalities are responsible of local air quality

assessments made by measurements and modelling. In many cases cities outsource that service for the

air quality experts (FMI or other consultants).

77



Member State Recommendation 1(a): Use AQ models for assessment of air quality levels to establish the extent


France The national air quality forecasting and mapping platform, PREV'AIR, allows to simulate background

concentrations over France with a 5 km resolution. It is sufficient and relevant to assess urban

background concentrations, but local exceedances near busy roads or industrial sites require other

means. To focus on the local scale, the local associations developed local modelling systems that are

operational as well. Exposure assessment and mapping tools are available and use together with in-situ

measurements from the automatic network and field campaigns. Therefore, geographical extension of

the exceedances and the number of inhabitants concerned, can be evaluated and reported according to

the air quality Directive.

Germany Without making AQ assessment by modelling mandatory with clear implementing provisions Germany

most probably will not follow this option.

Hungary We do not use models for the investigation and assessment of ambient air quality.

Ireland No. The baseline tabs address all these questions with the assumption that we would be developing our

modelling capacity with the requirements of CAFÉ in mind

Italy Yes, it does. The national model runs on a 4x4 km grid, provide hourly concentrations. finer resolution

models up to few meters of resolution ( fluid dynamics model) are applied as well.

Netherlands Yes, The Netherlands reports AQ data on the bases of measurements and modelling. The Netherlands

uses AQ models fo scenarios studies and effectiveness of measures. The Netherlands has an

operational (short term) AQ forecasting system. Considering costs for the Assessment and planning of

AQ levels: this is a rather complex system with many partners. We have the feeling that the cost items

asked here would require a major operation to get costs of every organisation for these items. Instead

we tried to make estimates and sort of fitted it into your items. So this is a very crude estimation.

Poland Yes, in order to support the annual air quality assessments, the modelling of tropospheric ozone is used

in the whole area of Poland, as well as for individual provinces. Such works are contracted on behalf of

CIEP by external institutions. In addition, 3 of 16 above. provincial inspectorates are implementing

modelling for air quality assessments for pollutants other than ozone.

Romania NO

Sweden Models are widely used to assess air quality in the larger towns and cities with the worst air quality

problems. Competencies and practice are, however, varied between authorites, but a number of highly

competent consultants are available for modelling studies. Modelling studies are also carried out on a

national level. Modelling results are, however, not currently being reported directly in accordance with

the Air Quality Directive's requirements.

UK Modelling is already used for the assessment of air quality levels to establish the extent of exceedances

and establish population exposure for compliance reporting according to decision 2004/461/EC

Member State Recommendation 1(b): Use AQ models for forecasting air quality levels for short term mitigation


Does your Member State currently conform to the standards or practices specified in the recommendation

listed above? Please clarify your response

Austria In order to perform such simulations extremely fast computers were needed, due to the high spatial

resolution required for those pollutants with short term limit values (e.g. NO2, SO2, PM10). Forecasting O3

is a different story and requires usually other types of models (CTM) covering the regional scale.

Belgium In BE a number of air quality forecast models are used to inform the public and to trigger short term

emission reduction measures (e.g. speed limits on high ways during smog episodes).

Croatia No, we are not providing air quality forecasts.

Finland Finland meets the requirements. Cities and municipalities are responsible of local air quality assessments

made by measurements and modelling. FMI offers AQ forecasting services for the biggest cities in Finland.

France Operational in France (see previous option with the implementation of the PREv'AIR system and local air

quality forecasting platforms). For information, partnerships with national and local TV chanels are

established to inform the general public of exceedances of limit values when air pollution episodes occur.

Moreover forecasting capacities of the PREv'AIR system are used to assess the effectiveness of short

terms action plans. In France a new regulation is about to be taken to implement short term action plans

when PM10 concentrations (daily means) exceed the 50 ug/m3 threshold over a given geographical area

for several days. The implementation of this provision is based on the results of forecasting systems.

Germany Germany's forecasting and information systems already fit with requirements.

78



Member State Recommendation 1(b): Use AQ models for forecasting air quality levels for short term mitigation


Hungary We use CHIMERE chemical transport modell to forecast air quality levels for Budapest. In the future we

plan to use air quality model to evaluate the air quality levels in Hungary.

Ireland No. This requirement should be a first step on the road to full compliance with requirements. It could be

quickly achieved by most member states at a smaller cost than the other elements. Already there is an

ensemble of modelled data available through MACC / GMES outputs for each country. Freely available

met data and back tracjectory tools could also be incorporated.

Italy Models are used to assess the impact of sources (point, linear etc) on demand. costs not estimated

Netherlands Yes. See under option 1a.

Poland Task in progress in terms of tropospheric ozone.

Romania NO

Sweden Some local authorities forecast air quality levels for short term mitigation and public information. We are,

however, unsure of the exact methods used and their complexity.

UK Modelling is already used for forecasting air quality levels for public information and warnings

Member State Recommendation 1(c): Use AQ models for source allocation to determine the origin of



recommendation listed above? Please clarify your response



application.

Belgium Source allocation is mainly performed via consultancy assignments e.g. in the framework of time

extension demands, plans and programmes when limit values are exceeded. In some Regions local

monitoring campaigns (frm 6 weeks to 6 months) are aimed to determine exposure and origin of

emissions for regulation strategies purposes. Theses programs are supported by mineralogic/chemical

analysis of PM (scanning electron microscopy coupled to a spectrometer energy dispersive analysis

(EDX), and the overall analysis using the X-ray diffraction (XRD)). Additional investments will have to be

done. However an estimation of the costs to conform to this standards is not feasible at the moment .

Croatia Yes. The source allocation techniques (trajectories, models) have been used. However permanent

scientific development and application of new methods is needed. The increase in reporting obligations

is foreseen.

.

Finland Not in operational use in FMI, only done in research projects

France Several field campaigns have been set up in French regions during the two last years to caracterise PM

episodes. Measurements of PM chemical composition and other parameters were available from these

campaigns and first investigations to allocate sources were done. Moreover when PM episodes occur a

specific network of chemical samplers is activated so that the composition of PM is analysed. Those

results are compared to simulations run with the chemistry-transport model CHIMERE to improve our

understanding of the source contributions. This is actually fundamental for planning control strategies.

Such a integrated system is running in France since 2008.

Germany Considerations according to option 1(c ) are already practice where necessary

Hungary We use AERMOD model for regulatory purposes to assess the future effect of a new investment.

Ireland would require development. This has been outlined previously, with estimate of time / staff requirements

given.

Italy source apportionment and origin of exceedances have been made for PM and NO2 over Italy. results

are documented in the communications to UE

Netherlands Yes. The modelling under Option 1a also gives information on the contribution of sources to the

concentration at a location. Costs are therefore integrated in Option 1a.

Poland We act in accordance with regulation of Minister of the Environment of 26 January 2010 on reference for

certain substances in ambient air (OJ No 16, 2010 item 87), where in annex No. 3. reference modelling

methodology of substances levels in the air are specified. In the case of applying for a permit/ license,

for the purpose of calculating the pollutants spread, a mathematical model, described in the above.

regulation has to be used.

79



Member State Recommendation 1(c): Use AQ models for source allocation to determine the origin of


Romania YES, using outsourced services. It has been used but not at the national level. Will be extended as a

compulsory request for air quality plans development.

Sweden Models are used to differing extents by Local Authorities for this application, but the larger cities

generally conform to these standards. It is however, possible, that practice could be improved for this

application.

UK Modelling is already used for source allocation

Member State Recommendation 1(d): Use AQ models for development and assessment of plans and measures

to control AQ exceedences



Austria The development of air quality plans with the help of air quality models can be quite extensive. In Austria

hundreds of measures in cities and regions were to be evaluated. This work is currently done by experts

at the province level often supported by Universities.

Belgium In BE deterministic models are used to evaluate emission reduction measures and to develop plans and

programmes. Not every Region uses deterministic models to evaluate measures but developments are

ongoing. Additional investments will have to be done. However an estimation of the costs to conform to

this standards is not feasible at the moment.

Croatia Yes, we have used models for development of national plans eg. for ozone. Models need to be


Finland Cities and municipalities are responsible of local air quality assessments made by measurements and

modelling. FMI offers dispersion modelling services for the cities.

France The CHIMERE model is run at the national scale by INERIS to assess the effectiveness of national and

sectoral control measures. At the local/urban scales local organisations run appropriate models to

evaluate additional local control measures. In fact both national and local measures are systematically

assessed by modelling systems.

Germany Is an essential part when setting up air quality plans

Hungary We use CHIMERE chemical transport model to forecast air quality levels for Budapest.

Ireland no.

Italy the effectiveness of national and regional measures has been assessed using both atmospheric

pollution model and GAINS Italy

Netherlands Yes. The calculations of the AQ levels over recent years and near future are made in the same

modelling system. So costs are integrated in Option 1a.

Poland Procedures for preparation and evaluation of the above. plans and activities are carried out in

accordance with the Regulation of Minister of the Environment (ME) of 26 January 2010 on the

reference of certain substances in the air (Journal of Laws 2010 No. 16, item. 87) and the Regulation of

ME of 11 September 2012 on air protection programs and short-term action plans (Journal of Laws of

2012, item. 1028). This regulation takes into account the obligations imposed by Directive 2008/50/EC.

Romania not yet, but it will be implemented as soon as possible

Sweden Models are used to differing extents by Local Authorities for this application, but the larger cities

generally conform to these standards. It is however, possible, that practice could be improved for this

application.

UK Modelling is already used for development and assessment of plans and measures to control AQ

exceedences

Member State Recommendation 1(e): Use AQ models for designing monitoring networks when models are used






application.

80



Belgium The use of AQ models in combination with monitoring to design monitoring networks is not a common

practice in BE yet. The use of these techniques will generate a supplementary cost for BE, however an

estimation of these costs is not feasible at the moment.

Croatia Yes, we have used model results for assessment and design of complience network. Further

applications are foreseen.

Finland Cities and municipalities are responsible of local air quality assessments made by measurements and

modeling. FMI offers dispersion modeling services for the cities and municipalities.

France Use of models to design monitoring networks is not routinely practised in France. The number of stations

and their locations are defined by the local monitoring networks according to the Directive requirements

and national and local needs. Recent studies started to assess the adequacy of the network. They are

based on model results.

Germany Without making the design of monitoring networks in combination with modelling mandatory together

with clear implementing provisions Germany most probably will not follow this option.

Hungary We do not have this type of modelling activity. We plan to do this.

Ireland no. currently ambient air modelling is not done

Italy different model techniques, detailed emissions on grid, GIS elaborations are currently used for the best

siting and to determine the minimum number of stations of the national network for HM, Hg, POPs, HPA.

Netherlands Probably yes. The Netherlands uses models to get insight in the distribution of the AQ. According to that

distribution and a number of other criteria the monitoring sites are chosen. We see no specification in the

recommendation here (the text is still very generic) so we cannot fully answer this question. Considering

costs: running the model for designing networks and to fix the locations (option 1f) is a small activity

compared to the yearly workload under option 1a so we have not specified this further.

Poland The network of measuring stations are designed in accordance with the results of the five-year air

quality assessments, but also using the results of modelling.

Romania No

Sweden Most monitoring networks operated in Sweden have been established for many years. We are unsure of

the exact role models played in their development. With regard to development of new monitoring

networks, the use of models for this application is varied. Some local authorities do consider models, but

it seems that the majority do not. This is an area in which practice needs to be improved and guidance

from FAIRMODE would be very welcome.

UK Modelling is already used as an input for the specification of monitoring networks when models are used


Member State Recommendation 1(f): Use AQ models for determining the number of fixed monitoring sites that

are required





application.

Belgium The use of AQ models to determine the number of fixed monitoring sites is not yet a common practice in

BE. The use of modelling for this purpose will generate a supplementary cost for BE, however an

estimation of these costs is not feasible at the moment.

Croatia Yes, we have used models for development of national plans eg. for ozone. Models need to be


Finland in Finland, the amount of fixed monitoring stations in various areas is determined by the amount of

population in the defined monitoring zone(according to the directive specifications about zones and

agglomerations).

France See answer for option e

Germany Without making option 1(f) mandatory together with clear implementing provisions Germany most

probably will not follow this option.

Hungary We have no activity in this field.

Ireland no. currently no ambient air quality modelling is completed

Italy see previous sheet

Netherlands Yes, see considerations under option 1e.

Poland The network of measuring stations are designed in accordance with the results of the five-year air

quality assessments, but also using the results of modelling.

81



Member State Recommendation 1(f): Use AQ models for determining the number of fixed monitoring sites that

are required

Romania No

Sweden With the odd exception, only monitoring results are used to determine the number of fixed monitoring

sites required. Guidance from FAIRMODE on this application would be welcomed.

UK Modelling is already used as an input for determining the number of fixed monitoring sites that are

required

Member State OPTION 2: MODEL QUALITY OBJECTIVES (MQO)



Austria No. In our opinion, the current MQO of the Air Quality Directive aren't ambiguous, especially the existing

guidelines helped to clarify open questions. We do not recommend to change the MQO. The

implementation of new, stricter MQO (DELTA-tool in its current version) will certainly increase time and

effort in respect to modelling in Austria due to the complex topography a lot. Additional costs can't be

estimated, as it is not clear whether state-of-the-art models are even able to provide results meeting the

new MQO.

Belgium BE currently uses data quality objectives (validation statistics) and will continue to actively follow up the

activities within the FAIRMODE working group concerning the model quality objectives. In some regions

model uncertainty is calculated for interpolation and/or dispersion models and model validation is done

with supplementary measurements from mobile measurement campaigns.

Croatia We are actively involved in the work of Fairmode regarding MQO and the revision of existing objective in

the Directive is needed.

Finland FMI supports the idea of preparing the guidance document for the quality assurance for air quality

modelling . FMI is interested to participate also in future in the preparation of the QA document.

France The MQO mentioned in the 2008 AQD are not precise enough for operational use. Therefore we are in

favour of more stringent and framed quality objectives for model use. We follow carefully the

developments of the FAIRMODE methodology to assess model quality. The CHIMERE model has been

tested against the FAIRMODE criteria. In 2013, the same criteria will be used to assess the quality of the

local models used in France.

Germany Quality assurance is integral part when modelling.

Hungary We have no activities on this field.

Ireland no. Clarification is definitely required

Italy At the present fully involved in Fairmode activities. MQO are tested continuosly.

Netherlands Yes, The Netherlands conforms to the MQD as mentioned in the 2008 AQ directive. Considering the

revision of the MQA: it is not clear yet what criteria and what targets for those criteria will be. So it is not

possible to answer this question further.

Poland The modelling results in Poland meet the requirements of the Directive 2008/50/EC in terms of data

quality objectives. This recommendation is vague, lack of a detailed cost estimate, with undefined data

quality objectives for modelling.

Romania Yes

Sweden FAIRMODE's existing guidance on calculation of model quality objectives is currently used within

Sweden, although their use is varied and by no means standard. However, when compliance of model

results with the quality ojectives has been assessed, they have been shown to easily comply with the

requirements. More stringent MQO's are needed and this recommendation is strongly supported by

Sweden.

UK We support and are involved with the work that FAIRMODE is undertaking to propose new data quality

objectives for modelling. This work is currently in progress and is not sufficiently developed for an

assessment to be made of the possible cost associated with new data quality objectives. FAIRMODE is

not recommending revising the DQOs at this stage.

82



Member State OPTION 3: FORUM OF EU AQ REGULATORY MODELLING

Does your Member State currently conform to the standards or practices specified in the recommendation

listed above? Please clarify your response

Austria No.

Belgium Yes, BE has nominated a competent authority for the FAIRMODE activities.

Croatia We do not have nominated competent authority for modelling and this Fairmode activity is fully supported.

Finland The idea of independent competent authority is good idea in order to ensure the quality of modeling made

by different players (FMI, consultants).

France The competent authority nominated for metrological issues (linked with AQUILA) is the Central Laboratory

for air quality monitoring to which belongs INERIS. Experts from the LCSQA (and INERIS) will be

nominated for air quality modelling issues as well.

Germany Answer depends on the concrete requirements. An estimate is not possible currently.

Hungary We are ready to take part.

Ireland no air quality modelling completed. There definitely is a need for clarification for both member states and

FAIRMODE members.

Italy ENEA researchers have been appointed by Ministry of Environment to partecipate in Fairmode groups,

moreover a national forum for air quality modellers follows the international debate

Netherlands RIVM will act as the Competent authority for modelling. Since we already do so we do not think it will cost

much extra. However, this depends on the activities and questions asked by FAIRMODE. There are no

specifications in this recommendation further so it is not clear to us where to conform to.

Poland The recommendation is not an obligation included in EU law, therefore, it is not carried out by Poland. The

inability to assess the cost without the prior presentation of a range of concrete proposals for action. the

body is the lack of such information in the poll. Inability to assess the costs without presenting detailed

proposals for the scope of activities of above. body – there is no such an information in this form.

Romania Yes

Sweden Sweden has already established a NRL for air quality modelling at SMHI. It was, however, only established

in 2012 and does not yet fully conform to all points in the list of supporting motivations.

UK We do not think that there will be any costs associated with this recommenation. Not possible to determine

a baseline.

Member State OPTION 4: QUALITY ASSURANCE AND CONSISTENCY OF EMISSION INVENTORIES





requirement.

Belgium In BE work concerning the improvement of the emission inventories is ongoing. If supplementary quality

objectives have to be met in the framework of FAIRMODE this will generate an investment, however an

estimation of these costs is not feasible

Croatia

Finland NA, Syke, tilastokeskus

France A national working group has been created in France to define the basic rules and methodologies that

should be used to develop local and national emission inventories. Common databases are used to this

end. In France each administrative region is covered by a high resolved emission inventory and France

is defined with the national emission inventory, so-called INS.

Quality assurance plans are required for local and national inventories and audits will started in 2013.

Germany This effort is undertaken contiguously. As long as specification of requirements is open we assume

status quo.

Hungary This can be fulfilled by the costs mentioned in Option 3.

Ireland No . This is a fair recommendation. The same set of criteria need to be set for the models themselves

also. As stated earlier, an expert on that particluar modelling system is needed on the EEA side of the

fence ( similar to AQUILA, and the intercomparision excerises that are carried out annually)

Italy ENEA harmonize national emissions with regional ones in order to check consistency of top-down and

bottom-up approaches

Netherlands Yes, The Netherlands has a competent authority that makes emission inventories (at 1x1 km scale),

reviews the emissions and reports them to international bodies. More detailed emission data (on street

83



level) are available via municipalities. These emissions are also used for the AQ assessments.

Poland Appropriate action is taken by the CIEP to refine the emission data for modelling.

Romania YES

Sweden There is a significant amount of work carried out within Sweden regarding the compilation and quality

assurance of emissions data suitable for AQ modelling. We are very supportive of this recommendation

and are keen to partake in this work.

UK This recommendation provides a general statement of intent which is pragmatic. However it is not

possible to assess the implications on costs of this recommendation.

84



Appendix D: Bottom-up approach. Graphs for each of the 14 questions

0

1000000

2000000

3000000

4000000

5000000

6000000 Total cost of baseline 1 - bottom-up approach



Additional staff





0

200000

400000

600000

800000

1000000

1200000

1400000

1600000




Additional staff





85



0

200000

400000

600000

800000

1000000

1200000

1400000 Total cost of baseline 3 - bottom-up approach



Additional staff





0

200000

400000

600000

800000

1000000




Additional staff





0

500000

1000000

1500000

2000000

2500000




Additional staff





86



0

500000

1000000

1500000

2000000

2500000

3000000Total cost of option 1a - bottom-up approach



Additional staff





0

20000

40000

60000

80000

100000

120000

140000

160000Total cost of option 1b - bottom-up approach



Additional staff





0

20000

40000

60000

80000

100000

120000

140000

160000

180000 Total cost of option 1c - bottom-up approach



Additional staff





87



0

50000

100000

150000

200000

250000Total cost of option 1d - bottom-up approach



Additional staff





0

50000

100000

150000

200000

250000

300000

350000Total cost of option 1e - bottom-up approach



Additional staff





0

50000

100000

150000

200000

250000

300000

Total cost of option 1f - bottom-up approach



Additional staff





88



0

20000

40000

60000

80000

100000

120000

140000




Additional staff





0

10000

20000

30000

40000

50000




Additional staff





0

200000

400000

600000

800000

1000000

1200000

Total cost of option 4 - bottom-up approach



Additional staff





89



Appendix E: FAIRMODE Consultation Survey of the top-down assessment

In this appendix, the results from the consultation survey on the top-down approach, conducted in

March 2013 are provided. The idea of the consultation survey was to verify the assumptions of the

minimum requirements for building up the capacity and competences needed for being able to fulfill

the FAIRMODE recommendations. For all questions, the member states were asked to assess the

estimated level of cost and to make a comment as follows:

Please chose a category and comment as appropriate

Categories Comment

100% too high

75% too high

50% too high

25% too high

Acceptable

25% too low

50% too low

75% too low

100% too low

Background

In November 2012, you received an Information Request asking for possible costs and resources

required for your Member State should the FAIRMODE recommendations be implemented. Thank

you to everyone who took the time to respond to this request. We have analysed the data received

and estimated the individual costs and resources across all Member States.

As a supplement to the results from the previous Information Request, we have made an alternative

assessment, using a top-down approach, where all countries are calibrated against an expert

estimate of the minimum requirements necessary for fulfilling the FAIRMODE recommendations.

We would like to provide you with the opportunity to review our findings to date and comment on

the validity of the results.

The following short consultation survey should take 15 minutes to complete.

We appreciate your participation in this consultation.

Introductory Questions

Question: What are your roles / responsibilities in your Member State?

Belgium

The Belgian Interregional Environment Agency (IRCEL-CELINE)

The telemetric air quality monitoring network in Belgium was initiated in 1979 and managed by the

federal Institute for Hygiene and Epidemiology (IHE). Following the transformation of Belgium into a

federal state with 3 regions, it was decided to split the network into 3 parts which were transferred

to the Flemish, the Brussels Capital and the Walloon Region in 1993.

It was decided to organize an interregional cooperation on a permanent basis through an official

cooperation agreement between the Regions (May 18, 1994).

90



This involves collaboration on the management of the monitoring networks and enforcing a

common scientific basis for collecting, processing and reporting air quality data.

The Belgian Interregional Environment Agency (IRCEL-CELINE) was the organisation created to

meet these goals.

Among other tasks, the agency runs :

an interregional calibration laboratory enforcing comparability of quality assurance and control

between the regional networks concerning air quality measurements.

an interregional air quality database where data from the regional networks are collected,

validated and processed into reports and studies.

an interregional data processing centre (IDPC) where air quality data collected in the regional

networks are acquired “real-time” , evaluated and used to predict and to inform the public during

these events. These real-time data are also published online: www.irceline.be.

air quality models used for air quality forecast, air quality assessment and scenario-analyses.

IRCEL-CELINE informs the regional, the federal authorities and the public on and during episodes

of enhanced air pollution (SMOG episodes), e.g.. informing or alerting the public when the ozone

EU information/alert thresholds of 180/240 µg/m3 are exceeded, activating the “SMOG90” measure

(speed limits on highways) during wintersmog episodes.

IRCEL-CELINE compiles the transmissions of all Belgian air quality and emission data to

international platforms such as EEA, WHO, OECD, EU, UNFCCC, EMEP-CLRTAP, EUROSTAT,

European commission, …

Staff members of IRCEL-CELINE are national experts in different European Air Quality expert

groups (e.g. the Air Quality Committee, FAIRMODE, AQUILA, EMEP TFMM, …).

IRCEL-CELINE is also the National Focal Point (NFP) for the European Environmental Agency

Croatia

I am leading the unit responsible for air quality research that includes the production of air quality

assessment reports and studies based on measurements and modelling data that are necessary for

regular air quality reporting in our country. Development of air quality modelling systems that would

be used for air quality management at national and local level is an important part of our work.

The Czech Republic

Relating to the work of FAIRMODE, one of the CHMI's responsibilities is to function as the

reference authority for modelling air pollution and evaluating air quality for European regulatory

purposes for the Czech Republic.

Ireland

Competent Authority with regard to the implementation of CAFÉ Directive and 4th

Daughter

Directive

Latvia

Latvian, Environment, geology and meteorology centre – authority responsible for Ambient Air

Quality Assessment in Latvia

UK

Competent Authority

91



Introduction to the FAIRMODE Options Analysis

This project is assessing the possible costs and resources for your Member State in case of a

possible implementation of the recommendations from FAIRMODE (The Forum for Air quality

Modelling in Europe). FAIRMODE is a joint response action of the European Environment Agency

(EEA) and the European Commission Joint Research Centre (JRC) (see also:

http://fairmode.ew.eea.europa.eu/).

The FAIRMODE recommendations propose the use of models for air quality assessments,

forecasting, source allocations, monitoring network optimization, amongst others. Furthermore, it is

recommended to revise the data quality objective for modeling, to establish a FAIRMODE forum for

competent authorities for modeling, and to improve the quality of emission data.

However, the recommendations are in their nature not very specific. Especially, the

recommendations do not address the requirements for modeling scale (regional, urban background,

urban street), number of chemical species or quality or resolution of models or emissions.

As air pollution covers all scales, we have therefore assumed in the expert estimate of required

resources that atmospheric modelling at all scales (regional, urban background and urban street) is

necessary, since it is not possible to take into account the effect of non-linear atmospheric

chemistry in e.g. cities, if the boundary conditions are not well described. As a basis for the expert

estimate, we have assumed that Member States need to run models at all scales to be able to e.g.

carry out assessments at all scales within the countries – both in rural and urban areas.

In the following sections we ask you to consider our assessment of the following: 1) Minimum

Personnel Resources, 2) Direct Operational Costs

Minimum Personnel Resources

The minimum requirements for any country is considered to be the competences of at least one full

time scientist or staff member at the senior level within the following four areas:

Task Minimum Resource Required

1 Regional Scale Modelling 1 modelling senior scientist

2 Urban Background Modelling 1 modelling senior scientist

3 Urban Street Modelling 1 modelling senior scientist

4 High-resolution emissions 1 emission senior staff member

It is assumed that a staff of this size – possibly counting more physical persons, and possibly after

a multi-year period of building up competences – are working almost entirely with modelling tasks

associated with the EU Ambient Air Quality Directive and the FAIRMODE recommendations. It is

irrelevant who pays the staff (government, regional agencies or local authorities or any other).

Question 1: Do you think that the estimate given above for minimum requirements for the personal

resources and competences needed for fulfilling the FAIRMODE recommendations in your country

is adequate?

Belgium

100 % too low since we estimated for the 5 baselines for the build-up phase (in the first

questionnaire) that a staff size of 22.5 was already needed. To comply with the recommendations

we also indicated that supplementary staff will be needed.

http://fairmode.ew.eea.europa.eu/

92



Croatia

50% too low. In our country high resolution emissions and the development and application of

models for urban scale still need to be built up.

The proposed number of scientists is acceptable if such modelling system is already at operative

level.

However in countries where this is a task to go at least two more scientists are needed.

The Czech Republic

25% too low. At first glance I would say yes. But one observation is that meteorological modelling

has not been included. And forecast operation might also require separate time. Therefore I have

said 25% too low.

I agree that a staff of such size might involve more physical persons.

I also think that in some circumstances, emission work would require more than 1 person's time,

but that it might be accommodated by the time from the other modelling tasks.

Ireland

25% too high. Initially, in Year 1 we would anticipate that the above resources would be a starting

point. Of course when you have never put such an extensive modelling project into operation

previously, only time will tell. From year 2 onwards, we would expect that 3 persons

would be a reasonable estimate.

Latvia

100% too low. Latvia work with Regional scale Modelling (Swedish company Opsis model

ENVIMAN). To get the job done we require 3 people as minimum.

UK

This estimate is too low by at least a factor of 3. The UK is currently using air quality models to

support compliance assessment, source apportionment, baseline projections, development if air

quality plans and review of air monitoring networks, so we have a good understanding of the costs.

Too low by more than 100%.

93



Direct Operational Costs

There will be direct operational costs, which are estimated as follows:

Resource Description Cost Description Total Estimated Annual Cost

One normal PC per staff member 4 PC x €1,500 / 3 years

€2,000 per year

Three computing servers for forecasting

and model calculations

3 servers x €10,000 / 3 years €10,000 per year

Licenses are needed for software for the

PCs and computer servers - GIS

software, database software, Fortran

and C compilers, visualization tools for

visualizing model results

€3,000 per year €3,000 per year

Travelling for participating in meetings

related to FAIRMODE and emission

quality

4 trips x €1,500 per trip €6,000 per year

Annual cost for training of staff member €667 per staff member on average €2,668 per year

Total Annual Direct Operational Costs €23,668 per year

This gives a total annual direct operational cost of €23,668 per year. Indirect operational costs as

e.g. administration, management, housing, etc. is not included here but assumed covered by

overhead.

Question 1: PC Requirements

Do you think that the estimate given for minimum PC requirements for the direct operational costs

(excluding overhead) needed for fulfilling the FAIRMODE recommendations in your country is

adequate?


One normal PC per staff member 4 PC x €1,500 / 3 years

€2,000 per year

Belgium

100% too low, since our staff estimates are much higher.

Croatia

50% too low. In accordance to the previous comment 2 more PC are needed, e.g. 3,000 per year

would be acceptable.

The Czech Republic

1 PC per person is adequate.

Ireland

Acceptable. One off purchase. Shouldn’t impact hugely

Latvia

Costs per PC Acceptable.

UK

Acceptable

Question 2: Server Requirements

Do you think that the estimate given for minimum server requirements for the direct operational

costs (excluding overhead) needed for fulfilling the FAIRMODE recommendations in your country is

adequate?

94




Three computing servers for forecasting

and model calculations

3 servers x €10,000 / 3 years €10,000 per year

Belgium

We accounted for 12 servers (each 7.000 euro) instead of 3 servers.

Croatia

This is acceptable.

The Czech Republic

Acceptable. I would say that 2 (larger) servers is the bare minimum. But it depends on the servers –

disk space, computing cores etc. And I'm not sure what 10k EUR buys these days.

3 servers allows one to run a forecast, a met model, and aq scenarios separately and therefore is

more convenient and makes running simulations easier. 3 would be acceptable.

Ireland

Acceptable. Again, the server is a once off purchase.

Latvia

3 servers are operated constantly. Offered rate per server Acceptable.

UK

Acceptable.

Question 3: License Requirements

Do you think that the estimate given for minimum license requirements for the direct operational


adequate?


Licenses are needed for software for the

PCs and computer servers - GIS

software, database software, Fortran

and C compilers, visualization tools for

visualizing model results

€3,000 per year €3,000 per year

Belgium

75% too low since licenses for GIS-software, computational software, etc. seems to be more

expensive then 3.000 euros a year.

Croatia

75% too low. GIS software that is compulsory tool for visualization of modelling results first needs to

be acquired and the price for the accomplishment of baseline is somewhat higher. Depending on

different software versions the price might be up to 30.000, Eur.

Licensing for compilers seems underestimated (e.g. one license for PGI Fortran for Linux

workstation is about 440 Eur and for 6 PCs it would be annual cost of 2625 Euro). Furthermore,

visualization tools and database software also might be at the range of 2,000 Eur per year.

If the price of GIS acquisition is not taken into account the estimated annual cost is approximately

75% too low.

95



The Czech Republic

Acceptable. These software items are necessary.

Ireland

Acceptable

Latvia

At the moment no new license costs are planned. If new licenses needed proposed costs are 100%

too low.

UK

Far too low, probably by more than a factor of 10. GIS costs are high, not all GIS systems are free,

similarly the air dispersion models that we use are also commercial software. Met data is also not

free.

Question 4: Travel Requirements

Do you think that the estimate given for minimum travel requirements for the direct operational


adequate?


Travelling for participating in meetings

related to FAIRMODE and emission

quality

4 trips x €1,500 per trip €6,000 per year

Belgium

Seems acceptable.

Croatia

This estimate is acceptable.

The Czech Republic

Acceptable. Cost per trip seems reasonable.

Ireland

50% too high. I think 1500 per person to cover travel and subsistence at a 2 day meeting in a

centralized European location is way too high

Latvia

Acceptable

UK

Acceptable

Question 5: Training Requirements

Do you think that the estimate given for minimum training requirements for the direct operational


adequate?


Annual cost for training of staff member €667 per staff member on average €2,668 per year

96



Belgium

The cost per staff seems acceptable, although in our estimates we have much more than only 4

staff members.

Croatia

50% too low. Since the high resolution modeling system needs to be established more training at

the beginning of the process is required. This includes the cooperation with the developed

modelling groups in EU and the mobility component needs to be taken into account.

Minimum estimate for this component would be around 1000€ per person and 6 persons are

proposed as a minimum for our country leading to the total estimated annual cost of 6000 Euro.

The Czech Republic

50% too low. We recently sent a colleague to Reading for 4 days of training. Total cost was about

1000 EUR. The cost of the actual training was free, so that number includes travel and

accommodation etc. only. Therefore I would say the estimate is too low.

Ireland

100% too low. In my own experience the training costs associated with models are high. They also

tend to be in one location in Europe ( eg calpuff, aermod training), so traveling costs and

subsistence have to be considered.

Latvia

Acceptable

UK

Acceptable

The Czech Republic

The competence building phase will be important because I perceive there to be a current lack of

availability of such staff with the required aq modelling-specific experience. In other words, the right

people might (do) currently exist but not all in the one institute and not having a role dedicated to aq

modelling for European regulatory purposes.

UK

Costs have been assumed to be the same in all MS (in terms of staff time). This is unlikely to be the

case. Costs are likely to be higher in MS with more complex air quality situations or more

exceedances. The costs will also be crucially dependent on the organisational structure of the MS.

In some MS air quality assessment and management is carried out by regional authorities and in

such cases the total cost for the MS may be many times higher.

Thank You

Many thanks for completing this consultation.

If you have any additional comments then please provide them here

97



Appendix F: Table 2.1, enlarged

Co

deRe

com

men

datio

nSu

b-O

ptio

n-FU

LLEC

Adm

inCo

stEC

Adm

inBu

rden

MSA

dmin

Cost

MSA

dmin

Burd

enAQ

Dire

ctAQ

Indi

rect

Info

rmat

ion

Man

agem

ent

Regu

latio

nEq

uity

Cohe

renc

e

1. A

sses

smen

t of a

ir qu

ality

leve

ls to

esta

blis

h th

e ex

tent

of e

xcee

danc

es a

nd

esta

blis

h po

pula

tion

expo

sure

neut

ral

---

--ne

utra

l++

++++

++++

+

2. F

orec

astin

g ai

r qua

lity

leve

ls fo

r sho

rt te

rm

miti

gatio

n an

d pu

blic

info

rmat

ion

and

war

ning

s

Neu

tral

neut

ral

---

++

++++

+++

+

3. S

ourc

e al

loca

tion

to d

eter

min

e th

e or

igin

of e

xcee

danc

es a

nd to

pro

vide

a k

now

ledg

e

basi

s for

pla

nnin

g st

rate

gies

neut

ral

---

--++

++´+

+++

++++

+

4. D

evel

opm

ent a

nd a

sses

smen

t of p

lans

and

mea

sure

s to

cont

rol A

Q e

xcee

danc

esne

utra

lne

utra

lN

eutr

alN

eutr

al++

++++

++++

+++

•D

esig

ning

mon

itori

ngne

twor

ksw

hen

mod

els

are

used

inco

mbi

natio

nw

ith

mon

itori

ng

neut

ral

neut

ral

+++

neut

ral

+++

+N

eutr

al++

Neu

tral

• De

term

inin

g th

e nu

mbe

r of f

ixed

mon

itorin

g si

tes t

hat a

re re

quire

dne

utra

lne

utra

l++

+ne

utra

l++

++

Neu

tral

++N

eutr

al

F22.

MO

DEL Q

UALI

TY O

BJEC

TIVE

SFA

IRM

ODE

reco

mm

ends

are

visi

onof

the

data

qua

lity

obje

ctiv

e fo

r mod

ellin

g-

--

-ne

utra

l+

++

+++

+

F33.

FO

RUM

OF

EU A

Q R

EGUL

ATO

RY

MO

DELL

ING

FAIR

MO

DEre

com

men

dsth

atin

para

llel

to

wha

tha

sal

read

ybe

enes

tabl

ishe

dfo

rth

e

mon

itorin

gof

air

qual

ity,

com

pete

nt

auth

oriti

esfo

rm

odel

ing

activ

ities

are

nom

inat

edby

the

Mem

ber

Stat

es(r

efAr

ticle

(3)

and

bulle

td)

qual

ityas

sura

nce

of

mod

ellin

g)

--

--

neut

ral

++

++++

++++

F4

4. Q

UALI

TY A

SSUR

ANCE

AN

D

CON

SIST

ENCY

OF

EMIS

SIO

N

INVE

NTO

RIES

FAIR

MO

DE re

com

men

ds to

inve

stig

ate

and

impr

ove

the

com

pila

tion,

cons

iste

ncy

and

qual

ity a

ssur

ance

of e

mis

sion

s dat

a su

itabl

e

for A

Q m

odel

ing

unde

r the

dire

ctiv

e

--

----

neut

ral

++++

++++

++++

Cost

s and

Bur

dens

Bene

fits

F1

1. O

N T

HE U

SE O

F M

ODE

LS F

OR

REGU

LATO

RY P

URPO

SE A

ND

TO

SUPP

ORT

AIR

QUA

LITY

PO

LICY

Sound analysis, inspiring ideas

BELGIUM – BULGARIA – HUNGARY – INDIA – THE NETHERLANDS – POLAND – RUSSIAN FEDERATION – SOUTH AFRICA – SPAIN – TURKEY – UNITED KINGDOM

P.O. Box 4175

3006 AD Rotterdam

The Netherlands

Watermanweg 44

3067 GG Rotterdam

The Netherlands

T +31 (0)10 453 88 00

F +31 (0)10 453 07 68

E [email protected]

W www.ecorys.nl

Review of: Provisions for Air Quality Measurement, Air ... 2 FAIRMODE Final pos… · Review of:...

Documents

Transcript of Review of: Provisions for Air Quality Measurement, Air ... 2 FAIRMODE Final pos… · Review of:...