Potential and limitations of using soil mapping ... · Soil plays a key role in hydrology since its...

Hydrol. Earth Syst. Sci., 15, 3895–3933, 2011www.hydrol-earth-syst-sci.net/15/3895/2011/doi:10.5194/hess-15-3895-2011© Author(s) 2011. CC Attribution 3.0 License.

Hydrology andEarth System

Sciences

Potential and limitations of using soil mapping informationto understand landscape hydrology

F. Terribile 1, A. Coppola2, G. Langella1,4, M. Martina 3, and A. Basile4

1Department of Soil, Plant, Environment and Animal Production Sciences, University of Napoli Federico II,Portici, Napoli, Italy2Department for Agricultural and Forestry Systems Management, Hydraulic Division, University of Basilicata, Potenza, Italy3Department of Earth and Geo-Environmental Sciences, University of Bologna, Bologna, Italy4Institute for Mediterranean Agricultural and Forestry Systems, National Research Council, Ercolano, Napoli, Italy

Received: 31 March 2011 – Published in Hydrol. Earth Syst. Sci. Discuss.: 17 May 2011Revised: 30 November 2011 – Accepted: 6 December 2011 – Published: 22 December 2011

Abstract. This paper addresses the following points: howcan whole soil data from normally available soil mappingdatabases (both conventional and those integrated by digitalsoil mapping procedures) be usefully employed in hydrol-ogy? Answering this question requires a detailed knowledgeof the quality and quantity of information embedded in andbehind a soil map.

To this end a description of the process of drafting soilmaps was prepared (which is included in Appendix A of thispaper). Then a detailed screening of content and availabilityof soil maps and database was performed, with the objec-tive of an analytical evaluation of the potential and the lim-itations of soil data obtained through soil surveys and soilmapping. Then we reclassified the soil features according totheir direct, indirect or low hydrologic relevance. During thisphase, we also included information regarding whether thisdata was obtained by qualitative, semi-quantitative or quan-titative methods. The analysis was performed according totwo main points of concern: (i) the hydrological interpreta-tion of the soil data and (ii) the quality of the estimate ormeasurement of the soil feature.

The interaction between pedology and hydrology pro-cesses representation was developed through the followingItalian case studies with different hydropedological inputs:(i) comparative land evaluation models, by means of an ex-haustive itinerary from simple to complex modelling appli-cations depending on soil data availability, (ii) mapping of

Correspondence to:G. Langella([email protected])

soil hydrological behaviour for irrigation management at thedistrict scale, where the main hydropedological input was theapplication of calibrated pedo-transfer functions and the Hy-drological Function Unit concept, and (iii) flood event simu-lation in an ungauged basin, with the functional aggregationof different soil units for a simplified soil pattern.

In conclusion, we show that special care is required inhandling data from soil databases if full potential is to beachieved. Further, all the case studies agree on the appropri-ate degree of complexity of the soil hydrological model tobe applied. We also emphasise that effective interaction be-tween pedology and hydrology to address landscape hydrol-ogy requires (i) greater awareness of the hydrological com-munity about the type of soil information behind a soil mapor a soil database, (ii) the development of a better quantitativeframework by the pedological community for evaluating hy-drological features, and (iii) quantitative information on soilspatial variability.

1 Introduction

1.1 Hydropedology and landscape hydrology

Soil plays a key role in hydrology since its importancein partitioning water between infiltration and runoff, stor-age, filtering, physical and chemical support to vegetation,etc. (Dunne, 1978). Soil scientists established a classifica-tion long time ago (Yaalon and Berkowicz, 1997). Sim-ilarly, there has been an increase of attention to quantita-tive taxonomy also in hydrology (McDonnell et al., 2007;

Published by Copernicus Publications on behalf of the European Geosciences Union.

http://creativecommons.org/licenses/by/3.0/

3896 F. Terribile et al.: Potential and limitations of using soil mapping information to understand landscape hydrology

Wagener et al., 2007) especially to face predictions in un-gauged basins, PUB, (http://eguleonardo2010.lippmann.lu/objectives.php). To this respect soil classification and soilmapping can also play a role in reducing uncertainty in hy-drological predictions especially when hydrological monitor-ing data are lacking (see general statements of this specialissue).

In this paper we shall refer to the term “landscape hy-drology” (also employed in other hydropedological studiessuch as Lin et al., 2006; Zhu et al., 2007) rather than theclassical watershed hydrology or basic hydrology. This isbecause many soil hydrological and hydropedological stud-ies/approaches employed in this paper deal with soils inthe landscape, which do not necessarily match to actualwatershed.

In recent years many advances have been attained by soilscientists in ameliorating soil information in both the esti-mate of hydrological parameters (e.g. by pedotransfer func-tions, PTFs, early defined by Bouma, 1989, as predictivefunctions of hydraulic properties from other more available,easily, routinely, or low costly measured properties as tex-ture, organic carbon, structure, etc.) and the spatial inferenceof soil information (e.g. McBratney et al., 2003; Grunwald,2009). In this regard, hydropedology (Bouma, 2006; Lin etal., 2006; Lin, 2011) has emerged as a new discipline devotedto the close interaction between soil science and hydrology,embracing multiscale process analysis in saturated and un-saturated soil conditions. This discipline promises to bothenhance the understanding and prediction of rainfall-runoffprocesses (Lin et al., 2008) and to be a powerful tool for en-vironmental policy research (Bouma, 2006).

Some evident examples of this potential are given in liter-ature on the relationships between soil (soil architecture) andrainfall/runoff processes. Amongst others, Lin et al. (2008)analysed the contributions of hydropedology to the under-standing and modelling of surface/subsurface runoff pro-cesses at microscopic (macropores and aggregates), meso-scopic (horizons and pedons) and macroscopic (hillslopesand catchments) scales; Bouma et al. (2008) showed the highpotential of hydropedology in addressing end user multiscaleland use problems; Bouma (1981) and Ritsema et al. (2005)related preferential flow paths to soil morphology and soilhydrophobicity, respectively; Coppola et al. (2009) studiedthe effects of a bimodal pore-size distribution and its vari-ability on a hillslope water balance. In an attempt to con-ceptualize the relationships between hydrology and pedol-ogy, Lin et al. (2008) and Lin (2010) have (i) framed hy-dropedology in the more general domain of earth’s criti-cal zone; (ii) created a hierarchical framework for bridgingsoil type distribution (forms) and soil processes (functions)in hydropedology; (iii) emphasised soil structural complex-ity at different scales (aggregates, horizon, profile, catena,etc.); and (iv) defined the Hydrologic Functional Unit (HFU)as the soil-landscape mapping unit with similar pedologicand hydrologic functions (Lin et al., 2008). Despite these

interesting conceptualizations, one basic point yet to be ad-dressed by hydropedologists concerns if/how whole soil datafrom standard soil mapping databases (which is often theonly available soil data) can be usefully employed by hydrol-ogists. To answer this question we need detailed knowledgeon the quality and quantity of information embedded in andbehind a soil map.

1.2 Soil mapping in the framework of landscapehydrology

Before approaching the use of soil database in hydrologywe indeed must report a preliminary synoptic view on soilmapping from a hydrological perspective. A more detaileddescription of this issue along with a glossary of basic soilscience terms (useful for reading this paper) is provided inAppendix A and B.

Here we must start acknowledging (Dokuchaev, 1883;Jenny, 1941) that any soil (or soil property) is formed throughthe interaction of the following environmental (or soil form-ing) factors: climate, organisms, topography, parent materialand time of soil formation (named CLORPT equation). Sub-sequently the analysis of these factors is indeed a strong helpfor understanding soils and soil behaviour.

This conceptual framework emphasises some importantpoints for hydrologists: (i) if we aim to understand soils,soil (hydrologic) properties, soil functions and soil spatialdistribution then we must adopt a robust multidisciplinaryapproach including the above environmental factors, (ii) as aconsequence it is not possible to derive soil information fromsolely geological or vegetation data, (iii) in real landscapethe spatial variability of climate, organisms, topography andparent material indeed forge the soil spatial variability. Tothis respect the above conceptual framework is very muchaligned with some new paradigms looking to the hydrologyof the future (McDonnel et al., 2007), more focused to dis-cuss on the “why” the heterogeneity exists rather than the“which” heterogeneity exists.

Unfortunately this appealing conceptual framework,where soils and soils distribution are strongly related to theabove environmental factors, has also a strong embeddedweakness since it is a rather difficult task to know/estimatethe state of these factors throughout the long time of soil for-mation at a specific landscape position.

Then it is not surprising that both conventional and digitalsoil mapping use the above soil forming factors as a very im-portant support (e.g. soil covariates) for determining the soilspatial distribution rather than parameters to be incorporatedin a mechanistic soil spatial distribution model.

A synoptic hydrological view of soil mapping must em-phasise that the production of any soil map include differ-ent activities typically performed by different experts, some-how all coordinated by pedologists: (i) describing and sam-pling soils at specific landscape positions, usually soil pro-files (typically performed by pedologists), (ii) performing

Hydrol. Earth Syst. Sci., 15, 3895–3933, 2011 www.hydrol-earth-syst-sci.net/15/3895/2011/

http://eguleonardo2010.lippmann.lu/objectives.php

http://eguleonardo2010.lippmann.lu/objectives.php

F. Terribile et al.: Potential and limitations of using soil mapping information to understand landscape hydrology 3897

chemical, physical and biological analysis on the sampledsoils (respectively performed by soil chemists, soil hydrol-ogists and soil biologists), (iii) classifying soils using inter-national systems (such as WRB, 2006; USDA, 2010) on thebase of both field and lab analysis, and (iv) producing a spa-tial distribution of either/both classified soil types and/or fieldand lab analysis into a coherent spatial framework as pro-vided by a soil map in an analog form (typically performedby conventional pedologists) or digital form (typically per-formed by digital soil map experts, geostatisticians, etc.). Ofcourse activity (iv) can indeed instruct activity (i) in locatingsoils to be described, sampled and then analysed.

The results from all these activities/expertise are then sum-marised in the final soil map and its corresponding soildatabase.

These soil maps, in the last century, have mainly reliedon qualitative approaches (Soil Survey Division Staff, 1993;McKenzie et al., 2008), based on standardized survey meth-ods, enabling to analyse and report the spatial distributionof soils (see Appendix A). These maps have been, and theyare still, largely employed as indispensable tools for planningproper land management throughout the world (e.g. Stolpe etal., 1998; Herrero et al., 2007).

Nowadays, some limitations and drawbacks of these con-ventional soil maps have become apparent and as a resultmuch advancement have been produced in mapping soils us-ing more quantitative approaches. More specifically Digi-tal Soil Mapping (DSM) has emerged as a credible alterna-tive to traditional soil mapping. It entails the use of newtools and techniques coming from different branches of abroader scientific community (e.g. spatial statistics, GIS, re-mote and proximal sensing, computer programming) in or-der to put into a quantitative framework the spatiotemporalstudy of soils (McKenkie and Ryan, 1999; McBratney et al.,2003). DSM overcomes some serious limitations of conven-tional soil mapping such as the description of soil variability.DSM emphasises the soil continuum, where soil propertiesat a given location depend also on their geographic positionand on the soil properties at neighbouring locations, and thenovercome limitations and coarseness of using large discretepolygons as a means of describing soil variability in the land-scape in both the geographic and the attribute domains.

One of the most interesting outputs of such DSM methodsmay be the spatial distribution of the prediction error (soiltype and soil attributes); this can be usefully employed inapplying hydrological modelling at the landscape scale.

Despite all the important advances by DSM, unfortunatelyit is rather evident (e.g. Jones et al., 2005) that, at the presentday, in most countries, regions, municipalities and so forth,classical soil maps still constitute the only real soil data avail-able – and usable – for landscape and watershed hydrology.This is not surprising considering that DSM is still a youngapproach in soil mapping and typically requires soil datasetswith associated higher costs (Manna et al., 2009) that must be

counterbalanced by higher earnings to be sustainable (Verha-gen et al., 1995).

Analysing the relationships between soil maps and hydrol-ogy it seems clear that there has been some progress. Forinstance in Table 1 are given references to few digital soilmapping works providing spatial analysis of soil parametersrelevant for hydrology. But despite the progress made on thesubject, the scientific literature (e.g. McBratney et al., 2003;Grunwald, 2009) is rather devoid of critical and analyticalevaluation concerning the use of soil map information (bothtraditional and/or obtained by DSM) aiming specifically tocontribute in watershed related hydrological studies.

The lack of scientific literature on this crucial question israther regrettable and indeed surprising and it is possibly re-lated to the evidence (by surveying the scientific literatureand from the few references quoted above) that at present hy-dropedology has been very much driven by soil hydrologistsrather than pedologists, who are indeed the experts producingsoil maps. In this regard, our contribution, which emphasisespedological issues, it may both provide a new outline for hy-dropedology and a new framework for approaching hydro-logical problems at the landscape scale.

In this context, going beyond the generic statement con-cerning the importance of hydropedology, we believe it is ofthe utmost importance to examine whether and to what extentsoil maps (and associated soil data), produced in accordancewith different aims, scales and procedures (conventional orby DSM), can play a role in hydrological applications at thelandscape scale. This work aims to address the above ques-tion, focusing on the hydrological potential and limitationsof soil surveys and soil mapping.

In this paper, having provided in Appendix A the descrip-tion of the process of making soil maps, we perform an an-alytical evaluation, from a hydrological perspective, of thedatabase associated to soil maps. We then show ways to re-inforce and to rethink the interaction between pedology andlandscape hydrology also exploring alternative strategies thatcan be followed according to the soil data availability, usingsome relevant case studies from Italy. For the sake of intelli-gibility we provided in Appendix C a short description of themodels applied in the case studies.

2 An evaluation of soil mapping data from anhydrological perspective

2.1 Content and availability of soil maps and soildatabase

The final result of soil surveying and soil mapping is thenthe production of a georeferenced soil database containingall the information obtained from field work and laboratoryanalysis, along with GIS vector data containing the geometryof the soil mapping unit polygons (see Appendix A). Hithertothe same information was produced in analog format.

www.hydrol-earth-syst-sci.net/15/3895/2011/ Hydrol. Earth Syst. Sci., 15, 3895–3933, 2011


Table 1. Selection of works involved in soil mapping and hydrologically relevant parameters.

Author (year) Attributes No Cartographic Study Areaobservations scale (1:x) (km2)

McKenzie and Austin (1993) clay content, CEC, EC, pH, 224 100 000 500bulk density and COLE

Zheng et al. (1996) available water capacity – – –

Cialella et al. (1997) drainage classes – 12 000 24

Voltz et al. (1997) wiliting point 426 100 000 17, 36

Obertur et al. (1999) soil texture classes 384, 208 100 000 192, 39

Chaplot et al. (2000) redoximorphic features, 182 – 0.2hydromorphy index

Lagacherie and Voltz (2000) wilting point 374 100 000 20

McBratney et al. (2000) clay content, CEC 95 2000 0.42180 200 000 1100734 500 000 45 600

Campling et al. (2002) soil drainage classes 295 + 72 50 000 589

Kravchenko et al. (2002) electrical conductivity, 107 – 0.2soil drainage

Jost et al. (2005) water storage 195 0.005

Agyare et al. (2007) texture, pH, OC, CEC, 600, 400 – 6, 0.64bulk density, saturatedhydraulic conductivity

Shrestha et al. (2008) water-holding capacity, 165 – 3550saturated hydraulicconductivity, hydrologicallyactive layer depth, soiltexture, rainfall

Joel et al. (2009) texture, groundwater – 50 000 to 7260level, occurrence of poorly 100 000permeable layers

Basically, in conventional mapping the final soil map andits corresponding database include (i)n soil mapping units;(ii) one (or more) main soil type for each of thesenmappingunits (typically named soil typological units); (iii) then ateach soil type are assigned field and lab data obtained froma representative soil. The choice of this representative soil(also named benchmark soil) is rather important and it isbased on expert (pedologist) judgement aiming to identifywhich of then soils of a specific type is both dominant andbest represents the modal concept. This soil can be either atrue soil (e.g. Soil Survey Division Staff, 1993) or a virtualsoil obtained after a weighted mean procedure (e.g. Lambertet al., 2003). Finally, (iv) some interpretations for practicalpurposes are also provided.

In the case of digital soil mapping, such final soil map(produced after a spatial inference system) more often re-fer to the mapping of specific soil properties but the starting“soil data” indeed include the same soil database of conven-tional mapping – but can also include (i) and (ii). Then both

conventional and digital soil mapping rely on the same basicsoil database (also in terms of potentials and limitations).

Here it is important to emphasise that, typically, thedatabase behind a soil map is rather ponderous. For instanceassuming a soil observation (profile) consisting of only threehorizons, the potential output for a standard field descrip-tion may include about 181 field data (31 for site descrip-tion; 50 characteristics to be described for each of the threesoil horizon) and 36 laboratory-based data for “each soilobservation”.

This ponderous spatial database can be very important forhydrologists because it can be easily assumed that most fea-tures are directly or indirectly connected to either/both waterfluxes and permanence of water in soils. An example of thekind of soil features that are described and then present insoil databases is given in Table 2 – later discussed – wherewe have highlighted in bold and italics those features that areof great relevance to hydrology.



Table 2. Example of soil features typically found in a pedological description. Numbers indicate the amount of sub-characters supplyingeach feature.

Field analysis at the Field analysis of Laboratory analysis Process for determining water andsoil survey station the soil profile of the soil profile temperature regime in soil taxonomy(31 features) (73 features (12 features (monthly data from weather stations

per horizon) per horizon) nearby)

Location Type of horizon pH (H2O) RainfallType of observation Matrix colour 4 pH (KCI) TemperatureSoil surveyors Concentrations 5 CEC ETp

Lithology Coatings 6 Exchangeable basesLand use Slickensides 2 CarbonatesVegetation Biological activity 2 TotalNMorphology Carbonates Electrical conductivityCurvature Roots 3 Organic carbonErosion 2 Mottles 4 GranulometryDeposition Coarse fragmentsRotting depth >2 mm) 5Depth to rock Texture 3Parent material 3 Consistency3Elevation Structure 6Slope Pores2Exposure Cracks 2Vegetation cover Boundaries2RockinessStoniness2RunoffCrack 3GroundwaterFloodInternal drainagePermeabilityEstimated AWC

Normal: soil features of low hydrological interest; italics: soil features of indirect hydrological interest; bold: soil features of direct interest in hydrology; underlined: quantitative

evaluation; not underlined: qualitative and semi quantitative evaluation.

Examples of such soil databases and vector data themesmay be found on the web (e.g. in the USA). Soil maps can beeasily accessed through services such as web soil survey pro-grams at national scale (e.g. in the USAhttp://websoilsurvey.nrcs.usda.gov) or also at regional level (e.g.https://applicazioni.regione.emilia-romagna.it/cartografiasgss).

The availability of soil maps (and associated databases)varies from country to country and their quality is often re-lated to the presence of soil survey agencies. It has beenestimated (Dobos et al., 2006) that over 500 000 detailed soilprofiles have been described in EU countries in the last 20–30 years. In the EU despite this potential, unfortunatelymany national institutions (which typically commissionedsoil maps) are unwilling to reveal soil data; they only pro-vide processed generalized products (Rossiter, 2004). Bycontrast, in the USA soil datasets are easily accessible (http://www.nrcs.usda.gov). The availability of soil maps obtainedthrough DSM procedures and with a hydrological content is

still rather limited (Table 1) but it embeds high potential infuture prospects.

Despite this complex and heterogeneous scenario, alreadyreported in Dobos et al. (2006) and Jones et al. (2005), manyEuropean regions are gradually moving to adopt standard-ization such as the case concerning the scale of conventionalsoil maps. At present, the latter are generally produced at theinventory (1:250 000) and semidetailed (1:50 000) scales.

The 1:250 000 scale aims at an inventory of regional soilresources. Most European countries either already have suchmaps or are in the process of obtaining them. The 1:50 000semidetailed scale, because of their better spatial informa-tion (higher number of soil observation per unit area), aimsto produce soil information directly usable in planning andland management (mainly in agriculture and forestry) butalso possibly in landscape hydrology. Availability of thesesemi-detailed soil maps varies greatly since there is a markeddiscrepancy in map coverage from several European regions


http://websoilsurvey.nrcs.usda.gov

http://websoilsurvey.nrcs.usda.gov

https://applicazioni.regione.emilia-romagna.it/cartografia_sgss

https://applicazioni.regione.emilia-romagna.it/cartografia_sgss

http://www.nrcs.usda.gov

http://www.nrcs.usda.gov


and most of the mapped areas refer to plains. This situa-tion, to be regretted from a hydrological viewpoint (plainsare just one component of a catchment), is due to the fact thatpublic agriculture departments, aiming at better agriculturemanagement, are the main financiers of such maps. At thisscale in many countries, hill and mountain areas are rarelyincluded in such maps. In this paper we do not aim to gofurther into the subject of soil mapping. However, soil mapcontent (e.g. data from specific soil mapping units) can beeasily browsed for specific areas by searching through themany dedicated websites (such as those given above).

From this scenario, regardless the scale, it is important toemphasise that soil mapping, being heavily financed by agri-cultural departments and organizations, is strongly focusedon providing answers to agricultural questions (e.g. chemi-cal and physical fertility, C stock, soil productivity) and thenhydrological answers must be found from this basis, whichis anyway a very valuable basis.

Moreover, standard soil mapping does not provide truequantitative information as to ensure detailed soil spatialvariability and related spatial uncertainty. In the scientificliterature there are a few cases where this information hasbeen produced ex post but in most localities this spatial in-formation cannot be retrieved mainly due to (i) loss of dataand (ii) customers (e.g. administrative regions) who commis-sioned the soil survey (in many cases long ago) only rarelyallow access to the original soil information in their posses-sion (if they still have it). In fact standard publication ofsoil maps, soil reports (both in analog and/or digital format)and even soil database do not include all the produced soilinformation but rather an extensive summary based on themain soil types (reference profiles) and the related landscapefeatures. Although a fairly standardized process producesthis summary (e.g. Soil Survey Division Staff, 1993) wheremuch care is taken to ensure internal consistency, it does notprovide a tool to investigate the spatial variability of soils be-tween and especially within the soil mapping units.

On this base, it is important to emphasise that in the fol-lowing sections we shall focus our hydrological evaluationto the only information contained in soil databases. Morespecifically, we shall not attempt any hydrological orientedexamination of other important soil documents such as soilmap legends/reports (as typically reported in shape files or inanalog maps) or soil classification classes. This is becausesoil map legends and reports provide a synopsis of soil infor-mation where parameters which originally may be importantfor hydrology are not anymore available being embedded andaggregated along with other information in the process of soilmapping. This situation also applies to soil classification.In fact pedologists typically classify soils, indeed using up-dated international soil classification schemes (e.g. Soil Tax-onomy: USDA, 2010; World Reference Base: WRB, 2006)that are still very much developing in the framework of agri-culture/forestry (e.g. FAO, USDA) rather than in hydrology.Then hydrological features (e.g. field estimate of infiltration

and runoff, water retention at specific pressure heads, etc.),if present, in the best of the cases have been employed asone of the many parameters enabling the soil classification.Then, in some cases, the only final soil classification maybe far from giving the indication of the soil hydrological be-haviour. The worse outcoming result is that soils classifiedwith the same name may have a rather different hydrologicalbehaviour. Then, from our viewpoint, it is better to focus ouranalysis on actual database rather than on soil classificationas given in standard soil legend/report.

2.2 An analytical evaluation of potentials andlimitations in using soil mapping data in landscapehydrology

Hydrological analysis of the soil map and its database mustreally start from its information content. Then in order toshed some hydrological light on these soil features occur-ring in typical soil databases, we reclassified them in ac-cordance with their direct, indirect and low hydrologic rel-evance. More specifically, in Table 2 we have highlighted inbold (direct) and italics (indirect) those features contained inthe soil mapping databases that are of great relevance to hy-drology. This classification has the form of a general sugges-tion, mainly because in some cases the boundaries betweenthese classes are not sharp.

In theory, any of these soil features, also those classified aslow hydrologic relevance, have a potential (direct/indirect)interest in hydrology; for instance soil pH, governing thecomposition of the soil solution and influencing ion exchangeon clay minerals, can greatly affect aggregation and hencesoil porosity and eventually hydrological behaviour. How-ever, soil pH, even if strongly influencing the soil system andthus potentially the hydraulic properties, has no direct quan-titative relationship with them. This is because pH is onlyone of the many soil features (and processes) occurring atdifferent spatial scales and governing soil porosity and thenin turn soil hydraulic properties, infiltration rate, runoff, etc.Then features, such as soil pH, have been considered havinglow hydrologic relevance.

An example of feature having an indirect hydrologic rele-vance is for instance the occurrence of iron coatings; in factin many soils they can be assumed as an indirect indicator ofsevere water stagnation (even if water may be not present atthe time of soil observation).

Among the data having a direct hydrologic relevance andalso directly applicable in hydrological modelling we just re-call the texture and organic carbon for an easy estimate ofhydraulic properties by PTF (Pachesky and Rawls, 2004),the horizons boundaries for the schematization of the waterfield flow, and the depth of the water table for fixing bottomboundary condition.

Given the complexity of the issue and the need to go intothe hydrological usability of data presented in soil databases,we limited our analysis at the only features (19) that have



Table 3a.Types of data of great (and direct) hydrological interest contained in soil map databases.

Type of soil Description of the feature Method Methodological description Main potential Main limitationsfeature (N. ofsub-feature

Cracks (3) Occurence of cracks at the SQT3 Estimate (using comparative tables and metric It is evidence of It is a strongly anisotropic parameter, non-soil surface measurements) of frequency, width and depth of processes of preferential linear function of the water content in soil

cracks water and pollutantflows (bypass flow)

Groundwater Occurence of groundwater QL2 Assessment made both on the basis of direct Basic environmental Strongly quality-based and timein the investigated soil observations in the soil profile along with indirect data dependent assessmentprofile information (interviews with farmers, land

reclamation consortia, etc.)

Floods Flood risk QL2 Estimate of morphometric, morphodynamic and Basic environmental Rather subjective evaluationhydraulic factors governing flood risk data

Runoff Runoff estimate QL2 The class of runoff (from very low to very high) Basic environmental It is a subjective measurement becauseKsatis established using a table. Slope angle andKsat data is rarely known. Hence assessment is made(or its estimate) must be known. by “expert best estimate”

Internal Estimate of water removal QL2 Assessment on the basis of slope, texture It is a feature governing Strongly quality-based assessment on adrainage rate in the soil profile skeleton, presence of horizons with low infiltration and runoff parameter which is very difficult to

permeability and also hydromorphic horizons process estimate

Estimated Estimate of Available Water QL2 Assessment based on texture, organic matter, It is a feature governing Many of the parameters required for thisAWC Capacity for vegetation bulk density, rock fragments, salinity, roots and infiltration and runoff evaluation are highly subjective and

soil horizon depth. Assessment is made by using processes qualitivetabular data and empirical formulas.

Boundaries (2) Lower limit of horizons. It is QT1 Metric measurements (cm) Essential basic This parameter can have marked spatialthe thickness of each horizon information variability

Mottles (4) Patches of different colours SQT3 It describes the colour (Munsell Tables), This information enables The size of the mottles depends also on(usually related to Fe and/or frequency (visual estimate using comparative the assessment of the the soil chamical conditions (e.g. pH) andMn), on the surface of the tables), size (metric measurement), contrast, (relative) degree of water on the behaviour of the different ionicaggregates produced by limits and location of the mottles. stagnation (even if water forms of iron and manganese. The mottleswaterlogging is absent at the time of may have been produced in a climate

soil description) different from the present day (e.g.paleoclimate)

Abbrev: 1 QT: quantitative;2 QL: qualitative;3 SQT: semiquantitative.

a great hydrologic relevance. Hence, in Table 3 their mainpotential and limitations have been reported along with theinformation if the method by which this data is obtained wasqualitative, semiquantitative or quantitative.

A detailed analysis of the table shows that at any fea-ture corresponds a hydrological potential; then this potentialcould become even larger considering that many soil featuresoccur in soils at the same time and then their ensemble canprovide integrated hydrological information. But in order toexploit this potential it is also fundamental to understand as-sociated limitations.

Then on this base below we discuss, using some exam-ples given in the table, the two main types of limitation thatshould be taken into account using these data, namely datainterpretation and quality of the estimate of soil features.

a. Data interpretationSoils are complex bodies and some features need an ex-pert interpretation. In order to show this point, one soilfeature worth mentioning, because having a close inter-action with soil hydrology, is the occurrence of coat-ings in soils. Those are soil material typically defloccu-lated, transported and deposited (flocculated) in lowerhorizons. Water is a crucial factor affecting the occur-rence of these soil features and then we would expect

that their occurrence and quantification could help us inunderstanding underlying soil hydrological processes.Unfortunately the issue is far more complex, in fact wemust consider the following: (i) coatings can be of dif-ferent type such as clay, silt, organic matter etc.; thenthe physics governing each of these type of materials isdifferent and it is very indeed much affect by the parti-cle size. For instance, the usually called clay coatingsare typically fine clay coatings (<0.2 microns) becausemainly fine clay is likely to easily pass through the fil-tering action of soils. Then this fine clay can moveboth in very small and very large pores (where pref-erential flow occurs) and then the hydrological signif-icance of coatings must be related to this complexity.The next issue (ii) is that coatings can move both ver-tically and horizontally and therefore their hydrologicalmeaning is not unique. Moreover (iii) the field quantifi-cation of these features is not always easy and its de-tecting (even with magnifying lenses) depends by thedegree of colour contrasts between the coatings them-selves and the soil matrix; finally, (iv) coatings can beformed (pedorelict) under different climatic condition(e.g. paleoclimate).



Table 3b.Types of data of great (and direct) hydrological interest contained in soil map databases.

Type of soil Description of the feature Method Methodological description Main potential Main limitationsfeature (No. ofsub-features tobe described)

Coarse Estimate of soil particles SQT3 Describes size (metric measurement), shape It is essential Visual estimate by comparative tablesfragments (5) larger than 2 mm (in the (using reference diagrams), lithology and information for

field) frequencey (visual estmate using comparative calculating watertables) of particles>2 mm balance in soil

Texture (3) Estimate of soil particles SQT3 Assessment, using standardized tactile tests Essential parameter to Evaluation of % data by tactile testsmaller than 2 mm (in the schemes (e.g. USDA), of textural class and/or % estimate many requires much experience an alsofield) estimate of sand, silt and clay hydrological calibration on the specific soils under

parameters investigation

Structure (3) Analysis of soil aggregates QL2 Description of type (comparison with diagrams), It is a feature strongly It is a feature described by a stronglysize (metrical analysis) and degree of destinctness governing the qualitative assessmentof the aggregates in soils dynamics of water and

pollutants in soil

Pores (3) Estimate of soil SQT3 Description of size, frequency and shape It is a feature, It is a rather difficult feature to bemacroporosity (pores (comparative tables) of macropores, using a connected to structure, determined. It is appraised by a strongly>0.1 mm) magnifying glass (10 X) which strongly quality-based assessment.

affects water andpollutant dynamics

Internal cracks Occurence of cracks within SQT3 Frequency estimate (comparative tables) of both Sign of potential Parameter strongly anisotropic with non-(2) a soil horizon width an depth of cracks (metric measurements) bypass flow processes linear function of the soil water content

in soil horizons

Consistence Soil features related to QL2 Description of consistence and plasticity of soil It is another feature It is a feature determined by a strongly(5) cohesion and adhesion aggregates by means of their resistance to hand that, by influencing quality-based assessment

breakage, type of breakage, degree of soil structure, can alsocementation, adhesiveness and plasticity. affect soil hydrologic

behaviour

Andic Occurence of QL2 Field test by pH estimate (by phenolphthalein It is a feature strongly It is a property determined by a stronglyshort-range order indicator) after sodium fluoride addition. Colour affecting hydrological quality-based assessmentclay minerals and/or intensity of the reaction is correlated with andic propertiesAl/Fe-humus complexes properties


Nevertheless, despite these cautions, indeed coatings, ifwell recognized and defined, can tell us something im-portant about which type of hydrological process take(took) place in the soils (e.g. preferential flow) and thentheir occurrence and their interpretation (between op-tions i, ii, iii and iv) can be much profitable from anhydrological viewpoint.

Another case worth mentioning concerns mottles. Theiroccurrence (frequency, size, location) is a very impor-tant index for assessing water saturation patterns, evenif water is not present at the specific time of soil survey.This is certainly true, but again some caution must betaken if these mottles refer to iron or manganese: thesetwo elements have rather different solubility at differentpH and redox potentials. For instance, if two soils showvery different pH values, then the same mottles may in-dicate rather different hydrological conditions. In addi-tion, mottles can occur as the result of ancient water sat-uration processes, such as those occurring for instancein many Italian palaeosols in the Po Valley. An expertpedologist knowing the study area can easily make thesecautionary interpretations of mottles.

A special case to be mentioned refers to the occur-rence of andic properties (also named here as andicfeatures). In this case much caution should also beused in applying PTF in soils showing marked andicfeatures (see Appendix B) for the occurrence of spe-cific and highly reactive clay minerals, namely loworder clay minerals (allophane-like). These specificfeatures lend these soils with distinctive physical andchemical characteristics that are not found in soils de-rived from other parent materials under the same veg-etation and climate. Particularly, they are well struc-tured (at the surface) showing both high specific sur-face (600 m2 g−1) and large volume of micropores andmacropores (Yong and Warkentin, 1975). Andic fea-tures are then responsible for the high values of waterretention and water conductivity but also high CationExchange Capacity (CEC); they all strongly affect waterand solute fluxes. As consequence (i) in these soils stan-dard PTF cannot be applied because the standard rela-tionships texture-hydraulic properties have been all setwith respect to soils having crystalline clays (Basile etal., 1999); moreover, (ii) due to this distinctive features,there are well known artefacts (Nanzyo et al., 1993) in



Table 3c.Types of data of great (direct and indirect) hydrological interest contained in soil map databases.

Type of soil feature Description of Method Methodological description Main potential Main limitations(No. of sub-features the featureto be described)

Clay coatings (6) Coatings of clay SQT3 It describe the colour (Munsell Tables), frequency (visual It is an evidence of Occurence clay coatings depends also bymaterial (usually estimate using comparative tables) size (metric processes of preferential soil chemisty (e.g. ions in soil solution,fine clay) as measurement) and location of clay coatings water and pollutant flows exchangeable sodium). Coatings may haveoccuring in been produced in a past climate differentpores and over from the presend day (e.g. paleoclimate)aggregates

Iron (Fe) and Fe and Mn SQT3 It describe the colour (Munsell Tables), frequency (visual It is an evidence of Occurence of Fe-Mn coatings depends alsomanganese (MN) coatings are estimate using comparative tables), size (metric reducing condition as by soil chemistry and on the diffent ioniccoatings (6) occuring in measurement) and location of Fe-Mn coatings occuring in water forms of iron and manganese in soil

pores and over stagnation. Fe coatings solution. Fe-Mn coatings may have beenaggregates more pronounced water produced in a past climate different from

stagnation as compared to the present day (e.g. paleoclimate)Mn coatings.

Granulometry Laboratory QT1 Analysis of frequency of coarse fragments, coarse sand, fine Basic information for Soils with similar granulomatry can stillanalysis of the sand, silt and clay (e.g. pipette method or hydrometer many hydrological have very different hydrological behaviourparticel size method). evaluations. It is a very This is especially the case for permeability,classes (at least Since soil particles can be agrregated by organic or inorganic robust parameter which depends on meso and macropores,sand, silt and cementing agents, real granulometric analysis (after governing many physical but also for water retention properties,clay dissolving all cements) or apparent granulometric analysis process which can change according to clay

(no pretreatments) can be performed. mineralogy (kaolinite versus smectite ratio)

Organic carbon Laboratory QT1 % of organic C (e.g. typically performed after dichromate It is a soil feature that Soils with the same organic carbon contentanalysis of oxidation method) strongly affects soil may have very different physical propertiesorganic C structure, porosity andcontent hence many physical

processes

Water Regime Simplified water QT1 This analysis is generelly performed to classify soils It is a rather synthetic but The assessment of the “water regime” isbalance according to the Soil Taxonomy (USDA) scheme. Examples very useful assessment. It very much affected by the coarse quality of

include Xeric, Ustic, Udic, Aquic moisture regime. This is strongly associated to inputs (i.e. monthly time based inputs). Thisanalysis consists in a simplified water balance on the basis of the physical reality of is the case for AWC, rainfall andmonthly rainfall and evapotranspiration data and of soil soils evapotranspiration. It is a description of aAWC. Evaluation is made using bucket-based models (e.g. rather “static” water balance.Billeaux, Newhall) for a specific soil depth (control section)


standard texture analysis due to the dispersion proce-dure engendering an apparent coarser texture than thereal one. Finally, (iii) solute transport in andic soilscan be characterized by the occurrence of both chem-ical and physical exclusion. The presence of large neg-ative charges on some surface particles results in thechemical repulsion of anions from these regions (Bolt,1982); moreover due to the occurrence of very high av-erage pore water velocity, a portion of the water in themicropores can be considered relatively immobile andtherefore excluded by the flow. Such unusual but crucialcharacteristics of these soils are difficult to be estimatedby indirect methods; hence typically they should be de-termined either experimentally or by carefully studyingthe limited available literature (Bartoli et al., 2007).

Then, as final remarks, all the above examples proveboth that some soil features present in soil database canbe potentially profitable for hydrologist but also that thehelp of an expert (hydro)pedologist is a must for a cor-rect interpretation of those features occurring in a spe-cific piece of landscape.

This is even more important considering that many soilfeatures (e.g. clay coatings and mottles) can occur atthe same time in soils and then their corrected com-pound interpretation (e.g. preferential flow and seasonal

water stagnation) can profit very much from this si-multaneous occurrence. Finally the simultaneous oc-currences of features (e.g. mottles, Fe coatings) havinga similar hydrological meaning (e.g. water stagnation)enable to have a further cross validation for a correctinterpretation.

b. Quality of the estimate/measurement of soil featuresIn soil databases almost all parameters have a sort ofquantitative formalization. In reality, the methods bywhich this information is obtained may be qualitative,semiquantitative or quantitative, as illustrated in Ta-bles 2 and 3 and these methodological differences areimportant if this information have to be effectively em-ployed in hydrology (e.g. parameterization of hydrolog-ical models).

For example, the analysis of water balance, usingbucket-based models, might induce one to assume that asoil database provides high quality data for hydrologicalapplications. Unfortunately this is not always the casebecause, for example, the AWC (Available Water Ca-pacity, the reference water storage in the rhizosphere) iscalculated on the basis of particle size classes by meansof a not calibrated PTF and not through direct measure-ment; furthermore, the particle size classes themselves



can be obtained by a laboratory measurement or evenmore by a field estimate (Table 3a). Nevertheless know-ing these cautionary notes, indeed simplified water bal-ance as occur in reports and even in some soil type clas-sifications (Soil Taxonomy) can be very informative anduseful in hydrology.

In Table 3 are also reported the generic potentiality inusing each reported feature. They range from (i) basicenvironmental information obtained by features such asflood, internal drainage, groundwater; (ii) direct data forinferring hydrological parameters (e.g. hydraulic prop-erties parameters from organic carbon, texture and/orgranulometry; modelling dissolved organic carbon con-centrations and fluxes in catchments and streams fromorganic carbon content); (iii) occurrence of distinct wa-ter flow processes (e.g. preferential and/or bypass flowin presence of cracks; wet regime for the presence of ashallow water flow impeding horizon).

The central point is how such potentiality can actu-ally be transferred in hydrological modelling. Whilethe information of the point (i) are of general inter-est and those referring to the point (ii) easily and di-rectly applicable, those referring to the point (iii) suchas for instance the occurrence of iron mottles (indicatingseasonal water stagnation) are not input parameters inany existing physically based hydrological models andtherefore need a further explanation to better appraisalthe potentiality embedded in soil map and database. InTable 4 we reported how some soil features can be fruit-fully employed for conditioning hydrological models.In other words, we reported how distinguish some dis-tinct water flow processes occurring in the soil with-out make specific hydrological test (e.g. infiltration, testwith tracers). In this sense we act at the early stage ofthe modelling procedure in the definition of which pro-cesses (i.e. presence/absence of preferential flow) can(should) be simulated. Then we have defined this step as“conditioning of hydrological modelling” and the list ofsoil features to be used has been compiled considering alarge bulk of literature attesting the relationship betweenthese features and hydrology. Between these featureswe included living roots (Aubertin, 1971; Warner andYong, 1991), macroporosity (Lin et al., 1999), mottles(Rabenhorst et al., 1998; Bouma et al., 1990), Fe andMn concretions (Stoops and Eswaran, 1985; Hseu andChen, 1996), colour of the matrix (Vepraskas, 2004),Fe and Mn coatings (Linbdo et al., 2010), clay coatings(Kuhn et al., 2010), carbonates coatings (Durand et al.,2010), gypsum coatings (Poch et al., 2010).

For example, the presence of the Fe concretions comesfrom alternating wet and dry condition. If they showa diffuse outline this means that a strong wet periodmay have occurred and therefore a fluctuating water ta-ble could be imposed as bottom boundary condition;

other way, if they show an abrupt outline this meansthat a strong dry period has likely occurred and there-fore the unit gradient can be imposed as bottom bound-ary condition.

Another case worth mentioning is the diffuse presencein the soil profile of coatings of manganese; they appearin soils having generally high soil moisture. In this case,the influence of the pressure head on the hydraulic gra-dient is reduced and therefore a simplified gravitationalflow can be cautionary assumed.

The last example, taken from the Table 4a, concerns bi-ological activity (observed in the field) and occurrenceof slickensides; these features are completely differentin their nature and genesis but both can produce macro-porosity, potentially inducing preferential flow paths. Inthis situation a model at two pores domains (e.g. com-posite porosity approach, the double permeability ap-proach) with or without an exchange term between thetwo domains should be applied. Furthermore, if themacroporosity produced by these two pedological fea-tures is anisotropic, a 2-D or 3-D domain of the flowfield should be applied even in flat area where the lat-eral component of the gradient is generally discarded.

For all the above cited examples and for many othersshowed in Table 4a we have to remark that the given hy-dropedological indications are not real fluxes measure-ments. These indications can be used but they shouldtake with caution crossing them with any other evi-dence, both pedological and hydrological, of occurrenceof the same process; or they should be used to program-ming field campaign of fluxes measurements.

But despite these cautions, when using hydrologicalmodels in ungauged landscapes (such as in PUB), thesefeatures can indeed represent a great help in modelsconditioning.

A conclusive evaluation on the features given in Ta-bles 2, 3 and 4 is required and it must emphasise theneed for new methods/methodologies aiming to betterquantify features described in the field. This need is notnew and it is very much under development (e.g. Rosselet al., 2010) in the area of geophysics, geochemistryand spectroscopic analysis but indeed very little is avail-able on soil hydrology. We believe that this area in-deed deserves much more emphasis in soil science andit may enable a better interaction between soil scienceand hydrology.

3 An application of pedological information tohydrological forecasting

In the following we show some examples of interaction be-tween pedology and hydrology toward hydrological fore-casting. Unfortunately we do not have a single case study



Table 4a.Indirectly usable soil features commonly occurring in databases and exploitable in hydrological modelling.

Soil features Simplified hydrological meaning of the soil features Examples of modelling conditioning (1-D, 2-D, 3-D)

Mottles Alternating wet and dry condition with a strongly wet period Bottom boundary condition: setting high fluctuating water table depth

Fe-Mn concretions Alternating wet and dry condition with a strongly dry period Bottom boundary condition: setting free drainage (hydraulic head(abrupt outline) gradient =−1)

Fe-Mn concretions Alternating wet and dry condition with a strongly wet period Bottom boundary condition: setting fluctuating water table depth(diffuse outline)

Greysh colour of Strongly redox condition induced by water stagnation Bottom boundary condition: setting low water table depththe soil matrix(not lithochromic)

Clay coatings Abundant water fluxes enable to be moved in macropores Use of preferential flow approaches (e.g. double permeability,composite porosity, etc.). Occurrence of strong 2-D/3-D flow field(not slope-induced)

Fe coatings Strongly redox condition induced by water stagnations Bottom boundary condition: setting low water table depth

Mn coatings High moisture in soils Simplified flow field during wet season (e.g. gravitational flow)

CaCO3 coatings Alternating wet and dry condition with a strongly dry period Bottom boundary condition: setting free drainage (hydraulic headStabilization of soil pores gradient =−1)

Gypsum coatings Very dry soil environment Bottom boundary condition: setting free drainage (hydraulic headgradient =−1)

High biological Presence of macropores and potential occurrence of preferential Use of preferential flow approaches (e.g. double permeability,activity; high flow paths composite porosity, etc.). Occurrence of strong 2-D/3-D flow fieldfrequency of (not slope-induced)macropores andliving roots

Slickensides Strongly alternating wet and dry condition inducing preferential Use of preferential flow approaches (e.g. double permeability approach,flow paths composite porosity approach, etc.). Occurrence of strong 2-D/3-D flow

field (not slope-induced)

High CEC Filtering ability towards xenobiotics (especially if cationic) Setting of parameters for solute transport (e.g. retardation factor>1 inconvective-dispersive equation)

High andic Filtering ability towards xenobiotics Setting of parameters for solute transport (e.g. retardation factor>1 inproperties High water retention and hydraulic conductivity convective-dispersive equation).

Not applicability of standard PTF

Table 4b. Directly usable soil features commonly occurring indatabases and exploitable in hydrological modelling.

Soil features

Horizon depthGranulometryOrganic CarbonCoarse fragmentsCracksElectrical conductivity

including all the most important alternatives of interactionbut case study 1, which is devoted to a land evaluation ex-ercise for maize production in a typical agro-ecosystem, itis an excellent methodological example to show the overallpotential of applying hydropedology even in hydrology. Weinserted other two case studies dealing with mapping of hy-drological behaviour for irrigation planning and managementat the district scale (case study 2) and flood event simulationin an ungauged basin (case study 3). The last case study wasdeveloped applying some novel data.

For the sake of paper readability we have reported in Ap-pendix C some further details of the three case studies in-cluding a detailed description of the applied modelling.

Hereafter, after providing a short description of the casestudies, we discuss their relevance for hydrology. This willalso be performed following the synoptic scheme reported inTable 5. Here in the first column we report, in an increasingorder, the crucial issue concerning which “soil map databaseis available” for each given case study (including sub-casestudies). Then, sequentially, in the other columns we show(i) the questions we have been asked to answer, then fromthat basis (ii) what we have performed, (iii) the appliedmodel, (iv) how and which hydropedological informationhave been employed to condition hydrological modelling,(v) what we have obtained, (vi) in general terms what welearn/get for hydrologic applications and finally and most im-portantly (vii) performance and costs of the given methods.



Table5a.S

ynopticschem

eofthe

hydropedologicalissuesin

the3

casestudies.

Soilm

apdatabase

availableQ

uestionsto

Case

Whatw

ehave

Applied

Hydropedological

Whatw

ehave

Whatw

egetfor

hydrologicapplications

Perform

ance6

Cost 7

(rankreflects

increasingbe

answered

study4

performed

model

information

conditioningobtained

information) 1

hydrologicalmodelling

Schem

eA

:Land

1.1A

pplyingsim

pleA

LMA

GR

AN

.A.

5Q

ualitativeestim

ationof

Sim

pleand

lowcostqualitative

evaluationofsom

e∗

∗

Only

representativesoilprofiles

evaluationm

ultiparametric

landattitude

(simple

andhydrologic

featuresor

processesatlandscape

scale ∗∗

data 2applied

to2.1

scheme

US

BR

Landlow

cost)(e.g.potentialrunoffoccurrence)

(∗)

(∗)

cropand

(FAO

-likeLE

)S

uitabilityirrigation

1.2A

LMA

GR

AN

oneS

imple

andlow

costqualitativeevaluation

ofsome

∗∗

suitability+

SW

AP

hydrologicfeatures

orprocesses

atlandscapescale

(hydrological(e.g.potentialrunoffoccurrence)

basedapproaches)

1.4S

imulations

CropS

yst–

Pedologicalevidence

of1-DQ

uantitativeestim

ationLow

costquantitativeevaluation

ofsome

hydrologic ∗∗∗

modelling

offlow

(absenceofhorizontal

ofpotentialyield(low


scale(e.g.daily

water

storage,w

aterbalance

preferentialflows)

cost)actualtranspiration,drainage

fluxes,etc.)and

cropyield

–N

oim

pedinglayer

–data

onfluctuating

water

table

1.3A

pplyingsim

pleA

LMA

GR

AN

oneQ

ualitativeestim

ationof

Sim

pleand

lowcostqualitative

evaluationofsom

e∗

∗

multiparam

etric+

SW

AP

landattitude

(simple

andhydrologic

featuresor


scaleschem

e(FA

O-LE

)low

cost)(e.g.runoffvulnerability)

Schem

eB

:Land

1.5S

imulations

CropS

yst–

Pedologicalevidence

of1-DQ

uantitativeestim

ationIn

simulating

soilwater

flowprocesses,the

concept ∗∗

Representative

soilprofilesdata

evaluationm

odellingof

flow(absence

ofhorizontalofpotentialyield

(low(choice)

oftherepresentative

soilprofileis

crucial+

appliedto

water

balancepreferentialflow

s)cost)

andbetter

performing

thanaveraging

procedureof

allotheravailable

soilobservations3

cropand

andcrop

yield–

No

impeding

layerm

anyother

soilobservationsirrigation

–D

ataon

fluctuatingw

atersuitability

table(hydrologicalbased

1.6C

ropSyst

–P

edologicalevidenceof1-D

Quantitative

estimation

∗∗∗

∗∗∗

approaches)flow

(absenceofhorizontal

ofpotentialyield(best

preferentialflows)

method

oncost/

–N

oim

pedinglayer

performance

ratio)data

onfluctuating

water

table



Tabl

e5b

.Syn

optic

sche

me

ofth

ehy

drop

edol

ogic

alis

sues

inth

e3

case

stud

ies.

Soi

lmap

data

base

avai

labl

e(r

ank

Que

stio

nsto

Cas

eW

hatw

eha

veA

pplie

dH

ydro

pedo

logi

cali

nfor

mat

ion

Wha

twe

have

obta

ined

Wha

twe

getf

orhy

drol

ogic

appl

icat

ions

Per

form

ance

6C

ost7

refle

cts

incr

easi

ngin

form

atio

n)1be

answ

ered

stud

y4pe

rfor

med

mod

elco

nditi

onin

ghy

drol

ogic

alm

odel

ling

Sch

eme

C:

Dis

char

gean

d3

Sim

ulat

ions

TO

PK

AP

I–

Agg

rega

tion

ofso

ilm

appi

ngQ

uant

itativ

ees

timat

ion

Defi

nitio

nof

soil

dept

h.R

ecog

nitio

nof

(∗∗)

(∗∗)

Rep

rese

ntat

ive

soil

profi

les

data

2flo

wpe

akin

mod

ellin

gof

units

acco

rdin

gto

ofdi

scha

rge

and

peak

impe

ding

laye

rre

duci

ngav

aila

ble

soil

dept

hfo

r+

wat

ersh

edflo

odpe

aks

hydr

oped

olog

ical

crite

riaflo

wat

wat

ersh

edw

ater

stor

age.

Est

imat

eof

diffe

rent

drai

nage

addi

tiona

lfiel

dw

ork

(fine

tuni

ng)

on–

Rec

ogni

tion

ofim

perm

eabl

ecl

osur

ese

ctio

nca

pabi

lity.

repr

esen

tativ

eso

ilpr

ofile

sto

frag

ipan

laye

ram

elio

rate

hydr

olog

ical

perf

orm

ance

Sch

eme

D:

Land

1.7

Sim

ulat

ions

Cro

pSys

t–

Spa

tialv

aria

bilit

yQ

uant

itativ

ees

timat

ion

Com

plex

and

high

cost

quan

titat

ive

eval

uatio

nof

∗∗∗

∗∗∗∗

Sch

eme

Ban

dC

eval

uatio

nto

mod

ellin

gof

–P

edol

ogic

alev

iden

ceof

1-D

ofpo

tent

ialy

ield

(no.

9so

me

hydr

olog

icfe

atur

esor

proc

esse

sat

+cr

op1.

8w

ater

bala

nce

flow

(abs

ence

ofho

rizon

tal

was

the

best

met

hod

onla

ndsc

ape

scal

ein

clud

ing

spat

ialv

aria

bilit

yad

ditio

naln

ewso

ilob

serv

atio

nssu

itabi

lity

and

crop

yiel

dpr

efer

entia

lflow

s)pe

rfor

man

ce)

issu

es.

(inte

nsiv

e)to

amel

iora

tesp

atia

l(h

ydro

logi

cal

1.9

–N

oim

pedi

ngla

yer

infe

renc

eba

sed

–D

ata

onflu

ctua

ting

wat

erap

proa

ch)

tabl

e

Irrig

atio

n2.

2S

imul

atio

nsS

WA

PR

epre

sent

ativ

eso

ilpr

ofile

and

Opt

imal

wat

erD

efini

tion

ofH

FU

(∗∗∗)

(∗∗∗∗)

man

agem

ent

mod

ellin

gof

spat

ialv

aria

bilit

ym

anag

emen

tsch

eme

atA

ggre

gatio

nan

ddi

sagg

rega

tion

ofso

ilm

appi

ngirr

igat

ion

wat

erdi

stric

tlev

elun

itsre

quire

men

t

Det

ails

conc

erni

ngm

odel

sde

scrip

tion

are

avai

labl

ein

App

endi

xC

.

1T

hese

data

base

sre

fer

toso

ilm

apat

abou

t1:5

000

0w

hich

isco

nsid

ered

asu

itabl

esc

ale

toad

dres

sso

ilhy

drol

ogic

alis

sues

atca

tchm

ents

cale

.

2E

ach

repr

esen

tativ

eso

ilpr

ofile

refe

rsto

asp

ecifi

cso

ilty

peoc

curr

ing

inth

est

udy

area

.T

hese

soils

wer

ech

osen

for

thei

rre

pres

enta

tiven

ess.

3A

llso

ilob

serv

atio

nsre

fer

toal

lsoi

lpro

files

and

min

ipit

desc

ribed

/ana

lyse

din

the

stud

yar

eare

gard

less

ofth

eir

repr

esen

tativ

enes

san

dst

ored

inth

eda

taba

se.

4T

henu

mbe

raf

ter

the

poin

tref

ers

toth

em

etho

dfo

rea

chca

sest

udy

(e.g

.1.2

mea

nsm

etho

dN

.2in

case

stud

yN

.1).

5N

.A.–

nota

pplic

able

6T

heP

ears

onco

rrel

atio

nin

dex

and

the

Rel

ativ

eva

rianc

ein

dex

wer

esc

aled

from

0to

1an

dav

erag

ed.

The

nth

ein

terv

al0–

1w

asdi

ved

info

urcl

asse

sof

perf

orm

ance

rang

ing

from

the

low

est(

0–0.

25,

∗)

toth

ehi

ghes

t(0.

75–1

.00,∗∗∗∗

).T

his

proc

edur

ew

asap

plie

dfo

rth

eon

lyre

sults

conc

erni

ngth

eca

sest

udy

1.E

valu

atio

ns

conc

erni

ngth

eot

hers

case

stud

ies

(rep

orte

dbe

twee

nbr

acke

ts)

wer

ees

tabl

ishe

dac

cord

ing

to“e

xper

test

imat

e”,f

ollo

win

gth

ere

ason

ing

give

nfo

rca

sest

udy

1.

7C

osts

wer

esc

aled

toth

elo

wes

tval

ueof

case

sst

udy

1(s

eeF

ig.2

b).

The

nth

ein

terv

alw

asdi

ved

info

urcl

asse

sof

cost

rang

ing

from

the

low

est(

<10

,∗)

toth

ehi

ghes

t(>40

,∗∗∗∗

).T

his

proc

edur

ew

asap

plie

dfo

rth

eon

lyre

sults

conc

erni

ngth

eca

sest

udy

1.E

valu

atio

nsco

ncer

ning

the

othe

rsca

sest

udie

s(r

epor

ted

betw

een

brac

kets

)w

ere

esta

blis

hed

acco

rdin

gto

“exp

erte

stim

ate”

,fol

low

ing

the

reas

onin

ggi

ven

for

case

stud

y1.



Fig. 1. Case study 1: approaches used and their main characteristics.

3.1 Case studies

3.1.1 Comparative land evaluation models: from theFAO framework to simulation modelling

Land Evaluation (FAO, 1976) is defined as “the process ofassessment of land performance when used for specific pur-poses ...”. The Land Evaluation models can be devoted to(i) agricultural and forestry productivity (land suitability-related) and (ii) land use, planning and management for en-vironmental protection (land vulnerability-related). In gen-eral Land Evaluation, normally based on morphological,(bio)physical and chemical data derived from soil survey, hasbeen the most commonly used procedure worldwide to ad-dress local/regional/national land use planning; it works byusing multi-criteria (many soil and land parameters) classi-fication (matching tables, see Appendix C). Generally, thistype of models are directly applied to soil map units then tosingle small and homogeneous area whose suitability or vul-nerability is to be determined, without any information onspatial variability.

In the current example, elaborated after the paper ofManna et al. (2009) devoted to a land evaluation exercisefor maize (forage) production in a typical agro-ecosystem(2000 ha in the Lodi plain, Po valley, Italy), we will showthe overall potential of applying hydropedology in hydrol-ogy using nine alternative methods (Fig. 1) at increasinglevel of complexity. They ranged from a simpler standardland Evaluation approach (using ALMAGRA model, De La

Rosa et al., 2004) to a more extensive use of Richards’based simulation modelling (SWAP, van Dam et al., 1997 andCropSyst, Stockle et al., 2003). For a brief summary of themain characteristic of the applied models see Appendix C.

Then we evaluated cost/benefit ratio of the differentmodelling approaches and then whether complex models(e.g. mechanistic simulation models) are really sustainableand appropriate for a specific landscape hydrological task.

We used as data input several information reflecting the4 schemes reported in Table 5:

1. A pre-existing 1:50 000 soil map with 22 soil units. Thesoil database is constituted by the only representativesoil profiles data.

2. Soil database constituted by Scheme A (representativesoils) and including all soil profiles and minipit de-scribed/analysed in the study area regardless of theirrepresentativeness.

3. An additional hydrology-oriented and limited field work(fine-tuning) was carried out on the only representa-tive soil profiles and laboratory measurements of wa-ter retention (Dane and Hopmans, 2002) and saturatedhydraulic conductivity (Reynolds et al., 2002) wereperformed.

4. A further additional intensive hydrology-oriented fieldwork of 100 sites located after a stochastic spatial sim-ulation annealing procedure (Aarts and Korst, 1989).



This procedure was performed on the basis of pre-existing soil map, geological data and reference soilprofiles data. The corresponding data set include par-ticle size distribution and main chemical parameters(e.g. pH, EC, OC, etc.).

On 50 of these 100 supplementary sites bulk density, waterretention curve (Dane and Hopmans, 2002) and saturated hy-draulic conductivity (Reynolds et al., 2002) were measured.In the remaining 50 supplementary sites, water retention wasestimated by the PTF of Vereecken et al. (1989) and saturatedhydraulic conductivity by the HYPRES PTF (Wosten et al.,1998).

For modelling purposes, data on groundwater level (lowerboundary conditions) were obtained from a monitoring site(Bonfante et al., 2010) and from the soil map report. Dailyclimatic data were obtained from the meteorological moni-toring network of Regione Lombardia (ARPA Lombardia –http://www.arpalombardia.it).

An independently estimate data set of maize greenbiomass was used to test the results of the different mod-els; it was obtained from the relationship (R2 = 0.83) be-tween NDVI images (several 16-days based MODIS VI,250× 250 m; one visible and near IR QUICKBIRD,2.4× 2.4 m) and 10 green biomass measurements at ground(Manna et al., 2009).

The comparison between the different methods is based onboth cost and “predictive performance”, defined as the abilityof a specific method to successfully discriminate areas withdifferent forage maize suitability. All the output estimates ofthe nine methods were expressed as suitability classes withrespect to the production of maize biomass. The predictiveability was derived using different statistical indexes includ-ing (i) the Pearson correlation coefficient (r) to express thedegree of existence of a relationship and (ii) the relative vari-ance (1−Vcl/Vtot), given from the complement to 1 of theratio between within-class and total variance, to express thesuccess of the classification in reducing the variance withinclasses.

In Fig. 2a are shown these statistical indexes vs. land Eval-uation methods where the higher is the value the better isthe predictive performance. In Fig. 2b is given the final costper unit area, normalized with respect to the less expensivemethod.

Scheme A deals with the first given example of Table 5 andit shows the case of having a soil map with the only represen-tative soil profiles data and without any additional associateddata or hydropedological information to condition hydrolog-ical modelling. The outcoming results using a classical LandEvaluation approach – methods 1.1 and 1.2 – are very poorover all the Land Evaluation performance indexes, but in-deed these poor results are associated to a very low budgetinvestment (Fig. 2 and last two columns of Table 5). Thesefindings confirm that the worldwide multicriteria FAO-likeLand Evaluation approaches best perform at regional scale

rather than the detailed scale such as our case study. De-spite these results, a coarse hydrologic benefit can still berepresented by the simple and low cost qualitative evalua-tion of some hydrologic features or processes at inventoryscale. For example, a simple scheme of Land Evaluation us-ing available data as texture, presence of impeding layers,structure, depth of the profile, morphology, etc. can indeedprovide, at the inventory scale, qualitative soil suitability to-wards potential runoff occurrence.

In Scheme A a good alternative to the classical Land Eval-uation can be represented by the method 1.4. Here the onlysoil information is still the soil map database based on rep-resentative profiles; but we have applied a physically basedmodel of water balance and crop yield (CropSyst, see Ap-pendix C). In this case a quantitative estimation of poten-tial yield is then obtained and the leap in the predictiveperformance is evident despite the correlation remain low(Fig. 2a). The good results of the method 1.4 may be ex-plained by some hydropedological evidences applied in con-ditioning hydrological modelling. Those evidences werebased on the following: (i) the rather homogeneous geomor-phologic setting consisting in an alluvial plain (but still hav-ing two very distinct terracing systems) may have made the1-D water flow assumption feasible, especially consideringthe absence of pedological evidences of horizontal preferen-tial flows (e.g. no cracks, no slickensides, very limited oc-currence of clay and silt coatings, etc.); (ii) the latter, cou-pled with no evidence of compacted soil horizons (e.g. ironpan, fragipan, etc.), enabled a more confident applicabilityof standard PTFs that could easily failed in presence of suchfeatures. In fact, despite many PTFs try to take into accountthe soil structure through the bulk density, this is very of-ten estimated by another PTF (Rawls, 1983) and not directlymeasured, especially in deeper horizon where, unfortunatelyfor the applicability of PTFs, the occurrence of compactedlayers is both more frequent and more difficult to get anundisturbed soil sample. Finally, (iii) the soil map has alsosome information on fluctuating water table permitting a bet-ter definition of the bottom boundary condition.

Therefore, in analogy to methods 1.1 and 1.2, this methodcan be beneficially applied also to coarse hydrology ap-plications. For example, a rather inexpensive quantita-tive evaluation of some hydrologic processes at landscapescale accounting for large hydrological difference betweensoils (e.g. daily water storage, actual transpiration, drainagefluxes, etc.) can be easily applied, through the application ofa 1-D Richards based model.

The Scheme B concerns those case studies (methods) hav-ing in addition to Scheme A (soil map database including theonly representative soil profile data) an associated databaseincluding all other existing soil observations. In this case noimprovement was observed after applying averaging proce-dure (method 1.2 vs. 1.3 and method 1.4 vs. 1.5, respectively)of the soil observations (see Table 5 and Fig. 2a), in agree-ment with the findings of Heuvelink and Pebesma (1999)


http://www.arpalombardia.it


Fig. 2. Case study 1: performance indexes and costs of the tested LE methods in relation to the level of complexity.

concerning the importance of working on real soils and onreal measured data rather than processing or averaging ob-servations of several different soils.

Scheme C concerns those case studies having the only soilmap database of representative soils integrated by an addi-tional fine tuning consisting of an extra field and laboratorywork to obtain measured soil hydraulic properties. Modelshave been conditioned by hydropedological appraisal as re-ported for the Scheme A. Method 1.6, is the one showingall good performance indexes at the lowest cost (but still 19-fold higher than method 1.1). Particularly, it shows a correla-tion twofold higher than method 1.4 (Fig. 2a) despite the twomethods differ by the only hydraulic properties (measuredvs. estimated by PTF). In terms of hydrologic application,we can deduce from these findings that in simulating soil wa-ter flow processes, the concept (choice) of the representativesoil profile, especially if hydrologically characterized, is cru-cial and better performing than averaging procedure of manynew soil observations. Thus, strictly speaking, this methodis the lowest cost method to be chosen, producing consistentresults.

As expected the best prediction results were obtained af-ter abandoning the support of the soil-mapping units, strik-ingly increasing the number of samplings and analyses andeventually performing geostatistical analysis (Method 1.9,Scheme D) as in the case of digital soil mapping processing(Fig. A1). Needless to say, this approach was 47 times morecostly than standard Land Evaluation approaches (Fig. 2b)and poses major questions on its sustainability. Here we mustemphasise that this last approach is similar to a large bulk ofscientific literature (e.g. Western et al., 1999; Grayson andBloschl, 2000; Lyon et al., 2006) proving the importance ofincorporating spatial variability issues in hydrological appli-cations discharging any evaluation concerning cost benefitratio.

Summing up, comparison between the Land Evaluation(FAO) framework and mechanistic simulation modelling dis-proved the assumption that an increase in model mechanicsand complexity always means an increase in its predictive

ability. Indeed, in this case study, the predictive abilityevolves discontinuously with respect to model complexity.Data quality is indeed the leading parameter affecting theperformance of land evaluation and also plays a major rolein determining final costs.

Our findings, being largely based on detailed character-ization of soil hydropedological behaviour, show great po-tential also in other hydrological-based applications. Thisis the case for instance when addressing environmentally re-lated topics (e.g. groundwater vulnerability, nitrate pollution,rainfall-runoff processes, etc.) where soil functions such asstoring, filtering, transformation and interface for runoff gen-eration are important.

3.1.2 Mapping of soil hydrological behaviour forirrigation management

One typical issue in landscape hydrology applied to agricul-ture is both the planning and management of irrigation at thedistrict scale (which are typically smaller than 104 ha). Thisis traditionally achieved by using soil information to derivein a simplistic way the land suitability classes for irrigation(USBR, 1981 and Appendix C). The purpose of this suitabil-ity approach is to assess whether a soil has an inherent capac-ity to pay off both the overall investment plan of the hydraulicsystem for irrigation and to provide appropriate added valueto farmers.

A case study in Sardinia (Arangino et al., 1986) can help inappreciating this procedure. More specifically an irrigationcapability class was assigned to each soil mapping unit us-ing the soil map information. Parameters employed for per-forming such assignment were both qualitative (i.e. drainageclass, degree of mineral weathering, risk of soil erosion,etc.) and quantitative (i.e. slope, stoniness, rockiness, claycontent, salinity, carbonates, profile depth). These parame-ters were combined into an empirical multiparameter scheme(Appendix C) in order to produce a soil suitability map forirrigation. Then this suitability has been employed to reviewall areas included in the project and also to identify new areassuitable for irrigation.



Fig. 3. Case study 2:(a) soil map and(b) hydrological functional units classified on the basis of the “functional property”d800.

This approach is a kind of land evaluation and it hasbeen largely employed in different parts of the world to ad-dress local/regional/national land use planning. Despite itswidespread use, the scientific community has largely criti-cized the procedure for its qualitative and empirical basis andtherefore we can consider that both disadvantages (qualita-tive and crude approach) and advantages (simple and low-cost qualitative evaluation of some hydrologic features orprocesses at inventory scale) are the same given for the casestudy 1.1 in Scheme A (Table 5).

In order to achieve a better irrigation management and de-parting from this scheme, we have performed a further steptaking into account the inherent soil spatial variability withineach soil unit, especially of those soil properties largely in-fluencing the soil water balance, such as water retention andhydraulic conductivity.

We have applied physically based numerical models,which are known to be a valuable tool to simulate soil waterflow, yielding the soil-vegetation-atmosphere water balance.In particular these algorithms, once calibrated and validatedto the specific conditions of a study site, can be used to im-prove the efficiency of irrigation, thus contributing to the ra-tional use of water resources (Bonfante et al., 2010).

Unfortunately, the application of these models at the land-scape scale is strongly limited by the availability of spatiallyvariable information for a correct description of the soil hy-draulic behaviour. By coupling pedological information, as

already available in a soil map, and a small number of hy-drological analyses, D’Urso and Basile (1997) proposed amethod to classify soils according to their hydrological be-haviour in an earlier application of the HydropedologicalFunctional Unit concept, and falling in the case in Scheme Dof Table 5.

The main points of the procedure were as follows:

– identification of the representative soil profile withineach soil mapping unit;

– additional field work on each soil horizon of the rep-resentative soil profiles for the characterization of thehydraulic properties (namely,θ(h) andk(θ) functions)and particle-size distribution;

– calibration of a specific soil unit quasi physically-basedPTF (Arya and Paris, 1981), through coupling the mea-sured hydraulic properties and the particle size distribu-tions (Basile and D’Urso, 1997);

– additional new soil observations for application of thecalibrated PTFs to several points in the whole area;

– definition of a specific “functional property” (i.e.d800,the time interval between two optimal irrigation fromthe output of the simulation model) and application ofthe SWAP model (Appendix C) in all the sampled soils;



– aggregation and disaggregation of soil units by a newclassification of soils on the basis of the “functionalproperty” and demarcation of new units hydrologicallyhomogeneous (HFUs).

The method was developed on an 11 km2 river plain ser-viced by the Sinistra Sele Irrigation Consortium (south-ern Italy); an irrigated area where a soil map was alreadyavailable (Fig. 3a). The drainage process following irriga-tion was simulated by SWAP a Richard’s based 1-D model(Appendix C) and the chosen output was the “functionalproperty” d800, defined as the number of days required toreach an average pressure head of−800 cm in the soil layerbetween 10 cm and 30 cm depth. The value of−800 cm isthe soil pressure head at which stress starts in the chosencrop (alfalfa) and therefore the calculated “functional prop-erty” represents the optimal interval between two irrigationsin the event of on demand irrigation supply. The functionalpropertyd800 ranges from a minimum of 4 to a maximumof 8 days. The soil classification map shown in Fig. 3a wasmodified to account for the hydrological similarities high-lighted by evaluatingd800. The new classification allows pro-duction of the HFU map shown in Fig. 3b. Specifically, somesoil mapping units were disaggregated; for example the soilmapping unit n. 11 (Fig. 3a) was divided in two sub areas, thelower withd800<5.5 with a short interval time between twoirrigations appears from rather draining soils while, the upperone, close to the Sele river, was not classified (n.c.) becauseof the high variability at the investigation scale. Others wereaggregates; for example part of the soil mapping units 10and 6 (Fig. 3b) merged in the same HFU (d800<5.5).

This new classification was the basis of several papersdevoted to optimal water management (D’Urso and Mi-nacapilli, 2006; Bastiaanssen et al., 2007).

It is useful to emphasise that this approach shows the largeflexibility in using HFUs; in fact once the procedure hasbeen adopted to perform a specific task (in our case thed800functional property), very little adjunctive work is neededfor addressing new tasks requiring for instance new (purposedriven) functional properties and maps.

3.1.3 Flood event simulation in an ungauged basin: thecontribution of soil data

In the Sangone basin (NW of Italy) the only data availablerefer to the meteorological and discharge time series at theclosure section. This basin of about 150 km2 shows consid-erable pedological and hydrological complexity. It is formedby three main geomorphologic systems: mountain ridges,slopes at different gradients and aspects, and valleys. Themain question we addressed is to what extent a parsimo-nious identification of soils (forms) can help in interpret-ing the hydrological complexity hidden in flood forecasting(functions).

The study was performed through a comparison of theresults obtained by using TOPKAPI, a physically based

distributed model of infiltration-runoff processes with parsi-monious parameterization (see Appendix C), applying blind(i.e. without any calibration) simulations relative to two dif-ferent levels of soil knowledge, namely: (i) soil characteris-tics retrieved after a global soil map resource, (ii) soil charac-teristics derived after a (parsimonious) soil survey campaign.The soils were accordingly parameterized relative to the dif-ferent approaches.

The approach with the most basic level of knowledge(Scheme A in Table 5) employed as dataset of soil typesthe FAO/UNESCO Soil Map of the World. This datasetwas developed under the SOTER Programme and the oldFAO legend was replaced by the update World ReferenceBase for Soil Resources (WRB, 2006). This dataset includes106 soil units of which four are in the study area, based onZobler’s assessment of the FAO/UNESCO Soil Map of theWorld. Soil properties such as soil depth and saturated hy-draulic conductivity are derived for each soil type from theFAO-UNESCO soil classification. The soil hydraulic prop-erties, i.e. water retention and unsaturated hydraulic conduc-tivity, represented by van Genuchten’s – Mualem formulas(van Genuchten, 1980) are estimated by PTF. This datasetclearly has the limitation of a coarse resolution and of qual-itative information on the soil characteristics. However, it isrepresentative of the minimum level of knowledge availablefor the implementation of a hydrological model. Moreover,the dataset was also used in several applications (Doll et al.,2003; Liu et al., 2008) where other (local) sources of datawere not available.

The second approach (soil hydrology driven) having ahigher demanding level of knowledge (Scheme C in Table 5)consisted of the following main steps:

1. A preliminary study of pre-existing available informa-tion (soil use map, 1:250 000 soil map, 1:50 000 soilmap available for the only lower plain area, 1:50 000geological map, 1:10 000 DTM).

2. A synopsis and definition of preliminary soil map-ping units obtained, along with standard thematic layers(e.g. geology, land use, etc.), using some environmen-tal covariates data. They resulted, from a pedologicalviewpoint, very useful for describing the specific soildistribution of the study area. Between them, we in-clude a fuzzy c-means clustering to classify the DEM(de Bruin and Stein, 1998), and a study of vegetation byanalysing the NDVI (Normalized Difference VegetationIndex) using Landsat and Modis images (Rouse et al.,1973; Tucker and Sellers, 1986; Wang et al., 2004).

3. A soil survey mainly limited to describe and to sam-ple supposedly “representative soil profiles” inside eachidentified preliminary soil mapping unit (about 50 soilprofiles and 50 minipits). The choice of representativesoils is indeed crucial (and questionable) in this proce-dure; but at the present day we do not think that there are



better “sustainable” options to the use of a traditionalpedological understanding (CLORPT).

4. Soil chemical analysis on about 220 soil samples (pHin H2O, pH in KCl, pH in NaF, EC, OC, CEC, EB, andthen Fe, Al and Si extracted in oxalate at pH = 3).

5. A synopsis of all acquired data and definition of aschematic soil map having 19 soil mapping units. Itmust be emphasised that this soil map produced frompoint 1 to 5 is named “schematic” because has muchless soil observations (and then lower costs) than thoserequired by standard soil maps (Appendix A).

Points 1 to 5 refer to the specific contribution of pedol-ogy to discriminate theforms (Lin et al., 2008) of thearea.

6. The use of TOPKAPI model (as many other hydrolog-ical models) for an application in PUB where nor ob-served neither gauged internal data are available in or-der to simplify the input dataset and then to aggregatethe 19 soil mapping units into a fewer number map-ping units. This point is crucial both from an hydro-logical point of view since it requires the use of par-simonious but efficient models and from an hydrope-dological point of view because it requires an aggre-gation of mapping units having, potentially, a similarhydrological behaviour. This has been performed us-ing the hydropedological reasoning given in Table 4.Therefore the original 19 soil mapping units were ag-gregated to form 8 mapping units named here as “soil-landscape” units, because strongly based on the land-scape features (Table 6). For instance this has been thecase for four soil mapping units referring to “ancientfluvio-glacial deposits, erosional slope of river terrace,moraine steep slope, moraine crest” aggregated intothe new soil-landscape mapping unit “ancient fluvio-glacial deposits”. In this case aggregation was ruled bythe occurrence of a flow impeding soil layer (fragipanhorizon) at shallow depth in very deep soils. Then inthis soil-landscape unit the soil water balance is mainlydriven by this fragipan horizon rather than soil hydro-logical properties and/or soil depth.

7. Determination of soil hydraulic properties at the eightsoil-landscape units. In some stony soils and thinhorizons, water retention and hydraulic conductivitycurves were determined through an inverse method fol-lowing a process of infiltration at predefined pressureheads (Simunek et al., 1998). Undisturbed soil sampleswere analysed in the laboratory by applying the fallowhead method (Reynolds et al., 2002) for the saturatedhydraulic conductivity, Wind’s method (Arya, 2002)for the unsaturated hydraulic conductivity and the wetbranch of the soil water retention curve, and the WP4-Tmethod (Bittelli and Flury, 2009) for the dry branch of

the soil water retention curve. Data were parameterizedaccording to the constitutive functional relationships ofthe TOPKAPI model.

8. TOPKAPI model requires a single set of parametersfor each soil profile, assuming a substantial homogene-ity along the soil profile. The consequent problem ofidentifying an effective set of hydraulic properties foran equivalent fictitious single-layered soil profile, giv-ing the same hydrological response as the real layeredsoil, was solved as follow: for each soil profile a 1-Dsimulation by means of Richards’-based Hydrus model(Simnek et al., 2008; see also Appendix C) was per-formed describing the profile according to the genetichorizons each one with own measured hydraulic prop-erties (point 7). Then an inverse estimation procedurewas applied assuming the profile as just one single hori-zon; the mean of each parameter along the profile wasassumed as initial estimate and the water storage,W(t),coming out from the multilayer simulation was used as“measured” hydrological response to be matched by theone-layer simulation. Under the same boundary con-ditions the equivalent homogeneous medium will havethe same amount of stored water as the heterogeneousone. It follows then the hydraulic functions obtainedfor this equivalent homogeneous medium are the effec-tive hydraulic functions of the heterogeneous medium.This approach has been largely adopted in the literature(among others: Yeh, 1989; Jhorar et al., 2004).

In the Fig. 4a and b the results of the comparison of thesimulations of the FAO soil map against those referring tothe soil-landscape units are shown. As can be easily seen,simulation from the first approach using the FAO soil mapreproduces none of the important features of the hydrographsuch as peak flow, the rising and recession limb. Differently,simulation from the second approach, which exploits the soildata collected in the field and the classification, is able to re-produce such features. The importance of the soil properties,which most affect the simulation performance, can also berecognized. The peak of the discharge is meaningful of therunoff mechanism. In the first parameter set, the soil is shal-low and with low permeability. The runoff volume is thenoverestimated because the soil capacity is exceeded in muchof the catchment and the volume infiltrated into the soil isvery low. In the second parameter set, the soil distributionis more accurate, enabling a distinction to be made betweenthe hillslope, with a shallow soil, and the valley, with deep,permeable soils. The peak flow is correctly estimated, whichmeans that also the runoff mechanism is captured.

The recession limb of the hydrograph is strongly con-nected with the baseflow. In the first approach the baseflowis underestimated since the transmissivity of the soils, espe-cially those beside the river, is not correctly parameterized.In the second approach the slope of the hydrograph is verysimilar to that observed. An important role here is played



Table6.H

ydrologicalrationaleused

insoilm

apunits

aggregationfor

thecase

study3.

OriginalS

oilMapping

Units

Rationale

(mainly

Soil-landscape

descriptionofthe

aggregatedsoilm

appingunits

Pedologicalfeatures

affectingN

ewaggregated

Soil

hydropedological)behind

Soil

hydrologicalpropertiesM

appingU

nitsM

appingU

nitsaggregation

Mountain

ridges,intermediate

1.verylow

water

storageM

ountainousareas

oforographicculm

inationw

itha

convexshape.

Soils

arethin

(25to

45cm

)w

ithfrequent

Soildepth:

25–45cm

;frequentcoarseM

ountainridges

shelves,summ

itbaredareas,

capability2.convex

areasw

ithcoarse

fragments

(30–35%

)w

ithstrong

granularstructure.

Topsoilorganicm

attercontent:

6–9%

;fragm

ents:30–35

%;

moraines

(ridgeof

uplandorographic

boundariespH

:4.3–4.5,siltloamand

loamy

sand.m

orphogenizedbody).

3.fastdrainage

South-facing

slopes,intermediate

1.moderate

water

storageS

outh-facingm

ountainslopes,w

ithconvex

orstraight(shoulder)

slopes(15

◦–35◦),elevation:

500–2000m

.M

oderatesoildepth

140–180cm

;S

outh-facingslopes

shelves,south-facingm

orainecapability

2.slopesouth

aspectS

oilsare

moderately

deep(140–180

cm)

with

comm

oncoarse

fragments

(5–15%

)and

stronggranular

comm

oncoarse

fragment:

5–15%

deposits.3.fastdrainage

structure.Topsoilorganic

matter

content:3–12

%;pH

:4.2–4.7;sandyloam

andsiltloam

texture.

South-facing

concave1.high

water

storagecapability

South-facing

mountain

slopeshaving

concaveshaped

slopes(15

◦–30◦);elevation

500–2000m

.S

oilsare

deepD

eepsoils

(250–500cm

);coarseS

outh-facingconcave

slope,intermediate

2.slope,southaspect3.fast

(250–500cm

)w

ithfrequentcoarse

fragments

(15–20%

)and

stronggranular


matter

fragments:

10–20%

;lowandic

slopesshelves,colluvial

drainagecontent:

5–10%

;pH:4.3–5.3;siltloam

texture.properties

(Alo

+0.5

Feo

%:

0.36–0.42)deposits,m

oraines(south-facing

concaveslopes).

North-facing

slope,intermediate

1.loww

aterstorage

capabilityN

orth-facingm

ountainslopes

(15◦–35

◦),fromconcave

toconvex

shaped;elevation:500–2000

m.

Soils

areS

oilsranging

fromshallow

(<

10cm

)N

orth-facingslopes

shelves,colluvial2.slope,north

aspect3.fastthin

(25–50cm

);coarsefragm

entsvaries

fromcom

mon

tofrequent(10–20

%);strong

granularstructure.

tom

oderate(50

cm);coarse

fragments:

deposits,moraines

(north-facingdrainage

Topsoilorganicm

attercontent:

5–18%

;pH:4.0–4.5;siltloam

andsandy

loamtexture.

5–25%

;andicproperties

arelow

toslopes).

moderate

(Alo

+0.5

Feo

%:

0.5–0.9)

Sum

mitbared

areas.1.no

water

storagecapability

Mountain

summ

itareasw

ithextrem

elyarticulated

slopes(0–90

◦);elevation:1400–2600

m.

These

areasR

ocksatthe

surfacew

ithonly

fewS

umm

itbaredareas

2.highlyeroded

3.veryfast

havebeen

subjectedto

strongglacialerosion,therefore

soilsare

almostabsentw

ithm

aximum

thicknessof

verythin

soils(

<10–15

cm)

drainage10–15

cm.

Mostofthe

areais

coveredw

ithloose

unaltereddebris.

Ancientfluvio-glacialdeposits

1.variablew

aterstorage

Pre-H

olocenedepositionallow

tom

ediumsloping

(1/2◦–20

◦)bodies

offluvio-glacial;elevation:500–800

m.

Deep

soils;frequentcoarsefragm

ents:A

ncientfluvio-glacial(cones),erosionalslope

ofrivercapability

accordingto

Soils

aredeep

(150–300cm

);coarsefragm

entsvaries

fromcom

mon

tofrequents

(10–30%

);stronggranular

10–30%

;occurrenceoffragipan

(water

depositsterrace,m

orainesteep

fragipandepth

2.ancientlowstructure.

Topsoilorganicm

attercontent:

about3%

,pH:5.5–5.8;siltloam

texture.im

pedinghorizon)

atabout30cm

slope,moraine

crest.sloping

fluvio-glacialbodiesdepth

3.fastdrainagedue

tofragipan

depth

Recentfluvialfan

deposits1.very

highw

aterstorage

Recentdepositionalbodies

(Holocene)

mainly

offluvialorigin;slopes:<

2◦

(depositionalsurface)−10◦

Verydeep

soils;lowcoarse

fragments

Recentfluvial

(cones),concavevalley

overcapability

2.recentfluvial(erosionalslope);elevation:

400–700m

.S

oilsare

verydeep

(>

300cm

);coarsefragm

entsvaries

fromabsent

(<

5%

);occurrenceofm

ollichorizon

depositsm

orainedeposits,m

orainegentle

deposits3.m

oderatedrainage

tolow

(0–5%

);stronggranular


matter

content:about2

%;pH

:5.0–5.7;siltloam(highly

poroussoillayer)

slope.texture.

Actualfluvialfan

deposits1.high

water

storagecapability

Actualdepositionalbodies

(lasthundredsofyears)

ofafluvialorigin,w

itha

slopevarying

between

1◦

and3 ◦,

Verydeep

soils;veryvariable

contentA

ctualfluvial(cones),deposits

atlowvalley

2.actualfluvialdepositsaltitude

between

350and

700m

a.s.l.T

hesoils

arevery

deep,more

than300

cmthick,the

coarsefragm

entsofcoarse

fragments

(5-50%

);deposits

position.3.m

oderatedrainage

frompoor

toabundant(5–50

%).

The

surfacehorizon

has6

%organic

matter,pH

between

5.6and

6.0,siltloam

texture,andgranular

structureofw

eakto

moderate.

Riverbed,ditch

ofriverincision.

Water

bodiesR

iverbed



(a)

(b)

Fig. 4. Case study 3: hydrographs of the Sangone catchment(a) from January 2002 to September 2007, and(b) at the main flood event inSeptember 2006. Symbols (+) are the observed discharge, dashed line is the simulation using the FAO dataset and black line is the simulationusing the soil-landscape data set.

Table 7. Statistical indexes of the simulated discharge applyingFAO and soil-landscape map.

Parameter FAO Soil-soil landscapemap unit

map

Nash-Sutcliffe coefficient 0.21 0.89RMSE (m3 s−1) 14.68 6.58

by the soils in the deposit (downstream) that, according tothe parameterization, have a high transmissivity (relative soilthickness and permeability) such that the low flow is alsocaptured.

In Table 7 are given the computed Nash-Sutcliffe (NS) andthe Root Mean Square Error (RMSE) coefficients aiming toassess the overall simulation performance. Both the NS and

the RMSE show a clear improvement of the model perfor-mance when the soil-landscape map is applied. Particularlythe NS coefficient, which is more hydrologically based, showa large increase since both peaks and baseflow are better cap-tured by the soil-landscape unit map rather than the FAO UN-ESCO soil map.

The (expected) different performance between approachesmeasures the improved knowledge gained by just using theinformation coming from the landscape soil mapping unitsclassification. The FAO soil map is taken here as the “nullhypothesis” of the knowledge which one could eventuallyget, while the landscape soil map proposed may be thoughtas the minimum sustainable compromise for more realisticmodel predictions. In other words if one should use a modelwithout a specific campaign to collect or elaborate hard data,its level of knowledge it not “nothing”, but – without any ef-forts – “something” i.e. those information already available.With this level of knowledge any improvements should becompared to measure its adding value.



4 Conclusions

The importance of knowing the soil distribution (forms) andprocesses (functions), which determines a sort of physicalsignature of the catchment, becomes very important whenmaking hydrological predictions in the absence of hydro-logical monitoring data (such as in ungauged basins). At-tempting to make the interaction between hydrology and soileffective – also in terms of catchment classification – thispaper aimed to explore potential and limitations of soil sur-vey and soil mapping. This aim has been performed byanalysing the quality and quantity of information embeddedin soil map and illustrated hydropedological analysis appliedin three case studies from Italy (i.e. comparative land evalu-ation modelling, mapping of soil hydrological behaviour forirrigation management at district scale, and flood event sim-ulation in an ungauged basin). Eventually we showed thatdata from soil survey have lots to offer but also that specialcare is required in handling soil database data if their fullhydrological potential is to be achieved.

Regardless the diverse and specific outcomes of each ofthese case studies some common key hydrological pointsmust be risen:

i. The complexity of the model to be appliedVery simple and widely applied models such as theLand Evaluation (LE) procedure has been widely criti-cized by the scientific community for its qualitative andempirical basis which makes it difficult to successfullyaddress many new soil-(agro-)environmental challengeswhich indeed require the dynamic characterization ofthe interrelated physical and chemical processes takingplace in the soil landscape.

On the other hand, many theoretically problems inthe hydrological application of physically based model(e.g. physics of heterogeneity, equations and parame-ters scale integration, etc.) are still unsolved, despitethey where posed many years ago (Beven, 1989).

Moving between these two extreme approaches, albeitnot linearly, increasing model complexity and hencethe accuracy of the described phenomenon requires anincrease in basic data parameters and thus generateshigher costs. This is even more important in ungaugedbasins where the lack of data is a crucial factor towardsthe proper modelling choice, according to “the rightresults for the right reasons” statement (Grayson andBloschl, 2000). In this respect, this paper support thata generalization of the methodological pathway fromsimple to complex modelling application in a land eval-uation procedure for maize production (case study 1)can be extended to landscape hydrology issues. It wasproven that, for this case study, the Richards’ equa-tion could be effectively applied also at landscape scale.This finding was tested also for others case studies,

where indirectly (case study 2) or through simple catch-ment response (case study 3) we evaluated the goodnessin using physically based model at the landscape scale.

These good results may also be ascribed to

ii. the good choice in selecting representative soil profilesfollowed by their hydrological characterizationThis was particularly true in the case study 3, char-acterised by a limited dataset of hydrological mea-surements. It was demonstrated the effectiveness of ahydrology-devoted and relatively inexpensive soil sur-vey (with respect to standard soil survey approaches)only aiming to identify the main soil mapping units andto select few representative soil profiles where to pro-duce a soil hydraulic characterization. In general terms,it must be emphasised that both the identification of soilmapping units and the selection of representative soils ismore difficult in an alluvial plain settings rather than intypical upper catchment landscape where both topogra-phy and soil outcrops better assist these critical choicesmade by the pedologist.

The other crucial issue, regardless the representative-ness of specific soils, is indeed the complex problem ofsoil heterogeneities, typically addressed through a spa-tial analysis taking into account intra-unit spatial vari-ability of hydraulic properties.

To this respect, a simple widely used approach is

iii. the spatial analysis of soil hydraulic characteristics (pa-rameters) through PTFs.This lead to assume that the spatial variability of textureand soil pore architecture (e.g. pore size distribution,pore connectivity, water retention, hydraulic conductiv-ity) are similar. This assumption can be rather con-troversial considering the different underlying physicsgoverning soil particles and soil pores, notwithstandinga large bulk of literature on PTFs use is based on it.To this respect, in case study 1 we showed that the het-erogeneity was better described, in terms of statisticsand spatial pattern, by real measurements of hydraulicproperties than those estimated by PTF. The issue hasbeen further explored in the case study 2 where a quasiphysically-based PTF was calibrated for each soil map-ping unit; in this specific case the inter-unit uncertaintywas taken on board by the representative soil profile hy-draulic properties measurement and the intra-unit het-erogeneity estimated by the coupling of PTF and theunit-specific calibration function. This procedure is ca-pable of transfer the “structural” information enclosedin the water retention curve (measured in each horizonof the representative soil profile) to the PTF applied inthe same horizons of the soil mapping unit.



iv. How the intra-unit and inter-unit variability can be re-organized according to hydrological reasons.The case study 3 shows how pedological-drivensoil mapping units were aggregated to form fewerhydrological-driven soil-landscape units (e.g. on the ba-sis of the effective soil depth) for water balance pur-pose. This is an important contribute of pedology serv-ing hydrology in cases a reduction of the discrepanciesbetween the model (i.e. TOPKAPI) scale and the soilobservation scale.

Another example worth mentioning concerning the soilmap unit reorganization was presented in the casestudy 2 showing how it is likely to classify soils bydefining some “functional properties”, which describethe hydrological behaviour of the entire soil profile(Wosten et al., 1986; Bouma et al., 2008). This wasperformed coupling concepts reported in points (i), (ii)and (iii) in order to produce a map of Hydropedologi-cal Functional Unit. It is useful to emphasise that thisapproach shows large flexibility; in fact once the pro-cedure has been applied very little adjunctive work isneeded for addressing new hydrological tasks.

v. Finally we aim to emphasise the key issue concerningwhich soil horizons (features, properties) can be effec-tive and then must be strongly acknowledged for hy-drological purposes.For example, the application ofPTF in soils with andic features can induce serious mis-takes. We have provided some evidences throughout ourcase studies including (a) the presence/absence of fragi-pan horizon was a key factor in the aggregation of soilunits into landscape units (case study 3); (b) soil pro-files with the occurrence of thick mollic horizons – soillayers having high water storage capacity – enabled toaggregate otherwise different soil mapping units (casestudy 3).

In all given case studies, soil map database have been in-tegrated with some extra soil hydrological data (fine tuning).The consequent high increase in cost for such an analysis canbe fully justified only in a more inclusive hydropedologicalframework where the same information – in terms offormsandfunctions– can produce integrated results in different ap-plications with a very similar data requirement.

In conclusion it is important to stress that an effective in-teraction between pedology and hydrology to address land-scape and watershed hydrology can be very fruitful for bothdisciplines but yet there is much work to be done. Morespecifically we believe that the following must be consideredkey issues:

– greater awareness on the part of hydrologists about howmuch and what information, both directly and indirectlyrelated to landscape hydrology, lies behind a soil map inthe attached soil database;

– awareness on the part of pedologist of the need to moveactual assessment of hydrological features/properties(i.e. runoff, cracking, permeability, flood, structure,mottles, etc.) from qualitative to quantitative or at leastsemi-quantitative schemes to better incorporate hydro-logical parameters in soil classification;

– the inclusion of quantitative information on soil spatialvariability (e.g. variance, semivariogram) and spatialdistribution of prediction errors within a new conceptionof soil maps. In this framework digital soil mapping canprovide a major contribution to hydropedology;

– in a context of a wider environmental management plan-ning, the use of a common base of mainly physical soilinformation can be a fundamental tool able to approachdifferent soil hydrological processes.

Appendix A

Soil mapping procedures

A1 Introduction

Dokuchaev and Jenny, respectively in 1882 and 1941, firstrecognized and then attempted to formalize soil formationby the following equation:s =f (cl, o, r, p, t , ...) (world-wide known as CLORPT), wheres is any soil property, clthe climate,o the organisms,r the topography,p the parentmaterial (the state of the soil at time zero),t the absolute ageof the soil, and the dots ... represents additional non-specifiedfactors or factors interaction.

In true landscapes, the above equation is both a powerfulconceptual framework but it also shows limitation in produc-ing a fine exhaustive spatial analysis of soils. This is becausesome of the factors of soil formation, formalized in the equa-tion, are very difficult to be determined such as the age of pe-dogenesis and/or the status of other soil-forming CLORPTfactors during the (long) life of a developing soil. Despitethese difficulties, in the last century the scientific commu-nity of pedologists used the CLORPT conceptualization andstandardized survey methods to analyse and report the spa-tial distribution of soils through the production of soil maps.These maps were then employed as indispensable tools forplanning proper land management.

Some recent conceptualizations, reviews and local stud-ies have been also performed on the use of mechanisticmodels of soil formation, partly based on the CLORPTconceptualization, at both point based and landscape basedscales (Samouelian and Cornu, 2008; Minasny et al., 2008;Salvador-Blanes et al., 2007).

More recently, soil scientists involved in spatial quantita-tive research have translated Jenny’s mechanistic CLORPT



formulation into the more quantitative and inference-basedSCORPAN model (McBratney et al., 2003):

S = f (s, c, o, r, p, a, n) (A1)

whereS is the target soil property/class,s stands for othersoil information at the same location,c is climate,o is or-ganisms,r is topography,p is parent material,a is age, andn stays for spatial position. Any factor can be formally ex-pressed as a function of spatial (x, y, z) and temporal (t)coordinates, such that the target soil property can assumeto the highest degree of complexity the formS(x, y, z, t).However pedometricians very often employ the simpler bi-dimensionalS(x, y), whileS(x, y, z) is a bit used in the spa-tial three-dimensional modelling (e.g. Park and Vlek, 2002;Lark and Bishop, 2007), and theS(x, y, t) template is some-times used in spatio-temporal geostatistics (e.g. Bilonick,1988; De Cesare et al., 2001a,b; Snepvangers et al., 2003).Here if not explicitly mentioned we refer to a simple bi-dimensional domain.

The SCORPAN formalization consists of an empiricalquantitative description of relationships between soils andenvironmental factors with a view to using these as soil spa-tial prediction functions for determining the spatial distribu-tion of soil types and soil attributes. It is an adaptation ofCLORPT, not for a mechanistic explanation of soil formationbut for an empirical representation of relationships betweensoil and other georeferenced factors. In SCORPAN this isobtained by extending the five soil forming factors with theaddition of the spatial positionn.

Digital Soil Mapping (DSM) is the production of maps ofsoil types and soil properties in digital format assisted by acomputer. A detailed discussion of both conventional andDSM methods may be found in some reference books (Dentand Young, 1981; McKenzie et al., 2008).

It should be pointed out that conventional and digital soilmapping are integrated fields and an overall presentationshould be given. However we prefer here a separate and briefdescription of these two complex procedures for the purposeof promoting an easy reading and understanding to the envi-ronmental hydrologists.

A2 Conventional soil mapping

The conventional approach describes the spatial complex-ity of soils in the landscape by means of an expert knowl-edge (mental) model developed by the pedologist using animplicit predictive model. It is strongly qualitative, com-plex and rarely communicated in a clear manner (it has anadaptation to each landscape). Basically the pedologists per-form a preliminary spatial study of climate, land use, topog-raphy, geology and then investigate the spatial relationshipsbetween these environmental features and soils observed inthe field and then analysed in the lab.

The approach has a high degree of subjectivity and un-certainty, and the soil spatial variability is generally not

described. The main steps typically employed in producingsoil maps are illustrated in Fig. A1 and include:

1. Acquisition of all available information on the spa-tial distribution of soil-forming factors (e.g. geologicalmap, geomorphology, DEM, climate data, etc.) includ-ing those obtained after remote and proximal sensing.

2. A synopsis of all such information for producing a pre-liminary landscape classification. This synopsis, as-sisted by the use of photointerpretation (either analog ordigital), consists in segmenting a region into many land-scape units supposed, in this preliminary step, internallyhomogeneous in terms of soil-forming factors (at leastthose available). In other words, the procedure employsthe strongly deterministic basis of the soil-forming fac-tors to segment, with a first approximation, the regionof interest into areas for carrying out soil sampling andanalysis.

3. In these segmented areas a preliminary soil surveyis then carried out. This survey consists in openingup holes and trenches (and also in performing handdrilling) where, following standardized procedures, avertical section of soil called the soil profile and the site(about 10 m2) where the profile is located, are described.

4. The description typically consist in recognising differ-ent horizontal layers, called pedogenic horizons, andalso in determining, for each of the identified horizons,specific features and properties that can be directly de-rived in the field (Soil Survey Staff, 1993; FAO, 2006).Finally, soils are sampled for chemical and physicalanalysis to be performed in the laboratory. An exampleof the kind of soil features that are described in a soilsurvey is given in Table 2 where we have highlighted inbold and italics those features which are of great rele-vance to hydrology.

5. Assuming a profile consisting of only three horizons,the potential output will include about 257 field data(including descriptions of the profile and the soil sam-pling station and in the theoretical case where that alltypes of soil features occur) and 36 laboratory-baseddata for each soil observation. Using the (qualitative,semi-quantitative, quantitative) field and laboratory dataobtained, soils are then classified into categories us-ing international systems of soil classification such asSoil Taxonomy (USDA, 2010) or World Reference Base(WRB, 2006).

6. On the basis of the results obtained after the preliminarysoil survey, a preliminary soil mapping units (SMU)map is produced after a synopsis of both landscape andsoil information (also named preliminary soil correla-tion). In this map, one or more soil types (Soil Typolog-ical Units also typically named as STU) are associatedto each preliminary SMU.



Digital Eleva+on Model (DEM) and Terrain Parameters (TP)

Digital Terrain Analysis

Photointerpreta+on (analogical or digital)

EMPIRICA

L

MEC

HANISTIC

MODELLING

…

Auxiliary and Exhaus+ve Inputs

Systema+c Soil Survey

Binary Data

Preliminary Landscape Classifica+on

Preliminary Soil Survey

Final Soil Mapping Units (vector data, report, db)

?

Preliminary Soil Mapping Units (Synopsis of landscape and

soil informa+on)

Soil Variable Spa+al Map

(pre

limin

ary

corr

elat

ion)

(f

inal

cor

rela

tion)

Test of soil map (by the organiza+on financing the soil map)

revi

sion

Environmental Datasets (CLORPT/SCORPAN)

Data Mining

geor

efer

ence

d so

il db

Probe Sensing remote

air-‐borne proximal

[case study #3]

[case studies #1,2]

[case study #1]

Fig. A1. Data flow of the main steps involved in both conventional (left panel) and digital (right panel) soil mapping. Some remarks:(i) the same set of environmental data can be used for conventional or digital soil mapping; (ii) to promote understanding the two mappingprocedures are separately depicted with few key interaction points (e.g. lab data are used to calibrate SCORPAN like models and to co-defineSMU).

7. Systematic soil survey (in accordance with standards re-quired by the organization commissioning the survey)and soil analysis on the basis of the preliminary SMUmap.

8. First draft of the final SMU map and soil legend (USDA,1993). In this drafting process, the aim is to organizeand produce a synthesis of all the soil knowledge inthe study area within a coherent framework (typicallynamed as final soil correlation). This rather complex

task is typically performed by aggregating all soil in-formation into a limited number of soil mapping units(SMU), each being represented by the dominant (andco-dominant) soil type (STU).

9. Field control to check soil mapping units and drafting ofthe final soil map (SMU) with explanatory notes wherefor each STU is also given a representative profile in-cluding field description and lab analysis.



Table A1. Example of standards employed in soil survey and conventional soil mapping.

Typical user Type of Mapping N. obs Remotely Minimum Costscale scale (in 100 ha) sensed polygon Euro ha−1

images (ha)

Country Schematic 1:1 000 000 1:100 000 4000 n.a.

Country Inventory 1:250 000 0.16 1:100 000 250 0.2–0.7Inter-regionRegion

Province Semi- 1:50 000 1–3 1:20 000 15 7–10District detailed 1:33 000Watershed authoritiesMountain communities

Municipal district Detailed 1:25 000 10/50 1:8000 2 n.a.

Municipal district Very detailed 1:5000 100 0.1 400–500Farms

10. Final test of the soil map, typically performed by theorganization commissioning the soil map.

11. Final review and publication and release (upon request)of .shp file and soil database.

What is evident here is the complexity of the process of sur-veying and mapping of soils, the large quantity of data tobe processed and the importance of a good synopsis of theobtained soil knowledge.

Here we must emphasise that the information content of aconventional soil map, is obviously highly dependent on thescale. For the sake of this specific paper, we only report someexamples of methods and standards in Table A1. From thistable it is possible to emphasise that soil maps are typicallymade at very different scales and that moving from more gen-eral scales towards more detailed scales a sharp increase inthe number of soil observations and their associated costs isobserved.

A3 Digital soil mapping

Digital soil mapping can rely upon, but is distinct from, soilmapping (Fig. A1). Digitized and georeferenced soil sur-vey information does not become DSM until they are usedto derive other soil related information within a softwareapplication. New spatial soil information is generated bycoupling (field and laboratory) observations at survey loca-tions with exhaustive auxiliary information using inferencesystems. A soil inference system is a pedometric (mathe-matical or statistical) SCORPAN-like model and it can bespatial or non spatial according to the domain of inference,that is the inference takes place in the spatial or in the at-tribute domain, respectively. Inference systems use auxil-iary information (SCORPAN factors) correlated to the targetsoil property/class (McKenzie and Ryan, 1999) and coming

from different fields such as remote sensing (hyperspectraland multitemporal imagery at relative high spatial resolu-tion), proximal sensing (e.g. electromagnetic induction scan-ning techniques), and digital terrain analysis (i.e. the calcu-lation of derived terrain parameters using a Digital ElevationModel; Wilson and Gallant, 2000). This auxiliary informa-tion is also said to be exhaustive because it is densely avail-able on a regular grid base to such an extent that every sur-veyed location and any unknown point are both covered. Theprediction of a soil property at a given site is thus obtainedfrom known observations about that property neighbouringthe site and from exhaustive auxiliary data at both the knownsoil observations and the unknown site to be estimated un-der different neighbourhood schemes according to the typeof the inference system used. In other words, a soil infer-ence system is a way to select amongst several auxiliariesinformation and use the remaining predictive information tocalibrate, validate and simulate the model at hand.

Much DSM work worldwide is based on the use of al-ready existing soil databases (as laboratory-measured datausing surveyed soil samples), and conventional and analogi-cal soil cartography itself is not necessarily required. DSMthus typically consists in creating soil information combin-ing point-based data (from field survey and laboratory analy-sis) and mostly exhaustive auxiliary information obtained atcheaper costs with models of inference in the spatial, tem-poral and/or attribute domains. According to the domain ofinference, Carre et al. (2007) distinguished the DSM sensustricto (DSMss) which is involved in the creation of soil in-formation in the space domain, from the DSM sensu lato(DSMsl) which can generate derived soil attributes from theoutputs of DSMss by mean of attribute domain inference sys-tems (e.g. soil water retention capacity).

Products are commonly assessed for accuracy and uncer-tainty (Oliver, 2010). Accuracy is a key aspect of DSM



procedures, also considering the linked structure of some in-ference systems (think to the cascade modelling using theDSMss and the DSMsl in sequence). The global accuracyof a DSM product depends upon the accuracy of the wholeset of soil data (localization, sampling, measurements, etc.),the auxiliary covariates and the inference systems used. Asan example of a DSM procedure a regression kriging model(Odeh et al., 1995) is here described in more detail (seeFig. 2.2 in Dobos et al., 2006):

1. There are soil profile observations that constitute thesoil database of the area of interest (observation points).

2. Primary soil attributes are measured at these locationsand the major inquiry is the knowledge of primary soilattributes at locations where no observations are avail-able (unknown points).

3. The SCORPAN framework suggests that any soil prop-erty (target variable) is quantitatively related to othersoil properties (point covariates) and to exhaustive aux-iliary information (gridded covariates) pertaining to soilforming factors.

4. Considering the sources of gridded covariates(e.g. probe sensing and digital terrain analysis)and hence the amount of forthcoming variables, thefirst step consists in reducing the dimensionality ofthe input space by a factor analysis (or a principalcomponent analysis). The resulting unobserved factors(or components) are queried at observation points tobuild the matching table (gridded to point covariatesare added to pre-existent point covariates), which isused for calibrating the regression kriging model.

5. Regression kriging can be thought of as a two compo-nent model plus an error term.

a. At first stage the deterministic component is solvedby a classical multilinear regression model (i.e. afixed effects model with a single error term; Sch-abenberger and Pierce, 2002).

b. The stochastic component is engaged at secondstage, in which the target variable residuals fromregression are analysed for spatial auto and crosscorrelation by means of the main explorative toolin geostatistics, the semivariogram. The so calledmodel of (co)regionalization is fulfilled by fittingallowed mathematical functions to experimentalsemivariograms, and then the (co)kriging system ismostly solvable.

c. Regression and geostatistical predictions at un-known gridded points are additively combined inorder to obtain the regression kriging predictions(gridded predictions).

6. A regression kriging map of the target soil attribute isfulfilled. It can be considered an output itself or can beused in cascade modelling as an input in order to addressfunctional properties of soils.

Appendix B

Glossary of soil terms (relevant for this specificpaper)

The following terms are given to elucidate definitions forthose unfamiliar with the soil survey and soil mapping fields.The soil survey glossary refers to terms given in FAO (2006)and in Soil Science Society of America (2003). The soilmapping definitions are partly extracted from the report ofthe Digital Soil Mapping Working Group (Dobos et al.,2006). Terms are in alphabetical order.

B1 Soil survey (field description): basic concepts anddefinitions

– Andic features:(from Japanese An, dark, and Do, soil) it refers to hori-zons resulting from moderate weathering of mainly py-roclastic deposits and dominated either by short-range-order clay minerals and/or and Al/Fe-humus complexes.

Andic features may be found both at the surface and inthe subsurface. Surface andic horizon generally containa high amount of organic matter (more than 5 %), arevery dark coloured have a fluffy macrostructure. Bothsurface and subsurface horizons exhibit smeary consis-tence, thixotropy, low bulk density, very high water re-tention and have silt loam or finer textures.

Since analytical difficulties implied in identifyingthe occurrence of short-range-order clay mineralsand Al/Fe-humus complexes the following chemi-cal/physical test are considered diagnostic: (i) bulk den-sity of the soil at field capacity (no prior drying) ofless than 0.9 kg dm−3; (ii) acid oxalate extractable alu-minium and iron (named Alox + 1/2 Feox) must be ei-ther larger than 2.0 % or between 0.4–2.0 % but in thiscase the occurrence other soil features (e.g. estimate ofvolcanic glass) are required; (iii) phosphate retentionof 70 % or more; (iv) volcanic glass content in the fineearth fraction and (v) thickness of at least 30 cm.

– Biological features:Biological features, such as krotovinas, termite burrows,insect nests, worm casts and burrows of larger animals,are described in terms of abundance and kind. In ad-dition, specific locations, patterns, size, composition orany other characteristic may be recorded.



– Carbonates:the presence of calcium carbonate (CaCO3) is estab-lished by adding some drops of 10 % HCl to the soil.The degree of effervescence of carbon dioxide gas is in-dicative for the amount of calcium carbonate presence.

– Coating:Layer of a substance completely or partly covering asurface of soil material; coatings can comprise clay, cal-cite, gypsum, iron, organic material, salt, etc. These fea-tures are described according to their abundance, con-trast, nature, form and location.

– Concentrations:are identifiable bodies within the soil that were formedby pedogenesis, including secondary enrichments, ce-mentations and reorientations (FAO, 2006). Some ofthese bodies are thin and sheet like; some are nearlyequidimensional; others have irregular shapes. Theymay contrast sharply with the surrounding material instrength, composition, or internal organization. Alter-natively, the differences from the surrounding materialmay be slight.

– Concretion:A cemented concentration of a chemical compound,such as calcium carbonate or iron oxide, that can be re-moved from the soil intact and that has crude internalsymmetry organized around a point, line, or plane.

– Drainage Classes (natural):Natural drainage class refers to the frequency and du-ration of wet periods under conditions similar to thoseunder which the soil developed. Alteration of the waterregime by man, either through drainage or irrigation, isnot a consideration unless the alterations have signifi-cantly changed the morphology of the soil. The classesfollow:

a. Excessively drainedWater is removed from the soil very rapidly. The oc-currence of internal free water commonly is very rare orvery deep. The soils are commonly coarse-textured andhave very high saturated hydraulic conductivity or arevery shallow.

b. Somewhat excessively drainedWater is removed from the soil rapidly. Internal freewater occurrence commonly is very rare or very deep.The soils are commonly coarse-textured and have highsaturated hydraulic conductivity or are very shallow.

c. Well drainedWater is removed from the soil readily but not rapidly.Internal free water occurrence commonly is deep orvery deep; annual duration is not specified. Water is

available to plants throughout most of the growing sea-son in humid regions. Wetness does not inhibit growthof roots for significant periods during most growing sea-sons. The soils are mainly free of the deep to redoximor-phic features that are related to wetness.

d. Moderately well drainedWater is removed from the soil somewhat slowly dur-ing some periods of the year. Internal free water oc-currence commonly is moderately deep and transitorythrough permanent. The soils are wet for only a shorttime within the rooting depth during the growing sea-son, but long enough that most mesophytic crops areaffected. They commonly have a moderately low orlower saturated hydraulic conductivity in a layer withinthe upper 1 m, periodically receive high rainfall, or both.

e. Somewhat poorly drainedWater is removed slowly so that the soil is wet at ashallow depth for significant periods during the grow-ing season. The occurrence of internal free water com-monly is shallow to moderately deep and transitory topermanent. Wetness markedly restricts the growth ofmesophytic crops, unless artificial drainage is provided.The soils commonly have one or more of the follow-ing characteristics: low or very low saturated hydraulicconductivity, a high water table, additional water fromseepage, or nearly continuous rainfall.

f. Poorly drainedWater is removed so slowly that the soil is wet at shal-low depths periodically during the growing season orremains wet for long periods. The occurrence of inter-nal free water is shallow or very shallow and commonor persistent. Free water is commonly at or near thesurface long enough during the growing season so thatmost mesophytic crops cannot be grown, unless the soilis artificially drained. The soil, however, is not contin-uously wet directly below ploughing depth. Free waterat shallow depth is usually present. This water table iscommonly the result of low or very low saturated hy-draulic conductivity of nearly continuous rainfall, or ofa combination of these.

g. Very poorly drainedWater is removed from the soil so slowly that free wa-ter remains at or very near the ground surface duringmuch of the growing season. The occurrence of internalfree water is very shallow and persistent or permanent.Unless the soil is artificially drained, most mesophyticcrops cannot be grown. The soils are commonly levelor depressed and frequently ponded. If rainfall is highor nearly continuous, slope gradients may be greater.

– Horizon:see soil horizon.



– Horizon Boundary:Horizon boundaries are described in terms of depth, dis-tinctness and topography. The depth of the upper andlower boundaries of each horizon is given in centime-tres, measured from the surface (including organic andmineral covers) of the soil downwards.

– Inundation occurrence:A record of the month(s) during which the inundationoccurs may be useful. Maximum depth of the inunda-tion, as well as the flow velocity, may be helpful.

– Minipit:a quick (and low cost) small soil excavation aiming ei-ther/both to check the occurrence and sample specificsoil horizons.

– Mottling:Mottles are spots or blotches of different colours orshades of colour interspersed with the dominant colourof the soil. They indicate that the soil has been subjectto alternate wetting (reducing) and dry (oxidizing) con-ditions. Mottling of the soil matrix or groundmass isdescribed in terms of abundance, size, contrast, bound-ary and colour. In addition, the shape, position or anyother feature may be recorded.

– Porosity:The total volume of voids in a soil sample (nonsolid vol-ume) discernible with a× 10 hand-lens measured byarea and recorded as the percentage of the surface oc-cupied by pores. Voids are described in terms of type,size, abundance, continuity and orientation.

– Reaction, soil:The degree of acidity or alkalinity of a soil, usually ex-pressed as a pH value. Descriptive terms commonly as-sociated with certain ranges in pH.

– Rock fragments:Unattached pieces of rock 2 mm in diameter or largerthose are strongly cemented or more resistant to rup-ture. Rock fragments are described by size, shape, and,for some, the kind of rock. The classes are pebbles,cobbles, channers, flagstones, stones, and boulders.

– Slickensides:Stress surfaces that are polished and striated producedby one mass sliding past another. Slickensides are com-mon below 50 cm in swelling clays subject to largechanges in water content.

– Soil colour (matrix):soil colour is one of the indicators of soil status and de-pends on many factors. The colour of the soil matrix ofeach horizon should be recorded in the moist condition(or both dry and moist conditions where possible) usingthe notations for hue, value and chroma as given in the

Munsell Soil Color Charts (Munsell, 1975). For exam-ple: 10 YR 6/4 is a colour (of soil) with a hue = 10 YR,value = 6, and chroma = 4.

– Soil horizon:A layer of soil or soil material approximately parallelto the land surface and differing from adjacent geneti-cally related layers in physical, chemical, and biolog-ical properties or characteristics such as colour, struc-ture, texture, consistency, kinds and number of organ-isms present, degree of acidity or alkalinity, etc. Soilhorizons include the following designation:

– O horizonsLayers dominated by organic material.

– A horizonsMineral horizons that formed at the surface character-ized by an accumulation of humified organic matter orhave properties resulting from cultivation, pasturing, orsimilar kinds of disturbance.

– E horizonsMineral horizons in which the main feature is loss ofsilicate clay, iron, aluminium, or some combination ofthese, leaving a concentration of sand and silt particlesof quartz or other resistant materials.

– B horizonsHorizons that formed below an A, E, or O horizon andare dominated by obliteration of all or much of the orig-inal rock structure and show one or more of the follow-ing (a) illuvial concentration of silicate clay, iron, alu-minium, humus, carbonates, gypsum, or silica, alone orin combination; (b) evidence of removal of carbonates;(c) residual concentration of sesquioxides; (d) coatingsof sesquioxides; (e) alteration that forms silicate clay orliberates oxides or both and that forms granular, blocky,or prismatic structure;

– C horizonsLayers, excluding hard bedrock, that are little affectedby pedogenic processes and eventually.

– R layersHard bedrock.

– Soil profile:Vertical section of soil horizons from upper layer to theparent material, showing the arrangement (configura-tion) of soil horizons typical for single soil types andused as a basis for soil classification.

– Soil structure:The combination or arrangement of primary soil parti-cles into secondary units or peds. The secondary unitsare characterized on the basis of size, shape (platy, pris-matic, columnar, angular, subangular, blocky, granular,



etc.), and grade (degree of distinctness: single-grain,massive, weak, moderate, and strong).

– Soil texture(of the fine earth fraction): numerical proportion (% bywt.) of sand, silt and clay in a soil (for particles lessthan 2 mm). Sand, silt and clay content are estimatedin the field, and/or quantitatively in the laboratory, andthen placed within the texture triangle to determine soiltexture class. Texture can be coarse (sand particles pre-dominate), medium (silt particles predominate), or fine(clay particles predominate).

B2 Soil mapping: basic concepts and definitions

– Collocated soil attributes:Any covariate available at the same observation pointsthan the target soil attribute. Typically these are othersoil attributes used to predict the target soil attribute.

– Covariates:All the variables related to the target soil attribute andpertaining to the SCORPAN factors. These variable canbe collocated soil attributes (e.g. other soil attributes) orcan be exhaustive auxiliary information (e.g. derivativesof a digital elevation model or a geophysical spectrum).

– Digital Soil Map:Visualization of a georeferenced soil database, whichshows spatial distribution of soil types and/or soil prop-erties; digital soil map can also be a digitized existingsoil maps.

– Digital Soil Mapping:It is the computer-assisted production of digital mapsof soil type and soil properties. It typically implies useof mathematical and statistical models that combine in-formation from soil observations with information con-tained in correlated environmental variables and remotesensing images.

– Exhaustive auxiliary information (or exhaustive ancil-lary data):The set of covariates derived from probe scanning, dig-ital terrain analysis or other digital/digitized maps. Thisset of data is exhaustive because of its gridded nature(generally the gridded points are not perfectly collo-cated with the target soil attribute recalling for a datamigration procedure). The term auxiliary refers to theaid this set provides within any soil inference system inproducing a digital soil map.

– Functional maps:Visualisation of soil database (a complex document) us-able in its current form to any further application, due toits complex description of how it was derived, what ac-curacy does it have (metadata), how to interpret, what it

can be used for; maps easy to use for practical purposes;multifunctional maps.

– Secondary soil properties:These are properties derived from primary soil proper-ties using various inference models (pedotransfer rulesand environmental models).

– Soil classification:Soils are named and classified on the basis of physi-cal and chemical properties in their horizons (layers).Soil classification schemes typically use colour, texture,structure, and other properties of the surface two metersdeep to key the soil into a classification system to helppeople use soil information (modified after Soil SurveyManual).

– Soil functions:Various ecologic and socio-economic roles of soils, asdefined in the COM179 (2002) regulation; the most im-portant soil functions are (a) soil biomass productivity,(b) organic carbon fixation, (c) groundwater protection,(d) support for raw material, (e) biodiversity, and (f) nat-ural heritage.

– Soil mapping:It is the process of mapping soil types or other proper-ties over a landscape. It relies heavily on distinguishingthe individual influences of the five classic soil formingfactors. This effort draws upon geomorphology, physi-cal geography, and analysis of vegetation and land-usepatterns.

Soil observations:Measured and observed data available from original soilsurvey.

– Soil spatial inference:A procedure or a set of procedures implementing a soil-landscape model also known as the “SCORPAN” modelused to derive soil properties or classes using availablesoil and auxiliary information.

– Soil survey:Describes the characteristics of the soils in a given area,classifies the soils according to a standard system ofclassification, plots the boundaries of the soils on a map,and makes predictions about the behaviour of soils. Thedifferent uses of the soils and how the response of man-agement affects them are considered. The informationcollected in a soil survey helps in the development ofland-use plans and evaluates and predicts the effects ofland use on the environment (Soil survey manual).

– Spatially predicted soil properties/classes:Interpolated soil properties or classes that are now avail-able at each location in the area of interest. This is theoutput from the soil spatial inference system.



– Target soil attribute:Any soil property/class analysed within a soil spatial in-ference framework to make a digital soil map.

Appendix C

Hydrological models

C1 ALMAGRA model – MicroLEIS DSS (De La Rosaet al., 2004)

This model is part of a more general agro-ecological landevaluation Decision Support System (DSS), which is the Mi-croLEIS DSS. This DSS is based on the multifunctional eval-uation of soil quality, using input data collected in standardsoil surveys.

It aims to define site-specific sustainable agricultural prac-tices and to point out the importance of using soil informa-tion in decision-making regarding the environmentally sus-tainable use and management of land.

Basically the ALMAGRA model uses a multiparametricapproach to determine classes of land productivity for a spe-cific crop. The land parameters correspond to the followingthree main factors: soil/site, climate and crop/management.

The employed soil parameters are selected as importantsoil indicators for specific soil functions such as the crop pro-ductivity. These parameters include functional depth, stoni-ness, texture, water retention, reaction, carbonates, salinity,and cation exchange capacity.

On the base of values of these parameters in the differentsoil mapping units (as given in the soil database), the Mi-croLEIS system classifies – according to the most limitingfactor and through a matching table – suitability classes forthe chosen crop. It also implements a specific module formaize crop.

The model can be freely downloaded athttp://www.microleis.com.

C2 Soil Water Atmosphere Plant (SWAP) model (vanDam et al., 1997)

Applied version 2.07 of SWAP was designed for simulationof water flow, solute transport, and plant growth in the soil-water-atmosphere-plant continuum. To calculate soil waterflow SWAP, using an implicit finite difference scheme, solvesthe 1-D Richards’ equation for soil water movement in thesoil matrix extended by the sink term,S:

∂θ(h)

∂t=

∂

∂z

[K(h)

(∂h

∂z+ 1

)]− S(h) (C1)

whereθ (cm3 cm−3) is the volumetric soil water content,h(cm) is the soil water pressure head,t (d) is the time,z (cm) isthe vertical coordinate taken positively upward,K (cm d−1)is the hydraulic conductivity, andS (cm3 cm−3 d−1) is the

water extraction rate by the plant roots. SWAP applies theequation both for the unsaturated and the saturated zones.The hydraulic conductivity is calculated as the arithmeticmean of the conductivities in the adjacent nodes. This hydro-logical model permits the definition of up to five soil layerswith different physical properties. Additionally, the soil issubdivided into a maximum of 40 compartments in whichthe soil water flow is calculated with a daily time step.

Soil water retention was described by the unimodalθ(h)

relationship proposed by van Genuchten (1980) and ex-pressed here in terms of the effective saturation,Se, asfollows:

Se =

[1

1 + (α|h|)n

]m(C2)

with Se = (θ−θr)/(θ0−θr), θr , andθ0 being the residual wa-ter content ath=−∞ and the water content ath= 0, respec-tively, and in whichα (cm−1), n, andm are curve-fittingparameters.

Mualem’s expression was applied to calculate the rela-tive hydraulic conductivity,Kr (Mualem, 1976). Assumingm= 1− 1/n, van Genuchten (1980) obtained a closed formanalytical solution to predictKr at specified volumetric wa-ter content:

Kr(Se) =K(Se)

K0= Sτe

[1 −

(1 − S

1/me

)m]2(C3)

in whichK0 is the hydraulic conductivity atθ0, andτ is a pa-rameter which accounts for the dependence of the tortuosityand partial correlation between pores of the same diameter.The condition at the bottom boundary can be set in severalways (e.g. pressure head, water table depth, fluxes, imper-meable layer, unit gradient, etc.).

SWAP simulated the water uptake and actual transpirationaccording to the model proposed by Feddes et al. (1978),where root water uptakeS is described as a function of thepressure head,h:

S(h) = α(h) Smax = α(h) Tp/zr (C4)

beingSmax (d−1) the maximal possible root water uptake un-der optimal soil water condition,Tp (cm d−1) is the potentialtranspiration rate,zr (cm) the thickness of the root zone, andα(h) an empirical function of pressure headh, varying be-tween 0 and 1. The shape of the functionα(h) depends onfour critical values ofh, which are related to crop/vegetationtype and to potential transpiration rates. Integration ofS overthe root layer yields the actual transpiration rateTa (cm d−1).If the simpler crop model routine is chosen, the root depth,the leaf area index (LAI), and the crop coefficient (KC) arespecified by the user as a function of the crop/plant develop-ment stage.

C3 CropSyst (Stockle et al., 2003)

CropSyst is a multi-year multi-crop daily time step simula-tion model. The model has been developed to serve as an


http://www.microleis.com

http://www.microleis.com


analytic tool to study the effect of cropping systems man-agement on productivity and the environment. The modelsimulates the soil water budget, soil-plant nitrogen budget,crop canopy and root growth, dry matter production, yield,residue production and decomposition, and erosion. Anal-ogy to Swap model, also CropSyst calculates the soil waterflow solving the Richards’ equation for soil water flow in thesoil matrix by an implicit finite difference scheme.

In CropSyst the soil hydraulic functions are described bythe analytical expressions of Campbell (1985). The soil wa-ter retention function is

h = hb for h ≥ hb

h = hb (θ/θ0)−λ for h < hb (C5)

wherehb is the air entry water potential (potential at whichthe largest water filled pores just drain), andλ is the slope ofln(h) vs. ln(θ). The hydraulic conductivity is described by

K(h) = K0(hb

/h

)(2+3/λ). (C6)

The condition at the bottom boundary can be set in severalways (e.g. pressure head, water table depth, fluxes, imper-meable layer, unit gradient, etc.). The number of soil lay-ers (horizons) can be selected by the user until to 11; thenCropSyst subdivides the layers automatically into sublayersof approximately 10 cm thickness.

In CropSyst model each layer water uptake is calculated asa function of (i) the difference between soil and xylem waterpotential and (ii) root conductance (Stockle et al., 1992). Thesoil conductance is assumed to be higher than root conduc-tance so water movement towards the roots does not limit thatwater uptake. The water uptake, WUi (kg m−2 d−1), fromeach soil layeri is given by:

WUi = 86 400Ci/

1.5 (ψsi − ψl) (C7)

whereψsi (J kg−1) is the soil water potential of soil layeri(Campbell, 1985),ψl (J kg−1) is the leaf water potential,Ci(kg s m−4) is the roots conductance of soil layeri, 86 400 isthe number of seconds per day and 1.5 is a factor that con-verts total root conductance to total plant hydraulic conduc-tance. The total water uptake WU is the sum of the wateruptake from each soil layer.

The crop growth is simulated for the whole canopy by cal-culating unstressed biomass growth as the minimum of twovalues of daily aboveground biomass rate. In fact, such rateis calculated as function of potential transpiration and of in-tercepted radiation. Unstressed biomass growth value is thencorrected by water and nitrogen limitations to simulate actualdaily biomass accumulation. The root growth is synchro-nized with leaf area growth (Stockle et al., 2003). The waterstress reduces biomass accumulation (and consequently LAIand roots development) proportionally to the actual to poten-tial evapotranspiration ratio. The user gives the maximumvalue of root depth as input and the root density is assumed

to decrease linearly with depth (Campbell and Diaz, 1988)with a maximum at the top of the soil profile and a value ofzero at the tip of the current root depth.

C4 USBR model for irrigability classes(http://www.fao.org/docrep)

Sophisticated methods of land classification for irrigatedagriculture were first evolved by the United States Bureauof Reclamation in the 1920s and 1930s.

The USBR classification system incorporates broad eco-nomic considerations from the start. This is important be-cause irrigation projects generally involve costly inputs andimprovements such as engineering works, irrigation anddrainage networks, land clearing and levelling, and others.

The USBR Reclamation Manual (1951) and subsequentReclamation Instructions lists the following principles of theUSBR classification system:

1. Prediction: the classification should reflect future condi-tions as they will exist after the project is implemented.This recognizes that changes will occur in relationshipsbetween soils, water and crops as a result of irrigationand land improvements and that the classifier should usethe classes to indicate whether these changes are likelyto be favourable or unfavourable.

2. Economic correlation: this assumes that a unique rela-tionship can be established during a classification, be-tween physical conditions of the land such as soils, to-pography and drainage and an economic measure of theclass ranges. The measure used is payment capacity,i.e. the residual available to defray the cost of water af-ter all other costs have been met by the farmers.

3. Permanent and changeable factors: the classifier mustdistinguish between permanent factors, such as soil tex-ture, soil depth, macro relief, etc., and changeable fac-tors, such as salinity, ESP, pH, micro relief, nutrient sta-tus, water table levels, etc. Thus the survey and clas-sification are directed to determining which inputs andimprovements to changeable factors are cost effective.

4. Arability-irrigability: land which is physically and eco-nomically capable of providing a farmer with an ade-quate standard of living, should water be available forirrigation is first classified. Such land is called “arable”(connoting a different meaning of the word to that incommon usage). Arable lands constitute areas that war-rant consideration for inclusion in a plan of develop-ment. Lands which are selected for inclusion in theplan of development are called “irrigable” lands. Thisdual stage procedure is copied in this publication inthe successive classification of “provisionally-irrigable”and “irrigable” land.


http://www.fao.org/docrep


The selection of lands for irrigation is phased into two parts:

– the selection of arable land on the basis of farm produc-tion financial considerations;

– selection of the irrigable area on the basis of the eco-nomics of the project plan, wherein irrigation bene-fits determined by economic evaluation equal or exceedproject irrigation costs.

Six land classes based on production economics are normallyrecognized. Brief descriptions are as follows:

1. Arable: lands that are highly suitable for irrigated farm-ing, being capable of sustained and relatively high yieldof climatically adapted crops at reasonable cost. Theselands have a relatively high payment capacity.

2. Arable: lands that have a moderate suitability for irri-gated farming. These are either adaptable to a narrowerrange of crops, more expensive to develop for irrigation,or less productive than Class 1. Potentially these landshave intermediate payment capacity.

3. Arable: lands that have a marginal suitability for irri-gated farming. They are less suitable than Class 2 landsand usually have either a serious single deficiency or acombination of several moderate deficiencies in soil, to-pography, or drainage properties. Although greater riskmay be involved in farming these lands than those ofClass 1 and 2, under proper management they are ex-pected to have adequate payment capacity.

4. Special use lands: lands that in the USA are only suitedto certain special uses (e.g. rice, pasture, or fruit) areclassified 1, 2 or 3 (to reflect relative payment capacity)along with the appropriate letter designating the landuse (crop).

5. Non-arable: this land is temporarily considered as non-arable because of some specific deficiency such asexcessive salinity, questionable drainage, flooding, orother deficiency which requires further studies to re-solve. The deficiency or deficiencies are of such a na-ture and magnitude that special agronomic, economic,or engineering studies are required to resolve the costsor effect on the land. Class 5 designation is tentative andshould be changed to either Class 6 or an arable classi-fication during formulation of the recommended plan ofdevelopment.

6. Non-arable: land that is non-arable under the existingor project economic conditions associated with the pro-posed project development. Generally, Class 6 com-prises steep, rough, broken, rocky, or badly erodedlands, or lands with inadequate drainage, or other de-ficiencies. In some instances lands considered to be

Class 6 in one area may be arable in another area be-cause of different economic conditions. In addition tovarious physical-type deficiencies that result in a non-arable classification, lands initially classified as arable(potentially irrigable) on the basis of payment capac-ity (farm financial analysis) may be found non-arableif subsequent economic analysis (benefit analysis) indi-cates that benefits from such lands are less than theircosts in a plan of development. Thus, the lower arableclass(es) of lands would be considered non-arable and,of course, non-irrigable for economic reasons.

Lower case letters that indicates the reason for the land be-ing downgraded to a lower class indicates subclasses. Thus,Class 1 land does not have subclasses, but other classes maybe appended with the letters “s”, “ t”, and “d”, singly or incombination to show whether the deficiency is in “soils”, “to-pography” or “farm drainage”. The basic subclasses of theland classes ares, t , d, st, sd, td and std.

The model developed by Arangino et al. (1986) is a mod-ification of this well known USBR model for adapting itat the Italian situation. It was adopted in many Italian ir-rigation projects of last 3 decades funded by Cassa per ilMezzogiorno.

The model classifies 4 irrigation suitability classes on thebase of a multiparametric scheme including soil, topographyand drainage features.

Soil features include texture, depth, stoniness, rockiness,permeability, mineral weathering, salinity and carbonates.Topographic features include slope angle and potential riskof erosion.

On the base of values of these parameters in the differentsoil mapping units an irrigation suitability map is produced.

C5 TOPKAPI (Liu et al., 2005)

TOPKAPI is a physically based, fully distributed rainfall-runoff model, which is based on the lumping of a kine-matic wave assumption in the soil, at the surface, and inthe drainage network and leads to transforming the rainfall-runoff and runoff routing processes into three nonlinearreservoir differential equations. Initially, the TOPKAPImodel (Ciarapica and Todini, 2002; Liu and Todini, 2002)was structured around five modules that represent the evapo-transpiration, snowmelt, soil water, surface water, and chan-nel water. In a more recent work (Liu et al., 2005), theground water component also has been developed. Many ap-plications of the model have been implemented, especiallyfor flood forecasting purposes (Martina and Entekhabi, 2006;Todini, 2007; Vischel et al., 2008).

The TOPKAPI model is based on the idea of combiningthe kinematic approach with the topography of the basin;the latter is described by a digital elevation model (DEM)that subdivides the application domain by means of a grid ofsquare cells whose size generally increases with the overall



dimensions because of the constraints imposed by the com-puting resources. Consequently, increasing the applicationscale of the model implies an increase in the dimensions ofthe cells, which at catchment scale may amount to severalhundred metres per side. Each cell of the DEM is assigneda value for each of the physical characteristics represented inthe model. The flow paths and slopes are evaluated, startingfrom the DEM, according to a neighbourhood relationshipbased on the principle of the minimum energy cost, namelythe maximum elevation difference. It considers the links be-tween the active cell and the four surrounding cells connectedalong the edges from the finite difference approach underpin-ning the model; the active cell is assumed to be connecteddownstream with a sole cell, while it can receive upstreamcontributions from more than one cell (up to three). The in-tegration in space of the nonlinear kinematic wave equationsresults in three “structurally similar”, zero-dimensional, non-linear reservoir equations. The first represents the drainagein the soil, the second represents the overland flow on satu-rated or impervious soils, and the third represents the channelflow.

The TOPKAPI model proposed is structured around threebasic modules that represent, in turn, the soil water compo-nent, the surface water component, and the channel watercomponent (drainage network component). The deep aquiferflow gives no significant response to storm events. The soilwater component is affected by a flow of water (interflow orsubsurface flow) in a horizontal direction in conditions of nosaturation, defined as drainage; drainage occurs in a surfacesoil layer, of limited thickness, and with high hydraulic con-ductivity because of its macroporosity. The drainage mech-anism plays a fundamental role in the model both as a directcontribution to the flow in the drainage network and, most ofall, as a factor regulating the soil water balance, particularlywith regard to activating the production of overland flow (To-dini, 1996). The soil water component is the most character-ising aspect of the model because it regulates the functioningof the contributing saturated areas. The surface water com-ponent is activated on the basis of this mechanism. Lastly,both components contribute to feed the drainage network.

The other important aspect which is strongly related to theintegration of the governing equations is the consequent un-avoidable parameters (and variables) averaging. The ques-tion is whether the average parameters at the model scale aresufficient to model the dominant physical processes or theinner parameters space-distribution is necessary for a correctrepresentation. That would imply, for instance, a model gridrefinement up to the scale at which the space-distribution canbe neglected.

In the context of rainfall-runoff modelling, we are mainlyinterested to the water flow both into the soil and over thesurface; it is worth then to test the influence of the param-eters averaging on these processes. It is convenient also todistinguish the vertical from the horizontal average.

Soil hydraulic behaviour is characterized by the soil wa-ter retention curve, which defines the water content (θ ) asa function of the capillary pressure head (ψ), and the hy-draulic conductivity function, which establishes relationshipbetween the hydraulic conductivity (K) and water contentor capillary pressure head. Simulations of unsaturated flowand solute transport typically use closed-form functions torepresent water-retention characteristics and unsaturated hy-draulic conductivities. Some of the commonly used func-tional relationships include the Brooks-Corey model, andthe van Genuchten model. Both models can be used tocompute the hydraulic transmissivity (T ) of a soil layerin non-saturated condition which is given by the followingexpression:

T =

L∫0

k (θ(z)) · dz (C8)

wherez is the vertical direction andL is the thickness of thelayer affected by the horizontal flow. The water flow can becalculated, assuming a kinematic wave approximation, fromthe actual soil moisture profile as:

q =

L∫0

tan (β) k (ϑ(z)) dz (C9)

whereβ is the topographic surface slope assumed to be equalto the water table slope andq is the water flow per unit soilwidth [m2 s−1].

As it can be seen from Eq. (C9) the horizontal water flowdepends on the particular soil moisture profile since thereis not a linear relationship between unsaturated conductiv-ity and volumetric water volume. However, due to the highconductivity value, caused by macropores in the top of thesoil, gravity will be the dominant mechanism driving wa-ter from the top of the soil to the bottom (impermeable orsemi-impermeable lower boundary). In this zone within therange of reasonable errors, the integration of the unsaturatedsoil vertical infiltration equation (namely Richards’ equation)and use only the average values for hydraulic conductivityand water content to compute the transmissivity (and conse-quently the water flow). As a matter of fact in the TOPKAPImodel the transmissivity is given by:

T = L · ks 2α (C10)

whereks is the saturated hydraulic conductivity,α is a pa-rameter dependent on the soil characteristics and2 is meanwater content along the vertical profile i.e.:

2 =1

L

L∫0

θ (z) dz. (C11)

The feasibility of this assumption has been validated by anumeric experiment where the transmissivity computed con-sidering the vertical profile of the water content (Eq. C8) is



Table C1.Principal parameters required for each grid cell in the new TOPKAPI model.

Model Component Parameter

Interception Vegetation interception capacity (Sr0)Vegetation storage capacity (SC0)Maximum leaf-area-index (LAI0)

Evapotranspiration Daily maximum sunshine hours (n)Monthly average air temperature (Ta)Crop factors (Kc)

Snowmelt Critical air temperature for determining the precipitation as snow or rain,Ts

Infiltration Infiltration coefficient depending on the land cover type (Kl )

Interflow and percolation Thickness of the upper soil layer (L)Horizontal and vertical saturated hydraulic conductivity of the upper soil (ksh1andksv1)

Effective soil moisture content of the upper soil (ϑe=ϑs−ϑr)Field capacity (ϑf ) of the upper soilExponent of the transmissivity law for the upper soil (αs)Exponent of the percolation law for the upper soil (αp)Vertical saturated hydraulic conductivity of the lower soil (ksv2)

Groundwater recharge Exponent of the vertical groundwater recharge equation for the lower soil,αr

Groundwater flow Horizontal saturated hydraulic conductivity of the lower soil (ksh2)Impermeable bedrock depth from the surface (d) and slope (Sb)Field capacity (ϑf ) of the lower soil

Surface flow Surface roughness (no)Surface slope, tang (β)

Channel flow Roughness for the channel according to the Strahler channel order (nc)Maximum and Minimum Channel widths (Wmax,Wmin)River bed slope (S0)

Lake/reservoir routing Routing curve

compared with that computed by considering only the meanvalue (Eq. C11).

However although it is physically based, the model stillneeds calibration because of the uncertainty of the informa-tion on the topography, soil characteristics and land cover. InTable C1 the main parameters required for each grid cell aregiven.

C6 HYDRUS 1-D (Simunek et al., 2008)

HYDRUS computer code numerically solves the Richardsequation (Eq. C1) for variably-saturated water flow andadvection-dispersion type equations for heat and solute trans-port. The flow equation incorporates a sink term to accountfor water uptake by plant roots (Eq. C4). The flow equa-tion may also consider dual-porosity type flow in which onefraction of the water content is mobile and another fractionimmobile, or dual-permeability type flow involving two mo-bile regions, one representing the matrix and one the macro-pores. The heat transport equation considers transport dueto conduction and convection with flowing water. Coupled

water, vapour and energy transport can be considered aswell. The solute transport equations consider advective-dispersive transport in the liquid phase, as well as diffu-sion in the gaseous phase. The transport equations also in-clude provisions for nonlinear nonequilibrium reactions be-tween the solid and liquid phases, linear equilibrium re-actions between the liquid and gaseous phases, zero-orderproduction, and two first-order degradation reactions: onewhich is independent of other solutes, and one which pro-vides the coupling between solutes involved in sequentialfirst-order decay reactions. Physical nonequilibrium solutetransport can be accounted for by assuming a two-region,dual-porosity type formulation which partitions the liquidphase into separate mobile and immobile regions.

The program may be used to analyse water and solutemovement in unsaturated, partially saturated, or fully sat-urated porous media. The flow region may be composedof nonuniform soils. Flow and transport can occur in thevertical, horizontal, or a generally inclined direction. Thewater flow part of the model can deal with prescribed headand flux boundaries, boundaries controlled by atmospheric



conditions, as well as free drainage boundary conditions. Thegoverning flow and transport equations are solved numeri-cally using Galerkin-type linear finite element schemes.

HYDRUS-1-D also includes a parameter optimization al-gorithm for inverse estimation of soil hydraulic and/or so-lute transport and reaction parameters from measured tran-sient or steady-state flow and/or transport data. Inverse meth-ods are typically based upon the minimization of a suitableobjective function, which expresses the discrepancy betweenthe observed values and the predicted system response. Soilhydraulic properties for this purpose are assumed to be de-scribed by an analytical model with unknown parameter val-ues. Initial estimates of the optimized system parameters areiteratively improved during the minimization process until adesired degree of precision is obtained

Minimization of the objective function is accomplishedby using the Levenberg-Marquardt nonlinear minimizationmethod that combines the Newton and steepest descendmethods, and generates confidence intervals for the opti-mized parameters.

Acknowledgements.The work was partially funded by the ItalianMinistry of Education and Research under the project “CUBIST –Relations between hydrological processes, climate, and physicalattributes of the landscape at the regional and basin scales” (Grant2007HBTS85).

Edited by: P. Claps

References

Aarts, E. H. L. and Korst, J.: Simulated Annealing and BoltzmannMachine, Wiley & Sons, Chichester, England, 1989.

Arangino, F., Aru, A., Baldacchini, P., and Vacca, S.: I suolidelle aree irrigabili della Sardegna, in: Regione Autonoma dellaSardegna, Piano Generale delle Acque Cagliari, 1986.

Arya, L. M.: Wind and hot-air methods, in: Methods of soil analy-sis, Part 4, Physical methods, SSSA Book Ser. 5, Madison, WI,916–926, 2002.

Arya, L. M. and Paris, J. F.: A physic empirical model to predictthe soil moisture characteristic from particle-size distribution andbulk density data, Soil Sci. Soc. Am. J., 45, 1023–1030, 1981.

Aubertin, G. M.: Nature and extent of macropores on forest soilsand their influence on subsurface water movement, USDA, For-est Serv. Res. Pap. NE-192, USDA, Washington, DC, 1971.

Bartoli, F., Regalado, C., Basile, A., Buurman, P., and Coppola,A.: Physical properties in European volcanic soils: a synthesis,in: Soils of Volcanic Regions of Europe, edited by: Arnalds,O.,Bartoli, F., Buurman, P., Oskarsson, H., Stoops, G., and Garcia-Rodeja, E., ISBN: 10:3-540-48710-7, 515–537, 2007.

Basile, A. and D’Urso, G.: Experimental corrections of simpli-fied methods for predicting water retention curves in clay-loamysoils from particle-size determination, Soil Technol., 10, 261–272, 1997.

Basile, A., De Mascellis, R., and Terribile, F.: Il suolo e la pro-tezione degli acquiferi: studio pedologico e idrologico dei suolidella piana del F. Sarno (Campania), Quaderni di Geologia Ap-plicata, 1, 251–261, 1999.

Bastiaanssen, W. G. M., Allen, R. G., Droogers, P., D’Urso, G., andSteduto, P.: Twenty-five years modeling irrigated and drainedsoils: State of the art, Agr. Water Manage., 92, 111–125, 2007.

Beven, K.: Changing ideas in hydrology – The case of physically-based models, J. Hydrol., 105, 157–172, 1989.

Bilonick, R. A.: Monthly hydrogen ion deposition maps for thenortheastern U.S. from July 1982 to September 1984, Atmos.Environ., 22, 1909–1924, 1988.

Bittelli, M. and Flury, M.: Errors in water retention curves deter-mined with pressure plates, Soil Sci. Soc. Am. J., 73, 1453–1460,2009.

Bolt, G. H.: Soil Chemistry, B, Physico-Chemical Models, Elsevier,527 pp., 1982.

Bonfante, A., Basile, A., Acutis, M., De Mascellis, R., Manna,P., Perego, A., and Terribile, F.: SWAP, CropSyst and MACROcomparison in two contrasting soils cropped with maize in North-ern Italy, Agr. Water Manage., 97, 1051–1062, 2010.

Bouma, J.: Soil morphology and preferential flow along macrop-ores, Agr. Water Manage., 3, 235–250, 1981.

Bouma, J.: Using soil survey data for quantitative land evaluation,Adv. Soil Sci., 9, 177–213, 1989.

Bouma, J.: Hydropedology as a powerful tool for environmentalpolicy research, Geoderma, 131, 275–286, 2006.

Bouma, J., Fox, C. A., and Miedema, R.: Micromorphology of Hy-dromorphic soils, Applications for soil genesis and land evalu-ation, in: Douglas, LA Soil Micromorphology: a basic and ap-plied science, Development in Soil Science, 19, Elsevier Ams-terdam, 257–278, 1990.

Bouma, J., de Vos, J. A., Sonneveld, M. P. W., Heuvelink, G. B. M.,and Stoorvogel, J. J.: The role of scientists in multiscale land useanalysis: lessons learned from Dutch communities of practice,Adv. Agron., 97, 175–237, 2008.

Campbell, G.: Transport Models for Soil-Plant Systems, Elsevier,Amsterdam, The Netherlands, 1985.

Campbell, G. S. and Diaz, R.: Simplified Soil-Water Balance Mod-els to Predict Crop Transpiration, ICRISAT, Patancheru, India,1988.

Carre, F., McBratney, A. B., Mayr, T., and Montanarella, L.: Digitalsoil assessments: Beyond DSM, Geoderma, 142, 69–79, 2007.

Ciarapica, L. and Todini, E.: TOPKAPI: a model for the represen-tation of the rainfall-runoff process at different scales, Hydrol.Process., 16, 207–229, 2002.

Commission of the European Communities: Towards a ThematicStrategy for Soil Protection, 179 final, Brussels, 2002.

Coppola, A., Basile, A., Comegna, A., and Lamaddalena, N.:Monte Carlo analysis of field water flow comparing uni- and bi-modal effective hydraulic parameters for structured soil, J. Con-tam. Hydrol., 104, 153–165, 2009.

D’Urso, G. and Basile, A.: Physico-empirical approach for map-ping soil hydraulic behaviour, Hydrol. Earth Syst. Sci., 1, 915–923,doi:10.5194/hess-1-915-1997, 1997.

D’Urso, G. and Minacapilli, M.: A semi-empirical approach forsurface soil water content estimation from radar data without a-priori information on surface roughness, J. Hydrol., 321, 297–310, 2006.


http://dx.doi.org/10.5194/hess-1-915-1997


Dane, J. H. and Hopmans, J. W.: Water retention and storage, in:Methods of soil analysis: Part 4-Physical methods, SSSA BookSer. 5, edited by: Dane, J. K. and Topp, G. C., SSSA, Madison,WI, USA, 671–720, 2002.

de Bruin, S. and Stein, A.: Soil-landscape modelling using fuzzyc-means clustering of attribute data derived from a Digital Ele-vation Model (DEM), Geoderma, 83, 17–33, 1998.

De Cesare, L., Myers, D. E., and Posa, D.; Estimating and mod-elling space-time correlation structures, Stat. Probab. Lett., 51,9–14, 2001a.

De Cesare, L., Myers, D. E., and Posa, D.: Product-sum covariancefor space-time modelling: an environmental application, Envi-ronmetrics, 12, 11–23, 2001b.

De la Rosa, D., Mayol, F., Diaz-Pereira, E., Fernandez, M., and dela Rosa Jr., D.: A land evaluation decision support system (Mi-croLEIS DSS) for agricultural soil protection with special refer-ence to the Mediterranean region, Environ. Modell. Softw., 19,929–942, 2004.

Dent, D. and Young, A.: Soil survey and Land evaluation, GeorgeAllen & Unwin, London, UK, 278 pp., 1981.

Dobos, E., Carre, F., Hengl, T., Reuter, H. I., and Toth, G.: Digi-tal Soil Mapping as a support to production of functional maps,EUR 22123 EN, Office for Official Publications of the EuropeanCommunities, Luxemburg, 68 pp., 2006.

Doll, P., Kaspar, F., and Lehner, B.: A global hydrological modelfor deriving water availability indicators: model tuning and vali-dation, J. Hydrol., 270, 105–134, 2003.

Dokuchaev, V. V.: Russian Chernozems (Russkii Chernozem), Is-rael Prog. Sci. Trans. Jerusalem, 1967, Transl. from Russian by:Kaner, N., available from U.S. Dept. of Commerce, Springfield,Va, 1883.

Dunne, T.: Field studies of hillslope flow processes, in: HillslopeHydrology, edited by: Kirkby, M. J., Wiley, New York, 227–293,1978.

Durand, N., Monger, H. C., and Canti, M. G.: Calcium carbonatefeatures, in: Interpretation of Micromorphological Features ofSoils and Regoliths, edited by: Stoops, G., Marcelino, V., andMees, F., Elsevier, Amsterdam, 149–194, 2010.

FAO: A Framework for Land Evaluation, Soil Bull., 32, FAO,Rome, Italy, 1976.

FAO: Guidelines for Soil Profile Description, 4, FAO, Rome, Italy,2006.

Feddes, R. A., Kowalik, P. J., and Zaradny, H.: Simulation of fieldwater use and crop yield, Simulation Monographs, Pudoc, Wa-geningen, 195 pp., 1978.

Grayson, R. and Bloschl, G.: Spatial Patterns in Catchment Hydrol-ogy: Observations and Modelling, Cambridge University Press,404 pp., 2000.

Grunwald, S.: Multi-criteria characterization of recent digital soilmapping and modeling approaches, Geoderma, 152, 195–207,2009.

Herrero, J., Robinson, D. A., and Nogues, J.: A regional soil surveyapproach for upgrading from flood to sprinkler irrigation in asemi-arid environment, Agr. Water Manage., 93, 145–152, 2007.

Heuvelink, G. B. M. and Pebesma, E. J.: Spatial aggregation andsoil process modeling, Geoderma, 89, 47–65, 1999.

Hseu, Z. Y. and Chen, Z. S.: Saturation, reduction and redox mor-phology of seasonally flooded Alfisols in Taiwan, Soi Sci. Soc.Am. J., 60, 941–949, 1996.

Jenny, H.: Factors of soil formation, McGraw-Hill, New York,USA, 1941.

Jhorar, R. K., van Dam, J. C., Bastiaanssen, W. G. M., and Fed-des, R. A.: Calibration of effective soil hydraulic parameters ofheterogeneous soil profiles, J. Hydrol., 285, 233–247, 2004.

Jones, A., Montanarella, L., and Jones, R.: Soil atlas of Europe,European Soil Bureau Network, European Commission, Ispra,128 pp., 2005.

Kuhn, P., Aguilar, J., and Miedema, R.: Textural features and re-lated horizons, in: Interpretation of Micromorphological Fea-tures of Soils and Regoliths, edited by: Stoops, G., Marcelino,V., and Mees, F., Elsevier, Amsterdam, 217–250, 2010.

Lambert, J. J., Daroussin, J., Eimberck, M., Le Bas, C., Jamagne,M., King, D., and Montanarella, L.: Soil Geographical Databasefor Eurasia & The Mediterranean: Instructions Guide for Elabo-ration at scale 1:1,000,000, Version 4.0, Office for Official Pub-lications of the European Communities, Luxembourg, 64 pp.,2003.

Lark, R. M. and Bishop, T. F. A.: Cokriging particle size fractionsof the soil, Eur. J. Soil Sci., 58, 763–774, 2007.

Lin, H.: Earth’s Critical Zone and hydropedology: concepts, char-acteristics, and advances, Hydrol. Earth Syst. Sci., 14, 25–45,doi:10.5194/hess-14-25-2010, 2010.

Lin, H.: Hydropedology: Towards new insights into interactivepedologic and hydrologic processes across scales, J. Hydrol.,406, 141–145, 2011.

Lin, H., Bouma, J., Pachepsky, Y., Western, A., Thompson, J., vanGenuchten, R., Vogel, H. J., and Lilly, A.: Hydropedology: Syn-ergistic integration of pedology and hydrology, Water Resour.Res., 42, W05301,doi:10.1029/2005WR004085, 2006.

Lin, H. S., McInnes, K. J., Wilding, L. P., and Hallmark, C. T.:Effects of Soil Morphology on Hydraulic Properties: I. Quantifi-cation of Soil Morphology, Soil Sci. Soc. Am. J., 63, 948–954,1999.

Lin, H. S., Brooks, E., McDaniel, P., and Boll, J.: Hydropedologyand Surface/Subsurface Runoff Processes, in: Encyclopedia ofHydrologic Sciences, edited by: Anderson, M. G. and Mcdon-nell, J. J., John Wiley & Sons, Chichester, 1–25, 2008.

Lindbo, D. L., Stolt, M. H., and Vepraskas, M. J.: Redoximorphicfeatures, in: Interpretation of Micromorphological Features ofSoils and Regoliths, edited by: Stoops, G., Marcelino, V., andMees, F., Elsevier, Amsterdam, 129–147, 2010.

Liu, Z. and Todini, E.: Towards a comprehensive physically-basedrainfall-runoff model, Hydrol. Earth Syst. Sci., 6, 859–881,doi:10.5194/hess-6-859-2002, 2002.

Liu, Z., Martina, M. L. V., and Todini, E.: Flood forecasting usinga fully distributed model: application of the TOPKAPI model tothe Upper Xixian Catchment, Hydrol. Earth Syst. Sci., 9, 347–364,doi:10.5194/hess-9-347-2005, 2005.

Lyon, S. W., Seibert, J., Lembo, A. J., Walter, M. T., and Steenhuis,T. S.: Geostatistical investigation into the temporal evolution ofspatial structure in a shallow water table, Hydrol. Earth Syst. Sci.,10, 113–125,doi:10.5194/hess-10-113-2006, 2006.

Manna, P., Basile, A., Bonfante, A., De Mascellis, R., and Terribile,F.: Comparative Land Evaluation approaches: An itinerary fromFAO framework to simulation modelling, Geoderma, 150, 367–378, 2009.



http://dx.doi.org/10.1029/2005WR004085





Martina, M. L. V. and Entekhabi, D.: Identification ofrunoff generation spatial distribution using conventional hy-drologic gauge time series, Water Resour. Res., 42, W08431,doi:10.1029/2005WR004783, 2006.

McBratney, A. B., Mendoca Santos, M. L., and Minasny, B.: Ondigital soil mapping, Geoderma, 117, 3–52, 2003.

McDonnell, J. J., Sivapalan, M., Vache, K., Dunn, S., Grant, G.,Haggerty, R., Hinz, C., Hooper, R., Kirchner, J., Roderick, M.L., Selker, J., and Weiler, M.: Moving beyond heterogeneityand process complexity: a new vision for watershed hydrology,Water Resour. Res., 43, W07301,doi:10.1029/2006WR005467,2007.

McKenzie, N. J. and Ryan, P. J.: Spatial prediction of soil propertiesusing environmental correlation, Geoderma, 89, 67–94, 1999.

McKenzie, N. J., Grundy, M. J., Webster, R., and Ringroase-Voase,A. J.: Guidelines for surveying soil and land resources, CSIROPublishing, Melbourne, Australia, 557 pp., 2008.

Minasny, B., McBratney, A., and Salvador-Blanes, S.: Quantitativemodels for pedogenesis – A review, Geoderma, 144, 140–157,2008.

Mualem Y.: A new model for predicting the hydraulic conductivityof unsaturated porous media, Water Resour. Res., 12, 513–522,1976.

Munsell Color (Firm): Munsell soil color charts, Munsell ColorCompany, 2441 North Calvert Street, Baltimore, MD 21218,1975.

Nanzyo, M., Shoji, S., and Dahlgren, R.: Physical characteristicsof volcanic ash soils, in: Volcanic Ash Soils: Genesis, Propertiesand Utilization, editd by: Shoji, S., Nanzyo, M., and Dahlgren,R., Development in Soil Science, vol. 17, Elsevier, Amsterdam,189–201, 1993.

Odeh, I. O. A., McBratney, A. B., and Chittleborough, D. J.: Fur-ther results on prediction of soil properties from terrain attributes:heterotopic cokriging and regression-kriging, Geoderma, 67,215–226, 1995.

Oliver, M. A.: Geostatistical Applications for Precision Agriculture,Springer, London, UK, 295 pp., 2010.

Pachepsky, Y. A. and Rawls, W. R.: Development of PedotransferFunctions in Soil Hydrology, Elsevier, New York, 2004.

Park, S. J. and Vlek, P. L. G.: Environmental correlation of three-dimensional soil spatial variability: a comparison of three adap-tive techniques, Geoderma, 109, 117–140, 2002.

Poch, R. M., Artieda, O., Herrero, J., and Lebedeva-Verba, M.:Gypsic features in soils and sediments, in: Interpretation of Mi-cromorphological Features of Soils and Regoliths, edited by:Stoops, G., Marcelino, V., and Mees, F., Elsevier, Amsterdam,195–216, 2010.

Rabenhorst, M. C., Bell, J. C., and McDaniel, P. A.: Quantifyingsoil hydromorphology, Soil Science Society of America SpecialPublication 54, SSA, Madison, Wisconsin, USA, p.258, 1998.

Rawls, W. J.: Estimating soil bulk density from particle size analy-sis and organic matter content, Soil Sci., 136, 48–55, 1983.

Reynolds, W., Elrick, D., Youngs, E. G., Amoozegar, A., Bootlink,H. W. G., and Bouma, J.: Laboratory methods, in: Methods ofSoil Analysis, Part 4, Physical Methods, SSSA Book Ser. 5, Soil.Sci. Soc. Am., Madison, WI, 802–816, 2002.

Ritsema, C. J., van Dam, J. C., Dekker, L. W., and Oostindie,K.: A new modelling approach to simulate preferential flow andtransport in water repellent porous media: model structure and

validation, Aust. J. Soil Res., 43, 361–369, 2005.Rossel, R. A. V., McBratney, A. B., and Minasny, B.: Proximal

Soil Sensing, Series: Progress in Soil Science, Vol. 1, 1st Edn.,468 pp., 2010.

Rossiter, D. G.: Digital soil resource inventories: status andprospects, Soil Use Manage., 20, 296–301, 2004.

Rouse, J. W., Haas, R. H., Schell, J. A., Deering, D. W., and Har-lan, J. C.: Monitoring the vernal advancement of retrograda-tion of natural vegetation, NASA/GSFC type II, Progress ReportRSC 1978-1, Greenbelt, MD, 371 pp., 1973.

Salvador-Blanes, S., Minasny, B., and McBratney, A. B.: Modellinglong-term in situ soil profile evolution: application to the gene-sis of soil profiles containing stone layers, Eur. J. Soil Sci., 58,1535–1548, 2007.

Samouelian, A. and Cornu, S.: Modelling the formation and evolu-tion of soils, towards an initial synthesis, Geoderma, 145, 401–409, 2008.

Schabenberger, O. and Pierce, F. J.: Contemporary Statistical Mod-els for the Plant and Soil Sciences, CRC Press, Boca Raton,Florida, USA, 738 pp., 2002.

Simunek, J., van Genuchten, M. Th., Gribb, M. M., and Hopmans,J. W.: Parameter estimation of unsaturated soil hydraulic prop-erties from transient flow processes, Soil Till. Res., 47, 27–36,1998.

Simunek, J.,Sejna, M., Saito, H., Sakai, M., and van Genuchten, M.Th.: The Hydrus-1D Software Package for Simulating the Move-ment of Water, Heat, and Multiple Solutes in Variably SaturatedMedia, Version 4.0, HYDRUS Software Series 3, Departmentof Environmental Sciences, University of California Riverside,Riverside, California, USA, 315 pp., 2008.

Snepvangers, J. J. J. C., Heuvelink, G. B. M., and Huisman, J. A.:Soil water content interpolation using spatio-temporal krigingwith external drift, Geoderma, 112, 253–271, 2003.

Soil Science Society of America: Glossary of Soil Science Terms,USDA Soil Survey Manual, 2003.

Soil Survey Division Staff: Soil survey manual, Soil ConservationService, U.S. Department of Agriculture, Handbook 18, 1993.

Stockle, C. O., Martin, S., and Cambell, G.: A model to assess envi-ronmental impact of cropping systems, Amer. Soc. of Agr. Eng.,92-2041, 1992.

Stockle, C. O., Donatelli, M., and Nelson, R.: CropSyst, a croppingsystems simulation model, Eur. J. Agron., 18, 289–307, 2003.

Stolpe, N. B., Kuzila, M. S., and Shea, P. J.: Importance of Soil MapDetail in Predicting Pesticide Mobility in Terrace Soils, Soil Sci.,163, 394–403, 1998.

Stoops, G. and Eswaran, H.: Morphological characteristics of wetsoils, in Wetland soils: Characterization, classification and uti-lization, IRRI, Los Banos, Philippines, 177–189, 1985.

Todini, E.: The ARNO rainfall-runoff model, J. Hydrol., 175, 339–382, 1996.

Todini, E.: Hydrological catchment modelling: past, present andfuture, Hydrol. Earth Syst. Sci., 11, 468–482,doi:10.5194/hess-11-468-2007, 2007.

Tucker, C. J. and Sellers, P. J.: Satellite remote sensing of primaryproduction, Int. J. Remote Sens., 7, 1395–1416, 1986.

USBR – Bureau of Reclamation Manual: Vol. V, Irrigated land use,Part 2, Land classification, Bureau of Reclamation, Dept. of In-terior, Denver Federal Center, Denver, Col. 80225, USA, 1951.


http://dx.doi.org/10.1029/2005WR004783

http://dx.doi.org/10.1029/2006WR005467




USBR – Bureau of Reclamation: Irrigated Land Use, Land Clas-sification, V, Bureau of Reclamation, Dept. of Interior, DenverFederal Center, Denver, 1981.

USDA – Soil Survey Staff: Keys to Soil Taxonomy, 11th Edn.,USDA-Natural Resources Conservation Service, Washington,DC, 2010.

van Dam, J. C., Huygen, J., Wesseling, J. G., Feddes, R. A., Ka-bat, P., van Walsum, P., Groenendijk, P., and van Diepen, C. A.:Theory of SWAP version 2.0, Dept. of Water Resources, WAU,Wageningen, Netherlands, Rep. 71, 168 pp., 1997.

Van Genuchten, M. Th.: A closed-form equation for predicting thehydraulic conductivity of unsaturated soils, Soil Sci. Soc. Am. J.,44, 892–898, 1980.

Vepraskas, M. J.: Redoximorphic Features for Identifying AquicConditions, Technical Bulletin, 301, North Carolina AgriculturalResearch Service, Raleigh, NC, USA, p.33, 2004.

Vereecken, H., Maes, J., Feyen, J., and Darius, P.: Estimating thesoil moisture retention characteristic from texture, bulk densityand carbon content, Soil Sci., 148, 389–403, 1989.

Verhagen, A., Booltink, H. W. G., and Bouma, J.: Site-specificManagement: Balancing Production Environmental Require-ments at Farm Level, Agr. Syst., 49, 369–384, 1995.

Vischel, T., Pegram, G., Sinclair, S., and Parak, M.: Implementationof the TOPKAPI model in South Africa: Initial results from theLiebenbergsvlei catchment, Water Sa., 34, 331–342, 2008.

Wagener, T., Sivapalan, M., Troch, P., and Woods, R.: Catchmentclassification and hydrologic similarity, Geogr. Compass, 1, 901–931, 2007.

Wang, J., Rich, P. M., Price, K. P., and Kettle, W. D.: Relationsbetween NDVI and tree productivity in the central great plains,Int. J. Remote Sens., 25, 3127–3138, 2004.

Warner, G. S. and Young, R. A.: Measurement of preferential flowbeneath mature corn, in: Preferential flow, Proc. of the NationalSymp, 16–17 December 1991, edited by: Grish, T. J. and Shir-mohammadi, A., Chicago, IL. ASAE, St. Joseph, MI, 1991.

Western, A. W., Grayson, R. B., Bloschl, G., Willgoose, G., andMcMahon, T. A.: Observed spatial organisation of soil moistureand its relation to terrain indices, Water Resour. Res., 35, 797–810, 1999.

Wilson, J. P. and Gallant, J. C.: Terrain analysis. Principles andapplications, John Wiley & Sons, New York, 2000.

Wosten, J. H. M., Bannik, M. H., De Gruijter, J. J., and Bouma,J.: A procedure to identify different groups of hydraulic-conductivity and moisture-retention curves for soil horizons, J.Hydrol., 86, 133–145, 1986.

Wosten, J. H. M., Lilly, A., Nemes, A., and Le Bas, C.: Us-ing existing soil data to derive hydraulic parameters for sim-ulation models in environmental studies and in land use plan-ning – Final report on the European union funded project 1998,DLO Winand Staring Centre, Wageningen, The Netherlands,Rep. 156, 106 pp., 1998.

WRB – IUSS Working Group: World reference base for soil re-sources 2006 – A framework for international classification, cor-relation and communication, 2, World Soil Resources ReportsNo. 103, FAO, Rome, 2006.

Yaalon, D. H. and Bekowicz, S.: History of Soil Science, in: Ad-vances in Geoecology, 29, Catena Verlag GMBH, 438 pp., 1997.

Yeh, T. C. J.: One-Dimensional Steady-State Infiltration in Hetero-geneous Soils, Water Resour. Res., 25, 2149–2158, 1989.

Yong, R. N. and Warkentin, B. P.: Soil properties and behavior,Elsevier, The Netherlands, 449 pp., 1975.

Zhu, Q., Lin, H. S., Zhou, X., Doolittle, J., and Zhang, J.: PrecisionSoil Mapping for Landscape Hydrology and Crop Managementin an Agricultural Landscape, Annual meetings of ASA-CSSA-SSSA, New Orleans, LA, 2007.


Potential and limitations of using soil mapping ... · Soil plays a key role in hydrology since its...

Documents

Transcript of Potential and limitations of using soil mapping ... · Soil plays a key role in hydrology since its...