geomorphological methods to characterise wetlands at the ... · the agreement between the two...

ment 375 (2007) 59–68www.elsevier.com/locate/scitotenv

Science of the Total Environ

Geomorphological methods to characterise wetlands at thescale of the Seine watershed

F. Curie a,⁎, S. Gaillard b, A. Ducharne a, H. Bendjoudi a

a Laboratoire Sisyphe, CNRS/Université Pierre et Marie Curie, Paris, Franceb Laboratoire IGARUN, Université de Nantes, France

Available online 29 January 2007

Abstract

Based on easily available morphological data within the Seine river watershed (76750 km2), two approaches were used forwetland delineation and characterisation. Their common assumption is that geomorphology largely governs the spatial distributionof wetlands, because it determines topography and the nature of deposits, thus water pathways and residence times. The firstapproach relies on the topographic index introduced by Beven and Kirkby [Beven KJ, Kirkby MJ. A physically based, variablecontributing area model of basin hydrology. Hydrol Sci Bull 1979; 24: 43–69.], that has been widely used to characterise saturatedareas in small catchments. We mapped this index for the Seine watershed using a 100 m resolution DEM typical of DEMs easilyavailable at this scale. The second approach relies on a geomorphological classification of river corridors which was specificallydeveloped for the Seine basin. It is based on genetic concepts, and defines 13 types of river corridors as a function of the geometryof the river bed with respect to bedrock (incised, aggraded, encased, stable), the nature of alluvial fills, and the small scalemorphology in the corridors. We used geological, hydrogeological and topographical maps of the Seine basin to delineate the rivercorridors and characterise the type of all the comprising streams with 2 km resolution.

Two cartographic sources that were not exploited by the above methods were used to assess their performances. The wetlandsdepicted on 1:25000 topographic maps cover 2% of the Seine basin but are limited. The waterlogged soils from two 1:50000pedologic maps are more reliable, but these maps only cover 5% of the watershed. In the river corridors, most wetlands fall in theencased and aggraded subsystems of the geomorphological classification, where the mean of the topographic index is significantlyhigher than in the other subsystems. High values of the topographic index are good general indicators of wetlands, even whencalculated from a 100-m DEM. The agreement between the two studied methods confirms that geomorphology is the major drivingfactor for wetland distribution, even in a sedimentary basin with a strong influence of aquifers on hydrology. These complementarymethods provide a powerful tool to complement the gaps of classical wetland databases at the scale of large watersheds.© 2007 Elsevier B.V. All rights reserved.

Keywords: Wetlands; River corridors; Geomorphology; Topographic index; Seine; Basin; Waterlogged soils

1. Introduction

Wetlands are special ecosystems able to achievemany environmental functions regarding biodiversity

⁎ Corresponding author. Laboratoire Sisyphe, CNRS/UniversitéPierre et Marie Curie, case 105, 4 place Jussieu, Paris, 75005, France.

E-mail address: [email protected] (F. Curie).

0048-9697/$ - see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.scitotenv.2006.12.013

(Fustec et al., 1998) and river functioning. Theirlocation at the interface between terrestrial and aquaticenvironments governs a wide range of buffering actionslike flooding control (Burt, 1997; Oberlin, 2000),sediment traps and sources (Dillaha and Inamdar,1997) and water quality improvement, in particular bythe retention of phosphorus and nitrogen, the mainnutrients responsible for eutrophication (Haycock et al.,

mailto:[email protected]

http://dx.doi.org/10.1016/j.scitotenv.2006.12.013

60 F. Curie et al. / Science of the Total Environment 375 (2007) 59–68

1997). These buffering functions explain the attentiondevoted to wetlands and their preservation, when theycover globally only 6% of land area (Lefeuvre et al.,2000). In the Seine river basin (France), our study area,as in most of the large watersheds in Europe, the spatialextent of wetlands is poorly known, apart for the mostobvious ones, such as swamps (e.g. swamps of SaintGond), alluvial plains or those presenting an ecologicalinterest and surveyed by the Natura 2000 program of theEuropean Union. Wetlands of much smaller extent arestill poorly delineated.

The general objective of this study is to contribute towetland delineation from data easily available at thescale of large watersheds as the Seine river basin. Thisobjective answers a need from stake-holders aiming atpreserving the numerous impacts of wetlands at thecatchment scale. The specific aim of this paper is tocompare two approaches for delineating wetland in theSeine river watershed which is a large sedimentary basin(Section 2). These two approaches rely on geomorphol-ogy, under the assumption that it largely governs thespatial distribution of wetlands in the landscape becauseit determines the topography and the nature of deposits.

Fig. 1. Topography of the Seine watershed from a 100 m resolution DEM provof Tonnerre and Saint Dizier. The study sites mentioned in this paper are ou

The first approach assumes that topography is thedriving force for water movement. We used the topo-graphic index (Section 3) developed in the hydrologicmodel TOPMODEL (Beven and Kirkby, 1979). Theseauthors showed that the spatial distribution of thesaturated areas in a catchment can be described from thistopographic index. To compute this index, one onlyneeds a digital elevation model (DEM). Given the wideavailability of DEM over the recent years, this methodpresents the advantage to be easy and possible over largeareas compared to remote-sensing or field surveys. Forthe first time, we evaluated the feasibility of thisapproach over the Seine river basin by using a 100 mresolution DEM typical of DEMs easily available at thisscale. The second approach makes use of a geomor-phological classification of fluvial corridors (Gaillard etal., 2001), which was specifically developed in the Seinebasin, in the framework of the PIREN-Seine multidis-ciplinary research programme (Section 4). This classi-fication is based on genetic and dynamic concepts. Thevarious classes of this classification correspond totheoretical differences in hydrologic functioning. Twoindependent cartographic sources (Section 5) were used

ided by Geosys. The rectangular boxes correspond to pedological mapstlined.

61F. Curie et al. / Science of the Total Environment 375 (2007) 59–68

to assess the skills of these two approaches to delineatewetlands (Section 6).

2. Study area

The Seine catchment (Fig. 1), in northern France,covers 76750 km2, approximately 12% of the country.The main river, the Seine, is 780 km long. The slopes aregentle: the median slope angle is 2.2° at the pixel scaleand the maximum slope angle is 30° in the Massif ofMorvan at south west of the basin. The altitude range is0–856 m above sea level but 89.5% of the basin is below300 m. The geological structure of the Paris basinresembles a stack of saucers with the most recent layersoutcropping in the centre and the oldest layers outcrop-ping on the outer edges of the basin. This structurecontains numerous aquifers of varying size and structure,about ten of them being very important in terms of waterresource. The mean annual effective rainfall varies from430 mm in the middle of the catchment to 990 mm in thesouth-eastern part of the basin (Morvan). Land use isdominated by crops, which cover 60% of the Seinecatchment according to the Corine land cover database(Bossard et al., 2000). Because of the strong humaninfluence, marked deterioration of natural wetlands hasoccurred. The disappearance of wetlands has not beenprecisely evaluated, but numerous big swamps as theSaint Gond swamp were reclaimed (Fustec et al., 1998).

Table 1Summary of studies using the topographic index as an indicator for wetland

Authors Location Catchmentsize (km2)

Cell size(km2)

Mra

Merot et al.(1995)

Crac'h (Brittany) 54 40

Kervijen (Brittany) 44 40 1

Curmi et al.(1998)

Coët Dan (Brittany) 12 30

Rhode and Seibert(1999)

Nasten (Sweden) 6.6 50 and 200

Kassjoan (Sweden) 164 50 and 200

Korsiberget (Sweden) 4.2 50 and 200

Hemberget (Sweden) 3.7 50 and 200

Merot et al.(2003)

France 6.3 50Netherland 10.6 10Poland 0.88 10

Spain 35.5 20Switzerland 16 25 1United Kingdom 0.84 20

Only works including comparison with validation data are presented.

3. Topographic index

3.1. Theoretical background: considering the importanceof wetland mapping

An important way to predict the spatial distribution ofareas with high water content is to use the watershedgeomorphology because topography is the main factorthat determines water pathways. Cappus (1960) proposedthe variable source area concept according to which thesub-catchment areas do not contribute equally to runoff.This concept was later developed by Hewlett (1961),Hewlett and Troendle (1975), Ward (1982), Burt andButcher (1985). The relationship between these contrib-uting areas and topography was formalised in thehydrologic model TOPMODEL (Beven and Kirkby,1979; Beven, 1986). This model is based on theassumption that the hydraulic gradient of the shallowwater table is equal to the local topographic slope angle. Asecond assumption states that the water table variationscan be assimilated to a succession of steady states withuniform recharge. This allows the authors to relate thelocal depth of the water table to a soil topographic index:

lnða=T tanbÞ ð1Þ

‘a’ is the drainage area per unit contour length, b thelocal slope and T the transmissivity. In most cases,

s

ean annualinfall (mm)

Topography Geology

700 Gentle slopes mean slope=0.8°max slope=11.6° alt 6.6 to 53.4 m

Granite

050 Medium slopes mean slope=3.1°max slope=18.4° alt 1.7 to 242 m

Brioverian shale

713 Medium slopes alt 65 to 136 m Brioverian shale

660 Gentle slopes mean slope=0.03°alt 18 to 55 m

Granite


Granite doleritegneiss


Metamorphicandesite


Fractured granite

730 Mean alt 95 m Granite761 Mean alt=55 m Tertiary clays601 Mean alt=116 m Sandy clay

moraine434 Mean alt=80 m Granite100 Mean alt=80 m Moraine karst800 Mean alt=90 m Moraine


transmissivity variations are negligible compared to slopeand drainage area variations. The resulting index is calledthe topographic index:

lnða=tanbÞ: ð2Þ

If the drainage area is large and the local slope isweak, the water quantity is important and cannot beeasily evacuated. Therefore, the value of this indexreflects the potential soil saturation.

3.2. Previous studies

Many authors used the topographic index as an index ofsaturation. Only a few established a comparison withvalidation data (Table 1). These studieswere performed onsmall watersheds with sizes varying from 0.6 to 164 km2.They are generally located in metamorphic terrain. TheDEM spatial resolution is generally high with cell sizevalues from 10 to 50 m. Merot et al. (1995) attempted topredict the extent of waterlogged soils by using the topo-graphic index, in two Brittany catchments in a metamor-

Fig. 2. Topographic index for the Seine watershed from a 100 m resolutipedological maps of Tonnerre and Saint Dizier. The study sites mentioned i

phic area. By comparing the topographic index maps withmaps derived from a 1:25000 soil survey, they demon-strated the suitability of the topographic index to determinethe intensity of waterlogging soils. By fixing a thresholdvalue, they showed that a catchment can be divided in twozones: a saturated area with high values of the topographicindex and an unsaturated one. The value of this thresholddepends of the DEM grid size and the study area. Theextent of waterlogged soils may not be the same for twosimilar catchments having the same topography butdifferent rainfall rates. The topographic index wasimproved by taking into account climate conditions.Merot et al. (2003) proposed a new index where thedrainage area a is multiplied by the mean annual effectiverainfall depth Pe (the part of the rainfall that is notevapotranspired). This climato-topographic index

lnða⁎Pe=tanbÞ ð3Þ

allows them to compare catchments with different rainfallamounts. This indexwas validated by comparing six studyareas representing a wide range of climatic (annual

on DEM provided by Geosys. The rectangular boxes correspond ton this paper are outlined.


effective rainfall varying from 67 to 592 mm), geologicand geomorphologic settings.

Rhode and Seibert (1999) showed that the topo-graphic index allows to estimate the position andextension of saturated areas that are not connected tothe hydrographic network, such as depressions inmoraine landscape. Curmi et al. (1998) established arelationship between soil hydromorphic classes andtopographic index. Each class of soil hydromorphycorresponds to a range of topographic index values.They demonstrated that the hydromorphy of topograph-ic origin is better estimated than the hydromorphy oflithologic origin because the topographic index takesinto account geomorphologic characteristics of basin butdoes not consider geological layers.

3.3. Construction of the topographic index map

Here, the topographic index for the Seine watershed(Fig. 2) was computed, from a 100 m resolution DEMprovided by Geosys. The DEM consist of 7673159 pix-els. The mean of the topographic index is 9.89 with astandard deviation of 2.43. The maximum is equal to27.36 and the minimum to 3.19.

4. Geomorphological classification

4.1. Principle

Gaillard et al. (2001) proposed a typologicalinventory of the stream corridors in the Seine basinusing a geormorphologic classification system. Thisclassification is based on genetic and dynamic conceptsrelated to the evolution of the hydrosystems over the last15000 years, and to their current functioning. There arefour levels to the classification. The first level ofclassification corresponds to the present delineation ofthe stream corridors as defined by the limit of the recentalluvium. The second level describes the connectionsbetween the current streams and the gravel deposits thatset up during the Weichselian Stage (120000 to10000 year BP). According to these stratigraphicrelationships, four subsystems are defined:

1. Incised subsystem: the fluvial erosion during the Lateglacial period (i.e. 15000 years BP) caused a linearincision in the gravel deposits and bedrock. Theincised subsystem is mostly found at the head of thebasin, corresponding to embanked, confined hydro-systems with steep slopes.

2. Aggraded subsystem: due to an extra load, fineHolocene sediments have aggraded over the gravel

deposits inherited from the Weichselian Stage. Thehydrosystems in that context develop large alluvialplains, mostly subhorizontal, with medium to steepslopes.

3. Encased subsystem: several erosion–sedimentationcycles occurred during the Late and Postglacialperiods. They gently incised the gravel deposits butdid not reach the bedrock. The encased subsystem ischaracterised by large alluvial plains which containhighly variable geomorphologic facies, with mediumto gentle longitudinal slopes.

4. Stable subsystem: the gravel deposits inherited fromthe Weichselian Stage were not modified neither byerosion nor by sedimentation.

The third level of the classification describes theHolocene alluvial deposits. Four types of alluvial fillsare considered: mineral deposits, organic deposits,mixed deposits (organic and mineral), and absence ofdeposits. The fourth level addresses the main hydro-geomorphologic facies that can be observed in thevalley floor: subhorizontal morphology, morphologywith natural levees and backswamps, depressions andhillocks. As a summary, this classification systemdescribes the river hydrosystems of the Seine basinand points out four subsystems at the second level, nineclasses at the third level, and 13 subclasses at the fourthlevel.

4.2. Construction of the geomorphological classificationmap

The typological inventory of the geomorphologicalclassification relies on a database describing the entireSeine watershed, with the exception of the Yonne sub-basin where the classification is not yet complete. Tocharacterise the different types of this classification weused: 1:50000 BRGM (Bureau de Recherche Géologi-que et Minière) geological maps of the Seine basin,1:25000 IGN (Institut Géographique National) topo-graphical maps and the 1:500000 hydrogeological mapof the Paris basin (Albinet, 1967). In this typologicalinventory, Gaillard et al. (2001) defined corridorelements of 2 km length. The precision of these datawas not sufficient to allow the identification and thedelineation of the different riverine wetlands that makeup the fluvial hydrosystems. The geomorphologicalclassification covers approximately 8% of the Seinewatershed. The second level of the geomorphologicalclassification with the four subsystems (incised,aggraded, encased and stable) in the Seine watershedis represented Fig. 3. Encased and aggraded subsystems

Fig. 3. Second level of the geomorphological classification representing the four subsystems defined by Gaillard et al. (2001) for all the sub-basins ofthe Seine watershed except the Yonne basin which is not available for the moment. The rectangular boxes correspond to pedological maps of Tonnerreand Saint Dizier. The study sites mentioned in this paper are outlined.


represent the major part of the corridors with respec-tively 38% and 42% of the total surface covered by thegeomorphological classification. The incised subsystemrepresents 15% of the corridors delineated and the stablesubsystem only 5%.

5. Validation method

The two approaches for delimiting wetlands (topo-graphic index and geomorphologic classification) werecompared with wetlands identified from two independentcartographic sources described below, and tested theconvergence of these approaches for delimiting wetlands.

5.1. Soil maps

The distribution of hydromorphic soils from high-resolution pedological maps is a very reliable way toidentify wetlands, and has been widely used to validatethe delineation of wetlands from topographic indexanalysis (Table 1). Two 1:50000 pedological maps of

INRA (Institut National de la Recherche Agronomique)are available in the Seine catchment, surrounding thecities of Tonnerre and Saint Dizier (Fig. 1). The fluvialdeposits and hydromorphic soils from these maps weredigitised and georeferenced, showing that 9.6% of theTonnerre map and 15.3% of the Saint Dizier map arecovered by waterlogged soils. These maps cover only 5%of the Seine watershed, and the 1:1000000 pedologicalmap covering all the France is not sufficiently detailed forour purpose. The soil types are defined according to theFrench Classification adopted by the Commission depédologie et de cartographie des sols (CPCS, 1967).

5.2. Topographical maps

We also estimated a wetland distribution from the sym-bols depictingmarshes on the 1:25000 IGN topographicalmaps, that are available in the entire Seine watershed. Thequality of this information probably depends on the datewhen these maps were last updated. These IGN wetlandscover 0.4% of the Seine watershed but the information

Fig. 4. Comparison of topographic indices and wetlands delineated from IGN maps: Swamp of Saint Gond (left) and Superbe river (right).


available is limited. This is shown by the comparisonbetween the IGNwetlands and thewaterlogged soils in theabove two pedological maps. There, only 0.6% of thewaterlogged soils correspond to IGN wetlands. Largealluvial floodplains, in particular, are poorly indicated byIGNmaps, as illustrated by Fig. 4, where the southern zonewith high topographic indices is the floodplain of the Seineriver, known as the Bassée.

6. Results and discussion

6.1. Comparison of topographic index maps andvalidation data

The first evaluation is a visual comparison betweenwaterlogged soils as indicated in the available high

Table 2Summary of the different types of waterlogged soils for the Tonnerre pedolo

Groupsof soils

Types of soils Mean oftopographicindex

TM SM

HS Hydromorphic soil 13.62 13.03AS Hydromorphic alluvial soil 12.16 12.47

Hydromorphic calcareous alluvial soil 11.61 12.36Humic calcareous alluvial soil 12.16Alluvial soil 12.71Calcareous alluvial soil 12.64 12.25Alluvial soil and calcareous alluvial soil 12.84

CS Hydromorphic colluvial soil 10.73 11.14Colluvial soil 10.97 10.54

OS Other types of soil 9.00 9.39

Soils are sorted in descending order of waterlogging intensity. In the colum«alluvial soils», CS to «colluvial soils» and OS to the «others soils».

resolution soil maps and topographical index. The highvalues of topographic index coincide with the occurrenceof waterlogged soils in the two pedological maps. Meansof topographic index in each class of waterlogged soilsfor the two pedological map allows to distinguish threegroups of soils (Table 2). The first group corresponds tohydromorphic soils, the second to alluvial soils and thethird to colluvial soils. The means of topographic indexin each group of soils are statistically significantaccording to the Wilcoxon rank test (significance levelof 0.01). These differences of means are determined bythe location of the soil types. Colluvial soils are located atthe bottom of hillslopes with stronger slopes than thoseof alluvial soils, located in the floodplains.

A threshold value of the topographic index on thetwo pedological maps was determined using the method

gical map (TM) and the Saint Dizier pedological maps (SM)

Standarddeviation oftopographicindex

Number of pixels Percentage recognisedfor a threshold value of

TM SM TM SM 11.5 TM 11.2 SM

3.25 2.96 3058 3004 82.5 80.93.11 3.08 479 6862 60.0 57.62.65 3.28 345 17791 46.7 69.93.25 1150 59.33.52 2911 70.43.96 4.03 4961 161 62.0 60.93.78 5838 66.22.10 2.36 1130 382 21.9 41.92.46 2.32 753 10502 32.3 29.91.89 1.98 194620 213636

n «groups of soils», HS corresponds to «hydromorphic soils», AS to

Table 3Characteristics of the four subsystems comprising the second level tothe geomorphological classification: number of 100-m pixel, mean andstandard deviation of topographic index (TI), percentage of wetlandsderived from IGN maps in the subsystems, within the entire Seinewatershed apart from the Yonne sub-basin; percentage of noncolluvialwaterlogged soil within the pedological map of Saint Dizier

Subsystems Numberpixels

MeanTI

TI standarddeviation

% IGNwetlands

% waterloggedsoils

Encased 199363 13.79 3.58 33.5 27.8Incised 222063 13.08 3.71 4.5 3.9Stable 79591 11.20 3.29 0.5 1.1Aggraded 26448 11.07 3.09 49.5 10.4Total 527465 12.96 3.70 88 43.2

All the differences in mean topographic index between the foursubsystems are statistically different according to the Student's test,with a significance level of 0.01.


of Merot et al. (1995). The threshold value wasincreased until the area covered by topographic indexvalues above the threshold value was equal to the areacovered by the waterlogged soils. Given the large scaleof pedological maps, the small streams in the topo-graphic index map were not considered so as not tounderestimate the threshold index. The threshold valuesare 11.5 for the Tonnerre map and 11.2 for Saint Diziermap. Table 2 shows the percentage of each soil typerecognised as wetlands for a threshold value oftopographic index equal to 11.5. Hydromorphic soilsare well recognised (82.5% to 80.9%). This efficiencyremains satisfactory for alluvial soils (46.7 to 70.4%)but drops to 21.9 to 41.9% for colluvial soils. Using thisthreshold value of 11.5, we estimate that 15.6% of theSeine watershed is covered by wetlands. This area islarger than the global estimation of 6% given byLefeuvre et al. (2000), and in the range of 1 to 30%found by Merot et al. (2003) in Brittany. It is importantto keep in mind that topographic indices inform aboutthe extent of potential wetlands. It does not take intoaccount the disappearance of wetlands because ofhuman influence.

The above estimation of the fraction of wetlands in theSeine catchment is also dependent on the soils consideredas wetlands to calibrate the threshold topographic index.The mean topographic index is significantly lower forcolluvial soils than for the other waterlogged soils, sugges-ting a lower potential for saturation: excluding the colluvialsoils in the calculation of the threshold value, the latterincreases to 12.5 for the Tonnerre map and 12.3 for theSaint Dizier map. The threshold of 12.5 defines that 10.9%of the Seine watershed is covered by wetlands. This newestimation agrees better with the qualitative knowledge ofwetlands extension in the Seine watershed, suggesting thatcolluvial soils are not waterlogged enough to be consideredas effective wetlands. A visual comparison was made bet-ween wetlands derived from IGN maps and topographicindex maps. This comparison was carried out on the entireSeine watershed. The results for two representative siteswere compared: the swamp of Saint Gond and the Superberiver (Fig. 4), which are listed as SCI (Site of CommunityInterest) by the Natura 2000 European network. As ex-pected, the IGN wetlands correspond to high topographicindices. The mean topographic index in these IGN wet-lands is 13.49 (with a standard deviation of 3.46). It is highbecause IGN wetlands are limited to obvious ones, suchas swamps. Topographic index analysis provides a morecomprehensive delineation of wetlands, as illustrated bythe floodplain of the Seine river in the southern part ofFig. 4, which is characterised by high values of the topo-graphic index but is not identified aswetlands in IGNmaps.

6.2. Comparison of geomorphological classificationand validation data

Table 3 displays the percentage of wetlands derivedfrom IGN maps inside the different subsystems of thegeomorphological classification in the entire Seine basinexcept the Yonne watershed. The results indicate that88% of IGN wetlands are included in the delimitation ofthe stream corridors. Aggraded and encased subsystemscontain more wetlands than do the incised and stablesubsystems. Within the Saint Dizier pedological map(Table 3), only 43.2% of the waterlogged soils areincluded in the stream corridor, because the geomor-phological classification has not yet been completed (itis completely lacking in the area covered by theTonnerre soil map, Fig. 3). The subsystems of thisclassification, however, are discriminant with respect towetland occurrence, and most waterlogged soils arelocated inside the aggraded and encased subsystems.Whatever the way to identify wetlands, they arepredominant in the encased and aggraded subsystems,which are therefore good indicators of wetland frequen-cy. This result was expected since these subsystems aredefined by features that are important for the develop-ment of numerous wetlands: large floodplains, gentlelongitudinal slopes and hydraulic annexes.

6.3. Comparison of topographic indices andgeomorphological classification

Initially, a visual comparison of topographic indexmaps with the first level of the geomorphologicalclassification was made. As in the alluvial plain of theMarne (Fig. 5), the high topographic indices are mainlylocated in stream corridors. The mean of topographic

Fig. 6. Frequency curves of the topographic index in the four types ofthe second level of the geomorphological classification.


index is 12.96 in these corridors, but it variessignificantly between the four subsystems comprisingthe second level of the classification (Table 3). Inparticular, the encased and aggraded subsystems have asignificantly higher mean topographic index than theother two subsystems. This confirms that these types aremore likely to include wetlands.

In the Marne watershed, a detailed analysis of thetopographic index distribution in the four subsystems ofthe geomorphological classification was undertaken(Fig. 6). The four distribution curves are bimodal. Thefirst peak is the same for all the curves and is locatednear 6.5. The second peak is close to 11 for the aggradedand encased subsystems and close to 14 for the stableand incised ones.

According to the geomorphological classification,encased and aggraded subsystems developed largefloodplains. In the corridors of these subsystems, onepixel corresponds to the river with a very high value ofthe topographic index and the other pixels to floodplainswith lower values than the river ones. The values oftopographic index of the river are high because theircontributing area is large. We observed three cases: (1)very high values of the topographic index in the corridorcorresponding to the river (wavelet at the end of thecurves), (2) high values of the topographic indexcorresponding to the floodplain (second peak) and (3)weak values of the topographic index corresponding byexample to hillock inside the corridor (first peak). Fig. 5displays an example of these small scale features, with acircular zone with low topographic index values in thefluvial corridor.

Stable and incised subsystems are characterised bynarrow valley with a width comparable to the cell size ofour DEM (100m), but they are not necessarily at the same

Fig. 5. Alluvial plain of the Marne river: comparison of topographicindex and the first level of the geomorphological classification.

place in the geomorphological classification and in the100-mDEMused to construct our map of the topographicindex. Two cases are possible: (1) the valleys in the twoapproaches coincide and the stream corridor of thegeomorphological classification contains the 100-mriver pixels with high contributing area and high topo-graphic index (2) the valleys are not superposed and thestream corridor contains weak values of topographicindex. The second case corresponds to the first peak of ourcurves and the first case to the second peak.

7. Conclusion

In this paper, we validated two methods for wetlanddelineation in large watersheds. The topographic indexhas the advantage to be easy and possible over largeareas as this method requires only a DEM. Clearly, highvalues of this index are good indicators of wetlands,even when it is calculated from a 100 m resolution DEMand in sedimentary context. At the time being, thegeomorphological classification is only available in theSeine watershed, but the method is transposable to otherlarge basins where the required information is available.As the two methods are coherent and both agree on thepredominance of wetlands in the encased and inaggraded subsystems, they are both good indicators ofwetlands. This confirms that geomorphology is the firstorder driving factor for wetland distribution, even in asedimentary basin with a strong influence of aquifers onhydrology such as the Seine basin. Therefore, itprovides a validation of the different assumptions ofthe geomorphological classification in the Seine riverbasin. The two approaches are also complementary. Theanalysis of the topographic index is not restricted to theriver corridors, and inside these corridors, it allows us toidentify zones with a higher water content. They providean efficient tool to complement the gaps of classicalwetland databases at the scale of large watersheds.


We calibrated a threshold value for the topographicindex against the waterlogged soils of two high-resolution pedological maps. We found thresholds of11.5 for the Tonnerre map and 11.2 for the Saint Diziermap, which are close in spite of the great geologydifferences between the two maps. Colluvial soils have asignificantly lower mean topographic index than theother waterlogged soils, suggesting that the former maynot be good indicators of wetlands. If we do not take thecolluvial soils into account to calibrate the thresholdtopographic index for delineating wetlands, we get avalue of 12.5 for the Tonnerre map and 12.3 for the SaintDizier map. Using these threshold values between 11.5and 12.5, we estimated that the area covered by wetlandsin the entire Seine basin ranges from 15.6 to 10.9%.

The topographic index does not account for thedisappearance of wetlands because of human influenceand the above fractions describe potential wetlands. Thisinformation is important to preserve but also to restorefunctional wetlands in large watersheds. In particular,the European Water Framework Directive was adoptedin 2000 by the European Union, with the objective toreach a chemical and ecological “good status” for watersurface and groundwater bodies in 2015. The interest ofthe geomorphological classification and the topographicindex to estimate important functions of wetlands withthat respect, as nutrient retention or flooding regulation,is the subject of ongoing research.

Acknowledgments

This study was supported by the PIREN-Seine re-search programme. The implementation of the geomor-phological classification of the fluvial corridors in theSeine river basin would not have been possible withoutthe help of Daniel Brunstein and Sylvain Théry and thesupport from the Agence de l'Eau Seine-Normandie. Theauthors are grateful to Shannon Sterling for her carefulproofreading of the manuscript.

References

Albinet M. Carte hydrogéologique du Bassin de Paris, 1:500000,dressée de 1963 à 1966. Paris: Bureau de Recherches Géologiqueset Minières; 1967.

Beven KJ. Runoff production and flood frequency in catchment oforder n: an alternative approach. In: Gupta V, Rodriguez-Iturbe I,Wood E, editors. Scale Problem in Hydrology. Norwell, Mass.:Reidel; 1986. p. 107–32.

Beven KJ, Kirkby MJ. A physically based, variable contributing areamodel of basin hydrology. Hydrol Sci Bull 1979;24:43–69.

Burt TP. The hydrological role of floodplains within the drainage basinsystem. In: Haycock NE, Burt TP, Goulding KWT, Pinay G, editors.

Buffer Zones: their Processes and Potential in Water Protection,Quest Environmental, Harpenden, UnitedKingdom; 1997. p. 21–32.

Burt TP, Butcher DP. Topographic control of soil moisture distribution.J Soil Sci 1985;36:469–86.

Bossard M, Feranec J, Otahel J. CORINE land cover technical guide—addendum 2000. Technical report N, vol. 40. Copenhague: EuropeanEnvironment Agency; 2000.

Cappus P. Etude des lois d'ecoulement. Application au calcul et a laprevision des debits. Bassin experimental d'AlranceLa HouilleBlanche; 1960. p. 493–520.

Commission de pédologie et de cartographie des sols (CPCS). Classi-fication des sols, travaux de 1963 à 1967. Laboratoire de Géologie-pédologie de l'ENSA-Grignon; 1967.

Curmi P, Durand P, Gascuel-Odoux C, Merot P, Walter C, Taha A.Hydromorphic soils, hydrology andwater quality: spatial distributionand functional modelling at different scales. Nutr Cycl Agreocosyst1998;50:127–42.

Dillaha TA, Inamdar SP. Buffer zones as sediment traps or sources. In:Haycock NE, Burt TP, Goulding KWT, Pinay G, editors. BufferZones: their Processes and Potential inWater Protection. Harpenden,United Kingdom: Quest environmental publ; 1997. p. 33–42.

Fustec E, Greiner I, Shanen O, Gaillard S, Dzana JG. Les zoneshumides riveraines: des milieux divers aux multiples fonctions. In:Meybeck M, de Marsily G, Fustec E, editors. La Seine en sonbassin: fonctionnement ecologique d'un systeme fluvial anthro-pise. Paris: Elsevier; 1998. p. 211–62.

Gaillard S, Sebilo M, Brunstein D, N'Guyen-The D, Grably M, Fustec E,et al. Typologie et fonctions des zones humides riveraines. In: CNRS,Universite Paris6, editor. rapport d'activite 2001 (PIREN-Seine),theme 1: Agriculture et qualite des eaux; 2001. 32 pp. http://www.sisyphe.jussieu.fr/internet/piren.

Haycock NE, Burt TP, Goulding KWT, Pinay G. Buffer Zones: theirProcesses and Potential in Water Protection. Harpenden, UnitedKingdom: Quest environmental Publ; 1997. 322 pp.

Hewlett JD. Soil moisture as a source of base flow from steepmountain watersheds, vol. 132. Asheville, NC: US Department ofAgriculture, Southeastern Forest Experiment Station; 1961. 11 pp.

Hewlett JD, Troendle CA. Non-point and diffused water sources: avariable source area problem. Irrigation and Drainage Symposium.Watershed Management, August 1975Logan: ASCE; 1975. 46 pp.

Lefeuvre JC, Fustec E, Barnaud G. De l'elimination a la reconquetedes zones humides. In: Fustec E, Lefeuvre JC, editors. Fonctions etvaleurs des zones humides. Paris: Dunod; 2000. p. 1–16.

Merot P, Ezzahar B, Walter C, Aurousseau P. Mapping waterlogging ofsoils using digital terrain models. Hydrol Process 1995;9:27–34.

Merot P, Squividant H, Aurousseau P, Hefting M, Burt T, Maitre V,et al. Testing a climato-topographic index for predicting wetlandsdistribution along an European climate gradient. Ecol Model2003;3221:1–21.

Oberlin G. Le controle des crues. In: Fustec E, Lefeuvre JC, editors.Fonctions et valeurs des zones humides. Paris: Dunod; 2000.p. 83–105.

Rhode A, Seibert J. Wetland occurrence in relation to topography: atest of topographic indices as moisture indicators. Agric ForMeteorol 1999;98/99:325–40.

Ward RC. The Fountains of the Deep and the Windows of Heaven:Perplexity and Progress in Explaining the Response of Rivers toPrecipitation. University of Hull; 1982. 32 pp.

http://www.sisyphe.jussieu.fr/internet/piren

http://www.sisyphe.jussieu.fr/internet/piren

geomorphological methods to characterise wetlands at the ... · the agreement between the two...

Documents