Analysing the properties and functional characteristics of heterogeneous soils containing several phases requires a correct estimation of the volume proportion of each phase. In the case of stony soils, the volume percentage of the content of rock fragments remains difficult to estimate in situ. This paper presents a method that uses field spatial electrical resistivity measurements to determine the volume proportion of rock fragments. Based on the hypothesis that the electrical resistivity signal noise increases as the proportion of rock fragments increases, a model was developed that uses the standard deviation of the apparent electrical resistivity measurements over a small area as an indicator of rock fragment contents. The model was tested on three study areas of several hectares containing soil units with varying quantities of rock fragments. The estimation of the rock fragment content was accurate, and the error estimation of about 6% was the same order of magnitude as the Bussian model (1983). The developed model strongly depends on the water content in the soil and the rock type and must be calibrated in each context. Nevertheless, estimations of the rock fragment content in stony soils can be performed efficiently in the surface horizon as well as all along the soil profile.

Recently, geophysical methods have been developed that can monitor soil characteristics spatially at high resolution. However, interpreting electrical measurements is difficult because geophysical data can be influenced by many soil variables, some of which vary over time. Our objective here was to use spatial measurements of electrical resistivity to define zones of homogeneity, to interpret them in terms of changing water contents, and to compare them with a soil map. Our underlying assumption was that the time variation of electrical resistivity at the field scale was only due to the dynamics of soil moisture in our studied field. Monitoring of soil electrical resistivity and soil moisture was performed at four dates during 2006 by two methods: by the use of the MUCEP (Multi Continuous Electrical Profiling) device, which gives measurements over a whole field, and by local gravimetric measurement of soil water content. Homogeneous zones were defined directly from measurements of the electrical resistivity and after ordinary kriging of the water content. Our analysis of spatial and temporal variability has permitted us to discriminate three temporally homogeneous zones, in terms of both electrical resistivity and water content, which were broadly related to the soil map. The use of electrical measurements enabled us to directly describe spatial and temporal changes in soil water content at the field scale and to describe some hydraulic processes, like lateral flows or upward capillary flows, that would be difficult to derive from soil maps.

Characterizing water processes at the field scale requires an accurate and high‐resolution description of the variability of representative state variables, such as water content. Obtaining these data, however, is unrealistic with conventional soil water sensors; we therefore propose an empirical approach to measure water content based on electrical resistivity. This geophysical method provides exhaustive and time‐repetitive information on the resistivity of a DC signal. Resistivity can be used to estimate water content, although the interpretation is not straightforward. As the approach is inexpensive and allows for repeated high‐resolution measurements, its potential benefits for the monitoring of soil water processes at the field scale warrant greater investigation. Soil electrical resistivity and soil water content were monitored at the field scale on four dates in 2006, by MuCEP (MultiContinous Electrical Profiling), which gives measurements at high spatial resolution, and by precise measurements of soil water content. The spatial organization of soil water dynamics was extrapolated from the geostatistical analysis of the spatial and temporal variability of electrical resistivity, and then improved data were obtained for the whole of the studied area. Homogeneous zones were then delineated and compared with soil units defined on a soil map. We demonstrated that the zones that had less than average soil water content over time for the whole area corresponded to those zones that had greater than average electrical resistivity values over time. The spatial similarity between these zones and the soil units defined on the soil map suggested that water flows were mainly vertical in the field. Nevertheless, our method could be used to characterize some specific hydrodynamic processes, such as lateral flows or upward capillary flows, that are usually difficult to characterize from the information derived from the soil map.

This study compares sequential Gaussian simulation (sGs), and collocated cokriging simulation (CCS) algorithms with respect to their success in modeling prediction uncertainty, and their accuracy in making point predictions of water content (w) in the soil cores of a 10ha area located in the Picardie region (Northern of France). The ability of sGs, and CCS in modeling uncertainty, and making point predictions was confronted with results achieved by ordinary kriging (OK), and collocated cokriging (CC) interpolation methods. A set of 81w samples was collected at the first 0.6m of soil. A first set of 51 measurements achieved through stratified random sampling was used for simulations, and interpolations. Thus, the remainder set of 30 measurements was kept for the validation. Electrical resistivity (ER1) of the first depth (0.5m) of investigation, which is linearly related to w and exhaustively sampled over the whole study area, was used as exhaustively sampled secondary information in the predictions, and the modeling of local, and spatial uncertainties of the target variable w using CCS and CC algorithms. In terms of the accuracy in making point predictions by simulation, and interpolation approaches, the results have shown that the approaches accounting for secondary exhaustive information (CCS and CC) are the more accurate. However, the difference between CCS and CC was not statistically significant stressing thus the convergence between a mean realization of a simulation algorithm, when the number of realizations is large enough, and the predicted map of an interpolation algorithm. As regards the modeling local uncertainty using accuracy plots, and goodness statistic (G), results have shown that CCS performed better the modeling prediction uncertainty than sGs that ignores the secondary exhaustive information in modeling uncertainty, and an improvement of local certainty on w was observed, through small values of standard deviations of the whole realizations at validation sites, for CCS compared to sGs. Regarding the spatial uncertainty, results revealed that the assessment of spatial uncertainty using simulation algorithms (sGs or CCS) were more revealing and more realistic than spatial uncertainty assessment using interpolation algorithms (OK or CC). The standard deviations varied much less across the study area for OK and CC compared to standard deviations across the study area for sGs and CCS highlighting that for interpolation algorithms, the variance of the errors is independent of the actual data values, and depends only on the data configurations.

Hedges are part of the landscape in many regions of the world. Among many important roles, they limit soil translocation. Hedges perpendicular to the slope at the lower end of sloping fields result in the formation of soil terraces. Quantification of fluxes of matter at the landscape scale has shown that terraces cannot be neglected. In this study, we try to quantify and explain the origin of the morphological and geochemical properties of terraces. The morphology of the terraces and corresponding stocks of soil material and chemical elements are assessed through a microtopographic study. The 11-ha study area is located on a rolling landscape in the Massif Central (France). The study is focused on three particular terraces. Two DEMs (2.5-m resolution) were established on the study area. The first DEM (DEM1) represents the actual elevation, using 4600 elevation spots. Elevation cross-sections were computed to determine the extent of the terraces, and a second DEM (DEM2) was then calculated, excluding all the elevation spots located in terraced areas. The thickness of soil material stored in each terrace is given by DEMst=DEM1−DEM2. It varies between 0 and 0.63m, representing a soil accumulation of 3–7m3m−1 hedge length. One-hundred and seventy-three samples were taken in the topsoil, and the content in some major and trace elements (Ca, Mg, K, Fe, Mn, Cr and Co) measured and mapped using ordinary kriging. The stock of these elements accumulated in the terraces was computed and compared to the stock eroded considering uniform erosion from the upper part of the fields. Results show a difference in stocks exceeding more than 20% for several elements, showing that uniform erosion is not a satisfactory explanation for the accumulations observed in the terraces. A higher contribution of the area located immediately upslope form the terraces results in a better agreement in the stocks comparison for most of the chemical elements studied. Evidence from coarse fragments study, particle size distribution, soil depth in the upslope part of the fields and microtopography show that the formation of the terraces is probably mainly due to redistributions through tillage. The geochemical properties of the terraces are probably exclusively the result of this mechanical redistribution, except for Mn and Co. Indeed, it is likely that since the plantation of the hedges, seasonal waterlogging conditions have significantly affected the mobility of these two elements through geochemical processes that resulted in their leaching downwards as well as perhaps out of the field.

Various approaches have been used to estimate metal pollutant element (TE) contents at unsampled locations in a 15-ha contaminated site located in the plain of Pierrelaye-Bessancourt (about 24 km Northwest of Paris). 87 samples of soil plough layer were randomly sampled at each mesh of a regular square grid over the whole study area and the total contents of Cd, Cr, Cu, Ni, Pb, and Zn were measured. A first set of 50 measurements, randomly selected from the 87 samples, was used for the prediction and another set of 37 measurements was kept for the validation. Topsoil organic carbon contents (SOC) were measured at 75 sites with 50 measurements sharing the same locations as TE. An aerial photography of the study area showing bare soils was selected for relating brightness intensities and SOC. Mapping procedures used were ordinary kriging (OK), cokriging (COK), collocated cokriging (CC), and kriging with external drift (KED). SOC maps used as exhaustively sampled information in KED and CC of TE were obtained by KED and CC procedures, respectively, accounting for 75 SOC measurements and the brightness intensities of numerical counts provided by the visible bands of the aerial photograph bare soils. Consequently, for each TE, four maps were generated: two maps resulting from KED and CC procedures (K-ED-SOC75P, CC-SOC75P), another one provided by standard cokriging (COK-TE50SOC75) accounting for TE prediction set plus 75 SOC measurements, and the last one corresponding to that estimated by ordinary kriging from only prediction set measurements (OK50). Three indices: (1) the mean prediction error (ME) and the mean absolute prediction error (vertical bar ME vertical bar); (2) the root mean square error (RMSE); and (3) the relative improvement (RI) of accuracy, as well as residuals analysis, were computed from the validation set (observed data) and predicted values. On the 37 test data, the results showed that the more accurate predictions were systematically those obtained by kriging accounting for SOC map predicted by KED from 75 SOC measurements and brightness values of the aerial photo (K-ED-SOC75P) followed closely by CC-SOC75P procedure, except for Cu and Zn where CC-SOC75P appeared to be slightly more accurate than KED-SOC75P. In regard to the RI of accuracy between prediction methods, the results confirmed once for all the benefit of accounting for SOC data set plus the exhaustively sampled information provided by the aerial photography regardless of the considered TE. Nevertheless, for Cd, Pb, and Zn, the RI of accuracy was less than 20% between the two most accurate methods (KED-SOC75P and CC-SOC75P) and standard colcriging in which the information provided by the aerial photography is ignored when mapping.The sensitivity of KED-SOC75P and CC-SOC75P approaches to the sampling density of the target variables (TE) was assessed using 10 random subsets of different sizes (25 and 33 observations) drawn from a prediction set that includes 50 data. Results have shown that the TE estimates by KED-SOC75P and CC-SOC75P approaches using only 25 TE samples were much more accurate than the estimates performed by OK50 and COK-TE50SOC75 approaches that use the whole samples of the prediction set. Moreover, the RI of accuracy was reduced by less than 15% if the original sampling density was reduced by a third. (c) 2005 Elsevier B.V. All rights reserved.

The aim of this study is to determine the factors that control the spatial variability of chromium in the topsoil at the landscape scale by using a combination of both individual soil pit informations and systematic soil sampling analyses. The 11 ha study area, located in Massif Central (France), includes contrasted bedrock and topography.In a first approach we determined the relation between the concentrations in chromium and other major elements reflecting the soil pedogenic processes and some particular minerals indicating the original materials the soils are derived from. For this purpose, we used data collected at the catena scale from 10 soil profiles in two toposequences matching all geological and topographic situations. In a second approach, we studied the spatial variability of Cr in both topsoil and alterite through a systematic sampling study.Chromium concentrations range between 16 and 355 mg kg(-1) in the alterite, and 31 to 129 mg kg(-1) in the topsoil. Chromium concentration in alterite presents a complex spatial distribution governed principally by the geological origin of the alterite. Two major differences between chromium concentration in the alterite and in the topsoil may be deduced. First, topsoil chromium distribution depends on topography. Increased weathering together with selective erosion of the topsoil resulted in smaller chromium concentration in the concave compared to the convex area. Secondly, chromium concentration in the topsoil varies spatially much less than in the alterite. Such differences can probably be observed in similar geomorphic conditions and for other chemical elements that can be considered as geochemically immobile through weathering. (c) 2005 Published by Elsevier B.V.

A field study was conducted to model the spatial and temporal variability of soil water content (PP) and to assess the potential role of selected permanent soil variables and topography to reveal scale dependence and the temporal persistence of W spatial patterns. The W was measured gravimetrically in the first 0.3-m of soil at 84 sites of a regular grid, and W measurements were repeated at the same locations 2-3 times a year, i.e., at sowing, tilling, and harvesting. Thus seven sets of Wwere collected between 2000 and 2003. The proportions of sand, silt and clay, pH, organic carbon, CaCO3 and total nitrogen (permanent soil variables) were measured in the upper soil layer (0-0.3-m) from soil cores sampled at the start of the experiment. Furthermore, the slope (P) and topographic wetness index (TWI) were derived from a DEM. The temporal variance of W measured at different dates, according to the spatial scale, was assessed by the co-dispersion coefficient, and the mean standardized relative difference (delta(ij)) of W was used to assess the pattern of relatively dry or wet zones. Partial least square (PLS) regression analysis was used to assess interrelationships between delta(ij) and permanent soil variables, and beta and TWT The behaviour and the relationships among the selected permanent soil variables and topography attributes were summarized through the kriged map of the latent vector (LV), which captures most of the systematic variation. The delta(ij) map was compared to the LV kriged map and to the maps corresponding to the main permanent soil variables assessed by factorial kriging analysis (FKA) in which the response variable (delta(ij)) was ignored. Results show that the 6 of W appeared to be an efficient tool to identify the wetter and drier zones. Furthermore, results also show that permanent soil variables allow zones to be detected across the study area where additional investigations are needed to elucidate the temporal variation of a dynamic soil variable such as W.

A field study was conducted to quantify spatial soil variability and to analyze correlations among soil properties at different spatial scales. Soil samples from 0 to 30 cm depth were collected from two adjacent fields in the southwestern Beauce Plain (France) which consisted of Haplic Calcisols and Rendzic Leptosols. Factorial kriging analysis (FKA) was used to describe the co-regionalization of nine soil properties. A linear model of co-regionalization including a nugget effect, and two spherical models were fitted to the experimental data direct and cross-variograms of the topsoil layer properties which were previously estimated. Co-kriged regionalized factors, related to short and long-range variation, were then mapped to characterize soil variation across the two fields. The potential value of ancillary sampled variables, such as yield data, to provide information on soil properties was tested. The relation between yield and measured soil properties appeared to be weak in general. However, the structures of the variation in yield appeared to be relatively stable for two years and showed similar patterns as the co-kriged soil factors. This suggests that information on the scale of variation of soil properties can be derived from yield maps, which can also be used as a guide to suitable sampling interval for soil properties and as a basis for managing fields in a precise way.

So far, soil-landscape models have been based on soil surface topographic information only. However, hillslope hydrology that affects soil distribution is also controlled by sub-surface flow pathways that may not entirely be explained by surface terrain features. This paper compares the accuracy of a model for predicting the spatial variations of the hydromorphic index (HI) using the surface topography with that of a model that uses the sub-surface topography. The study was conducted in an agricultural hillslope of the Armorican Massif (Western France) where two DEMs were generated from observations of the surface topography and the topography of the saprolite upper boundary. For both DEMs, the best correlations with HI occurred for altitude, elevation above the stream bank, and compound topographic index. Predictions of HI using the sub-surface topography greatly decreased prediction errors, especially for intermediary HI values at middleslope position. Finally, the use of subsurface topography as a novel approach to enhance existing techniques in new applications is discussed.

Kriging with topographic variable (beta) and electrical resistivity (rho) measurements used as external drifts was compared with universal kriging and kriging either with a topographic variable or with electrical resistivity measurements taken as external drift, to predict the depth of a limestone bedrock upper boundary (LUB) in the "Petite Beauce" region, southwest of the Paris Basin.Two sets of limestone depth apparition (LUB) were recorded. A first set of 67 measurements was used for the prediction and another set of 50 measurements was kept for the validation. The slope gradient (beta) and the resistivity (rho), which are linearly related to the limestone upper boundary depth (LUB) and exhaustively sampled over the whole study area, were used as external drift variables for kriging.Two indices, (i) the mean error (ME) and (ii) the root mean square error (RMSE), as well as residuals analysis, were computed from the validation sample (observed data) and predicted values. On the 50 test data, the results showed that kriging using two external drifts proved to be less unbiased and more precise compared to kriging using only one external drift variable in predicting the target variable. (C) 2002 Elsevier Science B.V. All rights reserved.

A nondestructive and spatially integrated multielectrode method for measuring soil electrical resistivity was tested in the Beauce region of France during a period of corn crop irrigation to monitor soil water flow over time and in two‐dimensional (2‐D) with simultaneous measurements of soil moisture and thermal profiles. The results suggested the potential of surface electrical resistivity tomography (ERT) for improving soil science and agronomy studies. The method was able to produce a 2‐D delimitation of soil horizons as well as to monitor soil water movement. Soil drainage through water uptake by the roots, the progression of the infiltration front with preferential flow zones, and the drainage of the plowed horizon were well identified. At the studied stage of corn development (3 months) the soil zones where infiltration and drainage occurred were mainly located under the corn rows. The structural soil characteristics resulting from agricultural practices or the passage of agricultural equipment were also shown. Two‐dimensional sections of soil moisture content were calculated using ERT. The estimates were made by using independently established “in situ” calibration relationships between the moisture and electrical resistivity of typical soil horizons. The thermal soil profile was also considered in the modeling. The results showed a reliable linear relationship between the calculated and measured water contents in the crop horizon. The precision of the calculation of the specific soil water content, quantified by the root mean square error (RMSE), was 3.63% with a bias corresponding to an overestimation of 1.45%. The analysis and monitoring of the spatial variability of the soil moisture content with ERT represent two components of a significant tool for better management of soil water reserves and rational irrigation practices.

This study investigates scale spatial specific dependence of some major and trace element contents in topsoil horizons and in alterite type horizons (subsoil horizons) over a weakly contaminated area. The study seeks also to compare the spatial distributions of chemical elements at both depths to the spatial variability of the alterite type in order to examine the effect of the parent material. The probability of occurrence of alterite type was estimated using indicator kriging (IK). Factorial kriging analysis (FKA) was used to analyze spatial variability in some soil chemical properties (Ca, Mg, K, Fe, Cr and Co) measured at two depths over a geologically contrasted area of 10 ha in central of France. The coefficients of the coregionalization matrix at different spatial scales reveal the dominant long-range autocorrelation and cross-correlation in all chemical elements in both depths except for trace elements (Co and Cr) where the short-range structure dominates the cross-correlation. The resulting structural correlation coefficients showed strong correlations between variables changing as a function of spatial scale. These relationships between chemical properties at different spatial scales were not revealed by the linear correlation coefficients. A principal component analysis was performed on the coregionalization matrices at each depth to summarize the relationships among the variables at the different spatial scales. Cokriging allowed mapping each spatial component for both depths. These maps were then compared with the probability map of alterite type estimated using indicator kriging. This comparison revealed that the spatial pattern of chemical elements in the subsoil horizons is almost certainly due to the alterite type effect, whereas the alterite type effect on the spatial pattern of chemical properties in the topsoil horizons was partly hidden by human activities and erosion.

*V. Souchere, INRA-SAD Ile de France, RN 10 (Route de Saint-Cyr), 78026 Versailles cedex (France); O. Cerdan, Y. Le Bissonnais, A. Couturier, and D. King; INRA-Science du sol, Centre de recherche d’Orléans, BP 20619, 45166 Ardon cedex (France); F. Papy, INRA-SAD Ile de France, BP 1, 78850 Thiverval Grignon (France). *Corresponding author: souchere@versailles.inra.fr ABSTRACT In loamy areas of Northern Europe, soil erosion is a widespread phenomenon, despite low rainfall intensity and a gentle topography. Interactions between meteorological conditions, farming operations and topsoil texture bring about rapid and significant changes in the hydraulic properties of topsoil. The processes involved in surface crusting are extremely dynamic and crust characteristics are often difficult to measure. Modeling infiltration into these crusts has led to the development of equations of varying complexity, ranging from simple empirical equations to numerical solutions of the Richards equation. Obtaining the parameters for the more mechanistic approaches remains a challenge. The objective of our work is to develop a simple runoff and erosion model based on field experiments and knowledge about crusting and agricultural practices (tillage direction, roughness, location of dead furrows, etc.). The model calculates the total runoff volume for a rainfall event at any point in the watershed. This model is able to serve as a simulation tool in order to test several anti-erosion schemes and choose the more efficient scheme for a given context.

The variability of Cr content is studied along a toposequence of soils developed on amphibolite and gneiss. Amphibolites are Ca-rich, gneiss are K- and Cr-rich. A mineralogical study shows that biotite is the main bearer of Cr, and explains the relationship between K and Cr in the alterites. This relation still exists in pedological horizons. The nature of the parent material and the erosion are the main factors explaining the variability of Cr along the toposequence. (C) 2001 Academic des sciences / Editions scientifiques et medicales Elsevier SAS.

The aim of this paper is to assess the performance of kriging with an external drift according to the sampling density of the primary variable and the precision of the auxiliary variable. The variable of interest is the thickness of the silty clay loam horizon (TSCL), which is known at 341 points (prediction sample) and 90 points (validation sample). Slope gradient derived from two Digital Elevation Models (DEM-LA and DEM-HA) is exhaustively known over the whole study area. It is chosen as an auxiliary variable. The DEM-LA was computed from contour lines of the topographic maps and the DEM-HA was obtained from altitude measurements taken with a laser tachometer on the field. The validation sample was used to compare the two maps of TSCL predicted using the prediction sample and slope gradient derived from DEM-LA and DEM-HA respectively. The results have shown that predictions by kriging with an external drift derived from the DEM-HA are more precise in our case, with mean gain in precision being about 24%. To examine the influence of observation density of the variable of interest (TSCL) on the performance of kriging with an external drift, five sub-samples were considered, including 50, 100, 150, 200 and 250 individuals randomly selected among the prediction sample. For each sub-sample, two predictions of TSCL were conducted by kriging using slope gradient derived from the DEM-LA and DEM-HA. The mean prediction errors using slope gradient from the DEM-LA are consistently higher than mean errors with slope gradient from the DEM-HA. The root mean square errors (RMSE) of the different sampling scenarios, confirm the progressive decrease of the estimation error of TSCL as sampling density of the variable of interest increases, regardless of the origin of the external drift. Values of the RMSE also show that 50 individuals associated with a slope gradient from the DEM-HA provide a better prediction than 341 individuals associated with a slope gradient from the DEM-LA.

This study examines two mapping method's sensitivity to the sampling density of the variable of interest, which is the thickness of a silty-clay–loam (TSCL) horizon. The two methods are simple linear regression (SLR) and universal kriging with external drift (UKEXD). As slope gradient (β) derived from a DEM, is available for the whole study area and linearly related to TSCL horizon, was used for TSCL prediction by SLR and by UKEXD. The accuracy performance of TSCL prediction using these methods was assessed by comparison with another group of 69 sample points (validation sample) where the TSCL is actually measured. In the validation procedure for the two methods, two indices were calculated from the validation sample (measured values) and predicted values. These two indices are the mean error (ME) and the root mean square error (RMSE). The results showed that UKEXD was more accurate than the SLR. The improvement of the accuracy of the prediction from SLR to UKEXD was about 38%. To examine the effect of sampling density of TSCL (variable of interest) on the performance of both mapping methods, five subsets of 40, 50, 75, 100 and 125 observation sites of TSCL were randomly selected from the 150 sites of the prediction sample. For each subset, a prediction of TSCL was performed over the study area by: (i) SLR; (ii) UKEXD. The validation sample was used to compare the performance of the two methods according to the sample size of the variable of interest. The results show that whatever the sample size may be, UKEXD performs on average more accurate predictions than SLR. Moreover, the results indicate that UKEXD performed better when the sample size of the variable of interest increases. On the contrary, the performances with the linear regression remain stable whatever the sample size may be.

The initial objective of this study was to identify relationships between terrain attributes and soil cover over a Loessic flat area covering a limestone plateau. The second objective was to determine those energy factors (flow path, solar radiation, wind intensity) which could improve the understanding of morphology and soil genesis. We describe 341 field observations taken over 1600 ha of an experimental area used for monitoring the water and nitrate supply in the Petite Beauce Region (100 km to the Southwest of Paris). For each observation, several soil variables were encoded. One was the presence of a non-calcareous clay–loam (NCCL) horizon. Relief is very smooth in this region (mean slope around 0.5%). The main terrain attributes were derived from a Digital Elevation Model (DEM) at 20×20 m and assigned to the pedological observations. A multiple logistic regression was used to analyse the relationship between the NCCL horizon and terrain attributes. Special statistics were used for aspect, because of the circular nature of this variable. The results show a strong relationship between the presence of NCCL horizon and slope gradient and slope aspect, while hydrological parameters are not correlated with this horizon. The mean angle of the aspect frequency of the NCCL horizon was calculated and compared to the mean angles of wind direction and solar radiation balance. There is a small difference between wind direction and aspect frequency of the NCCL horizon. This result confirms the role of wind in the spatial pattern of soils. Further data are required to better understand the combination of several factors (role of vegetation) and the age of the reshaping.

The present study involves the possibility of using geology and relief to map soil classes. We initially focused on two small catchments considered as representative of a 6000-ha forested area overlying a sandstone bed. The catchments differed in the stratigraphic sequence of sandstones, i.e., rich or poor in weatherable minerals. In one, the depleted bedrock was downstream and the rich was upstream, and the converse obtained for the second. Relationships between soil classes and environmental factors were modeled using two discriminant functions corresponding to the two types of stratigraphic sequences found in the catchments. More than 70% of the soil class distribution in small catchments can be explained by the nature of the substratum and attributes derived from a digital elevation model (DEM). These relationships were then applied to a larger region. An automatic catchment delineation was first carried out with the DEM and was then combined with geologic maps. The choice between the two discriminant functions was based on the stratigraphic sequences in each catchment. Predicted soil classes were compared to soil classes conventionally mapped in 1978 at the scale of 1:100,000. The results show that the model reproduced the soil map over 55% of the area studied. Disagreements were due primarily to the existence of superficial deposits not mentioned on the geologic maps and to an altitude effect that is not sufficiently considered in the study of small catchments.