SSSAJ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (9)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Schoorl, J. M.
Right arrow Articles by Bouma, J.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Schoorl, J. M.
Right arrow Articles by Bouma, J.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Schoorl, J. M.
Right arrow Articles by Bouma, J.
Related Collections
Right arrow Watershed and Landscape Processes
Right arrow Soil Erosion
Right arrow Water Content
Soil Science Society of America Journal 66:1610-1619 (2002)
© 2002 Soil Science Society of America

DIVISION S-6—SOIL & WATER MANAGEMENT & CONSERVATION

Modeling Water and Soil Redistribution in a Dynamic Landscape Context

J. M. Schoorl*, A. Veldkamp and J. Bouma

Laboratory of Soil Science and Geology, Department of Environmental Sciences, Wageningen University, P.O. Box 37, 6700 AA Wageningen, The Netherlands

* Corresponding author (Jeroen.Schoorl{at}aio.beng.wau.nl)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Soil suitability assessments for land use planning are commonly based on on-site specific topographic, soil, and climatic characteristics, and are often neglecting the effects of physical landscape processes by water. The spatial and temporal variability of the landscape requires specific input data and modeling procedures. Existing studies aiming at the landscape level often are data-driven and operating at detailed resolutions of seconds and hours for single slopes or catchments. This study aims at the coarser level of multiple catchments over a period of 10 yr. A dynamic modeling approach is applied to a study area for the effects of soil redistribution within a landscape (run-on, run-off, erosion, and sedimentation) on subsequent soil water availability. Simulation scenarios include factors of water routing, soil depth, and erodibility. Different approaches for surface run-off routing have a major influence on the magnitude and spatial patterns of soil redistribution. Also initial conditions such as soil depth, parent material, and erodibility have spatial impacts upon soil erosion and sedimentation within the landscape. Locally decreasing water storage capacity (on-site) may cause increased run-off and erosion at lower positions in the landscape (off-site). Localized soil redistribution can cause significant changes in actual soil depth and indirectly affect available soil water. The changing patterns of soil redistribution for the different scenarios are both related to modeling techniques as well as to the implemented boundary conditions. This study indicates that at the landscape scale spatial variability in for example soil properties is inherent to both the complexity of the landscape (parent material) and on-site and off-site effects of controlling processes.

Abbreviations: AW, available water • DEM, digital elevation model • GIS, geographical information system • GN, gneiss • Kes, erodibility of the soil surface • LAPSUS, landscape process modeling at multi-dimensions and scales • LM, limestone • MA, marls • MF, multiple flow direction routing • MO, molasse • PL, phylite • RI, river terraces • SD, steepest descent flow routing • SP, serpentinite


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
TRADITIONALLY, SOIL SUITABILITY ASSESSMENTS and land evaluation are largely based on the soil profile data (e.g., Food and Agricultural Organization, 1976). This profile based one-dimensional view has governed soil science for many years. However, soil science is gradually evolving towards a three-dimensional landscape context (Jacob and Nordt, 1991; Bouma, 1999). Consequently, an increasing number of studies is directed at the slope and catchment level (e.g., Beven et al., 1984; Nearing et al., 1989; Gessler et al., 2000). Nevertheless the detailed spatial and temporal resolution of such studies hamper the implementation at higher landscape levels such as multiple catchments on varying lithologies.

Landscapes are the result of and shaped by a set of interrelated and nonlinear processes (Milne, 1991). These processes all show specific impacts in space and time. In most landscapes, water and soil redistribution (erosion and sedimentation) are the main processes that form topography, soils, and ecosystems (Holling, 1992; Rigon et al., 1994). Results of these landscape processes are that they connect soil pedons in specific ways such as toposequences and catenas, causing changes in soil properties (from soil depth to available water) to have both on- and off-site effects. On-site effects are considered locally forced changing conditions (e.g., reduced infiltration by crust forming and erosion), while off-site effects are the result of changed conditions elsewhere. However, these off-site effects within landscapes such as deposition, changes in run-on or even land use have been mostly neglected in land evaluations and other agricultural research (Fresco, 1995; Veldkamp et al., 2001). In contrast with current procedures, land use suitabilities should also be evaluated for both on-site and off-site effects. While creating simulation scenarios, in addition to average rainfall data and soil profile data, run-off, and run-on should be taken into consideration for each land unit. Also aspect, slope gradient, and wind direction may result in different impacts of radiation, evaporation, and temperature upon available water, factors that are not commonly expressed in simulation models of land evaluation. Furthermore, relevant time scales should be considered ranging from a growing season for annuals (<1 yr) up to multiple decades (>10 to 100 yr) for perennials and natural vegetation.

Aside from soil properties, soil redistribution (e.g., erosion and sedimentation) affects the productivity of each landscape element (Hall and Olson, 1991). Consequently, similar soils at different landscape positions may have different productivity potentials. The resulting landscape processes can have large impacts in sloping (Lal, 1997). However, most prior studies are mainly focussed on the erosion aspects of the landscape processes while deposition could have an equally large impact. Payton and Shishira (1994) published an example illustrating both aspects of erosion and deposition along a soil catena. They showed how within one century a good productive district turned into a marginal area with very limited productivity. Productive soils were either eroded or buried as a result of mismanagement of arable land. A similar example for a less sloping and drier environment in Niger is described by Bromley et al. (1997).

Evaluating land use systems thus requires a combined on- and off-site approach in such a way that it not only includes soil processes at the pedon level, but also takes into account three-dimensional landscape processes. Such an approach should aim at exploring natural soil and landscape processes in such a way that desirable land use can lead to continuing productivity without reducing soil quality. This aim will require a landscape approach focusing not only on evaluating pedon characteristics but also on introducing landscape components with on- and off-site impacts for both short and long time-spans (e.g., Pennock and van Kessel, 1997; Schoorl and Veldkamp, 2001).

Ideally, the results of these landscape-scale studies should be used in a dynamic landscape evaluation to gain more insight in the complex functioning of ecosystems and soils at the landscape scale (Gessler et al., 2000). Furthermore these type of studies should include not only agricultural production systems but also many nonagricultural land uses as well. Concerning the biophysical aspects of such a evaluation there are many sophisticated hydrological and geomorphological models available (e.g., Beven et al., 1984; Nearing et al., 1989). However, they all require high temporal (up to seconds) and spatial (slope segment) resolution for the input data and are therefore unsuitable for the proposed soil landscape suitability assessment with limited input data.

This paper describes a simple modeling approach to integrate three-dimensional landscape processes with readily available soil survey data. Thus, it provides a new dimension to soil survey interpretations that currently do not consider the multi-dimensional landscape context. Examples of scenario building will show the impact and unexpected effects of this annual integration over a period of 10 yr at the multi-catchment level. Model simulations will allow assessments of current and possible water and soil redistribution for on-site effects (local changes in terms of boundary conditions) and off-site effects (caused by changes elsewhere) within the landscape.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Study Area
A study area of about 239 km2 has been selected in the Guadalhorce River basin ranging in altitude from 100 to 1200 m (Fig. 1) . The area is situated near the village of Álora (long. 04°41'57''W, lat. 36°49'10''N) in the province of Málaga, southern Spain and was selected because of its lithological diversity and resulting soils. General climate is summer-dry Mediterranean with a mean annual temperature of 17.5°C and mean annual rainfall of 534 mm, mainly from October to April (Álora Estación). This landscape is composed of geological and soil units shown in Fig. 1b (Ruiz et al., 1993). Comparing Fig. 1a and 1b, it is clear that steeper areas (high contour density) reflect more resistant parent material like limestone (LM) at the northeast, serpentinite (SP), molasse (MO), and gneiss (GN) at the central western zone and phylites (PL) at the southeast corner. The marls (MA) are found at the less steeper areas (less contour density) from the southwest to the northeast. River terraces (RI) are located in the middle of the research area. This same broad geological outline and the resulting soils of the area are reflected in the simplified soil depth map (Fig. 1c). This map has been developed from field survey of the dominant land units, georeferenced sampling, and spatial interpolation (Wielemaker et al., 2001).



View larger version (133K):
[in this window]
[in a new window]
  		Fig. 1. Study area in the lower Guadalhorce River basin with (a) DEM of the area with contour line interval 50 m, (b) major soil units according to parent material: river terraces (RI), gneiss (GN), molasse (MO), marls (MA), limestone (LM), phylites (PL), and serpentinite (SP), and (c) initial soil depth map.

 
Modeling
Topography, represented by a digital elevation model (DEM) is the driving force behind the geomorphic processes. This readily available DEM consists of a raster map with a grid-cell resolution of 100 m (SGE, 1997). With this resolution, the study area is divided into 132 rows and 181 columns. Because of the relevant 1 to 10 yr temporal resolution and the multiple catchments spatial resolution, the LAPSUS model was used (Schoorl et al., 2000). Consequently, input data and resulting modeling structures were implemented at an annual basis. LAPSUS is a finite element model and uses sediment transport equations based on works of Kirkby (1971) and Foster and Meyer (1972)(1975) on the continuity equation for sediment movement. Assuming that the transport capacity C and the detachment capacity D or settlement capacity T remain constant within one finite element, the annual sediment transport rate S [m2 a-1] is evaluated as follows:

[1]
where S over the length dx (m) of a finite element is calculated as a function of transport capacity C (m2 a-1) compared with the amount of sediment already in transport S0 (m2 a-1). Please note that S is expressed as soil volume per unit grid width per year. To convert to erosion or deposition rate in mass per area per year, S is divided by the grid length (dx) and multiplied by soil bulk density. Term h (m) refers to the transport capacity divided by detachment capacity D (m a-1) or settlement capacity T (-m a-1). Assuming, amongst others, that detachment and settlement capacity is proportional to a certain shear and that the drag coefficient is constant, we obtain:

[2]

[3]
where Kes is a lumped surface factor (m-1) indicating the erodibility of the surface and Pes a similar factor indicating lumped sedimentation characteristics (m-1), Q is the annual run-off flux (m2 a-1) and {Lambda} the height difference or slope (-). Note that the erosion conditions for D or sedimentation conditions for T will result in opposite signs for the change in S.

This approach simulates net annual soil redistribution by mimicking one average yearly event to shape the landscape. Therefore, annual sediment transport rates are driven by the topographical potentials imposed upon the run-off water, taking into account annual infiltration and evaporation losses by using the amount of net annual run-off reaching the main drainage system. Total amount of annual run-off water and the down-slope gradient determine transport capacity. When this transport capacity is higher than the actual sediment transport rate, the sediment in transport can be increased by detachment (erosion). The rate of detachment is controlled by the Kes factor, which stands for the erodibility of the soil surface (e.g., Beven and Kirkby, 1979; Kirkby, 1987). By definition this Kes factor incorporates many properties of the soil surface including crusting and land use. This factor should not be confused with the universal soil loss equation (Wischmeier and Smith, 1958) erodibility factor K. Nevertheless both factors comprise numerous local surface characteristics and resulting variability is considerable (Torri et al., 1997). When the sediment transport rate becomes higher than the transport capacity because of decreasing gradients in slope or discharge, the excess of sediment is deposited by sedimentation taking into account the settlement factor Pes.

Run-off routing is simulated with both steepest descent and multiple flow directions. The steepest descent flow routing directs the run-off towards one single cell with the steepest gradient (Moore et al., 1991). In the multiple flow direction routing, all down-slope neighbors receive a fraction of the run-off following (Holmgren, 1994):

[4]
where fraction fi of the run-off in direction i, is equal to the difference in height or slope gradient {Lambda} (tangent) in direction i powered by factor p, divided by the summation of {Lambda} for all (never more than 8) down-slope neighbors j powered by factor p.

Available water for crop growth is an important land quality within the dynamic landscape context. Factors determining available water for different soil units are for example, texture, bulk density, organic matter, and effective soil depth which are related to the effective rooting depth of the crop considered (Bouma and Droogers, 1999). In this study, we define available water for any soil unit as the difference in water content between -10 and -1500 kPa and as a function of the effective soil depth (see Fig. 1c). The on- and off-site effects on soil water are reflected through changes in soil depth due to erosion and sedimentation.

Input Parameters and Scenarios
In addition to topography, precipitation is also a principal input parameter. The spatial and temporal distribution of rainfall in this Mediterranean area is highly variable (Pardo Iguzquiza, 1998; Renschler et al., 1999). For example mean annual rainfall ranged from 248 mm in 1994 to 1052 mm in 1996 at Álora Estación. However, since we want to demonstrate the spatial impact of slope and soil related landscape sensitivities, we assume constant mean annual rainfall for all scenarios, neglecting these effects of climate variation. The effect of altitude on annual rainfall amounts was calculated following the linear regression (r2 = 0.63) established by Pardo Iguzquiza (1998) giving annual rainfall Ra (m) at altitude a (m) as:

[5]

The mean annual rainwater retention and evaporation loss in this region is ~75% of the precipitation, based on measured mean annual discharges in various tributaries in the Guadalhorce River basin, leaving 25% of the annual rainfall as run-off (e.g., CHS, 1974). Since land use in the region is related to the parent material, effects of different land uses are considered constant in both time and space to allow for specific evaluation of the main scenario parameters. We formulated four scenarios (A to D) with increasing topographic and soil related complexity. The last scenario (D) is considered as the most realistic one (Table 1). The simulation period was 10 yr with a time step of 1 yr, resulting in a yearly update of the input parameters such as soil depth, topography, and available water storage capacity.


View this table:
[in this window]
[in a new window]
  		Table 1. Main scenario input parameters for water routing, soil depth and erodibility (Kes).

 
Scenario A.

Scenario A is the baseline condition for all experiments, showing only few landscape dynamics, as implied by the aforementioned inputs. All soil related parameters are assumed uniform and constant, soil depth is set at 2 (m) and storage capacity at 0.132 (m). The Kes and Pes factors are calibrated for the whole area, giving soil losses comparable with other studies in this region (e.g., Poesen and Hooke, 1997; Renschler et al., 1999). The calculations of the transport rate and the distribution of the collected run-on are made by routing the run-off towards the steepest down-slope (SD) neighbor cell (Moore et al., 1991).

Scenario B.

Scenario B introduces the first step towards a more dynamic landscape concept by applying the multiple flow (MF) direction algorithm of Eq. [4]. The weight factor p of Eq. [4] is set at 4.0 for the topography in the research area, following the experiments of Holmgren (1994). This scenario will introduce the diverging properties of the topography, since the grid-cell size of 100 m in this study is far below the mean slope length in the research area of over 400 (m) resulting in a sufficient number of grids representing the slope.

Scenario C.

Scenario C implements the initial soil depths of Fig. 1c. As a result every grid cell has its own separate capacity for annual infiltration and storage. Due to storage limits, shallower soils will generate more run-off than other areas with deeper soils. As a result within (sub)catchments redistribution of the run-off will take place. We calibrated the model by adjusting the infiltration loss so that the total mean annual run-off leaving the catchments and subcatchments equals to the previous scenarios.

Scenario D.

Scenario D, provides different Kes (erodibility of the surface) factors for each soil type. A simple relationship is used in assigning Kes with lower values to the more resistant parent materials in relation to their actual soil depth. Data on different Kes factors were aggregated from qualitative field observations, bulk density, and literature review on different parent materials (e.g., Kosmas et al., 1997; Cerda, 1999; Romero Diaz et al., 1999). Final Kes factors used in this study ranged between 1.10-5 and 3.10-5 (m-1) (Table 1).


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Water Redistribution
The run-off pattern of Scenario A is presented in Fig. 2a . It shows a typical pattern generated by many of the standard modeling approaches and geographical information systems (GIS) using a steepest descent flow routing (Moore et al., 1991). The central channel in the middle of the graph collects most of the run-off and transports the water out of the area at the lower middle (south) boundary. Note that the total run-off does not change as a result of different routing algorithms in Scenario B as long as the annual precipitation, evaporation, and infiltration balance remains the same. However, the pattern inside subcatchments does change considerably going from Scenario A to B (Fig. 2b) because the use of more divergent routing. These changes, after implementing the multiple flow direction algorithm of Eq. [4] in Scenario B, can be found on both the slopes and in the channels of the subcatchments. In general, the water is allowed to diverge more on the irregular and convex slopes, increasing the length of the flow paths and narrowing the water divide (compare Fig. 2c and 2d). As a result in upland areas less water is flowing directly towards the upper parts of the channels although the total discharge from each catchment does not change (Desmet and Govers, 1996; Tarboton, 1997).



View larger version (140K):
[in this window]
[in a new window]
  		Fig. 2. Water redistribution with (a) run-off patterns of annual rainfall excess under Scenario A, (b) changes in run-off between Scenario A and B using multiple flow direction routing (MF) instead of steepest descent flow routing (SD), (c) detailed area indicated in (a) with SD, (d) detailed area with MF and (e) changes in run-off between Scenario B and C.

 
Resulting differences in run-off comparing Scenario A and B to Scenario C (Fig. 2e) are directly related to the effects of different soil depths, allowing less or more water to be stored and infiltrated. In this way these imposed scenario conditions cause the shallow soil areas (e.g., soils on LM and SP) to generate more run-off, while areas with deeper soils (e.g., soils on MA) generated less run-off which is consistent with field measurements (van Wesemael et al., 2000). When only considering Scenarios C and D, the initial run-off patterns will not have any influence since they will be the same for each scenario. However, after the first time step of 1 yr, erosion and sedimentation patterns started to change initial soil depths and topography throughout the area differently for each of the scenarios. As a result after the first year during the next time step run-off patterns started to change as well between scenarios. This implies that when analyzing terrain attributes from a DEM (e.g., Boer et al., 1996; Tucker et al., 2001), such attributes need to be evaluated for each step in time of the model.

Soil Redistribution
An example of soil erosion and sedimentation patterns for Scenario D is plotted in Fig. 3 . In general, lower erosion rates are situated on low gradient slopes especially near the center of the study area. Higher erosion rates can be found along the major tributary channels (high discharges) and, for example, on some steep slopes in the north of the study area (Fig. 3a). Sedimentation in the landscape is located on the longer slopes and valley bottoms of the area (Fig. 3b). Higher sedimentation rates can be found mainly in the central RI where tributary channels lose their energy in the flatter river plain. This can also happen locally in the tributary channel itself because of changes in channel gradient and lateral sediment input.



View larger version (113K):
[in this window]
[in a new window]
  		Fig. 3. Annual soil redistribution in the landscape for Scenario D showing: (a) erosion and (b) sedimentation.

 
In Table 2 simulation results for the different scenarios are presented with mean erosion and sedimentation rates and net soil loss for the whole area. In general, erosion rates are dominant over the sedimentation rates although both react differently to the implemented scenarios. In this case erosion rates increase some 8%, while sedimentation rates increase up to 64% after introducing landscape dynamics (Scenario D). Comparing all scenarios, the largest change occurs for erosion from Scenario B to C, indicating the effect of increasing run-off rates on steep slopes with shallow soils. Implementing the multiple flow directions resulted in the largest change in sedimentation rates between Scenario A and B. This is clearly an effect of increasing flow divergence and therefore lowering the dominant down-slope flow gradient and increasing the lengths of flow paths (Schoorl et al., 2000).


View this table:
[in this window]
[in a new window]
  		Table 2. Mean annual erosion and sedimentation in megagrams per hectare and percentages of change between scenarios.

 
To find out where these changes occur in our study area, we have plotted differences larger than 5% in maps of both erosion (Fig. 4) and sedimentation (Fig. 5) between different scenarios. Figure 4a shows the shifting erosion patterns from Scenario A to B. Changes occur randomly over the whole area more or less consistent with the shifting of the drainage patterns indicated in Fig. 2b, stressing the importance of flow routing algorithms used in modeling (e.g., Quinn et al., 1991; Desmet and Govers, 1996; Tarboton, 1997; Yin and Wang, 1999). Sedimentation patterns in Fig. 5a show the dominant increase related to the improved simulation of divergent properties of the landscape.



View larger version (131K):
[in this window]
[in a new window]
  		Fig. 4. Changes in soil erosion patterns, shown with 5% decreasing or increasing threshold for (a) Scenario A to B, (b) Scenario B to C, (c) Scenario C to D and (d) total change from Scenario A to D.

 


View larger version (77K):
[in this window]
[in a new window]
  		Fig. 5. Changes in sedimentation patterns, shown with 5% decreasing or increasing threshold for (a) Scenario A to B, (b) Scenario B to C, (c) Scenario C to D and (d) total change from Scenario A to D.

 
Introducing initial soil depths from Scenario B to C (Fig. 4b) shows a general trend of larger areas with more erosion (the higher areas of Fig. 1) and the lower areas with less erosion. As indicated above, this is the effect of the shallower soils on the steeper slopes that cause more run-off and generate more stream-power. An unexpected effect is that not only soil units as whole are affected by either increased or decreased erosion, the off-site effects also influence areas located down-stream. For example increased erosion is found along major tributary channels (Fig. 4b) and increased sedimentation is found in the central river area (Fig. 5b). Sedimentation shows on-site effects on local spots with isolated increased sedimentation rates due to excessive sediment delivered to these locations and changes in channel gradients (Fig. 3b).

Patterns of change after implementation of different Kes factors, which represent mainly the erodibility of the parent material, are shown in Fig. 4c and 5c. This case represents the opposite effect than in Fig. 4b. Lower areas show increased erosion while the higher areas reveal less erosion (Fig. 4c). Also sedimentation decreases although not for the soil unit as a whole but much more segregated on the slopes (Fig. 5c). This is the effect of the softer and more erodible parent materials on lower positions in the landscape, while harder parent materials including bedrock are situated in the higher elevated areas. Therefore, when less soil material can be detached upslope, it causes less soil to be redeposited on the lower slope positions.

Finally the maps of Fig. 4d and 5d show the cumulative results of changes from scenario A to D, implementing the most dynamic scenario in our simulations. Although areas of increased erosion dominate in Fig. 4d, this effect is compensated by an increased sedimentation in the RI (Fig. 5d) and decreased erosion in the upland areas giving the same net soil loss for both scenarios (Table 2). Thus, without spatial explicit information about on-site and off-site effects, general indications of changing erosion or sedimentation rates remain difficult to interpret. With adequate field data the present methodology offers the possibility to locate and quantify these landscape processes. For example, the 137Cs-technique (Walling and Quine, 1990) can provide landscape wide soil redistribution data at the adequate temporal resolution (years and decades).

Available Soil Water
Spatial distribution of changes in available water (AW) in the study area is presented in Fig. 6 for Scenario D. The majority of areas with a decreasing AW are located in the north and along tributary channels especially in the GN and SP area. Here the shallow soils are more prone to soil loss, as opposed to increases in AW of more than 10% in the central river valley. To illustrate the effects of our scenarios, four major soil units were chosen comprising different parent materials (Table 3). Since the AW is related to effective soil depth, we have only considered Scenarios C and D here.



View larger version (120K):
[in this window]
[in a new window]
  		Fig. 6. Changes in available water storage capacity for Scenario D compared with Scenario A.

 

View this table:
[in this window]
[in a new window]
  		Table 3. Initial mean soil depth available water (AW) as a percentage of soil volume, AW mean amounts for initial soil depths and changes in AW and soil loss after 10 yr of simulation for the study region and landscape units.

 
Comparing different lithologies under Scenario C (Table 3), GN shows greater soil and AW losses, the latter up to 0.6 mm, while the MO and MA show less soil and AW losses than the mean regional soil loss rates (e.g., soil loss is erosion plus sedimentation, see Table 2), indicating an increased net soil loss from these areas. Under Scenario D, however, soil and AW losses are increased for the MA and decreased for the MO. At the RI very high sedimentation rates, or a net soil and AW gain were found under both Scenarios C and D.

Again these trends indicate that within a dynamic landscape, different soil units attribute in their own way to the overall picture. For example, the soil AW increases in the RI where soils are already deep and well developed (Table 3). Since this area already has more than 100 mm available water for vegetation these changes will not be significant. This is in contrast, however, with other areas such as SP and GN that are loosing valuable topsoil and AW. These decreases can be serious if the soil depth drops below the threshold for vegetation growth (e.g., Kosmas et al., 2000).

This study indicates that each soil unit should be taken into consideration in the landscape context. Comparing for example MO and MA, the latter unit has much deeper soils and is therefore less sensitive to a small change (Table 3). In this case, the GN is most vulnerable with shallow soils and high erosion rates. Consequently, the mean AW is likely to be depleted rapidly. Although the loss of AW seems hardly serious as it appears only to be a few percent, we have to consider that these values are the means of the entire area and the local variability can be detrimental at some sensitive locations.

A selected catena in the GN is also presented in Table 3. In this case, the upslope area shows low rates of change (GNup), with an AW loss of 0.1 mm. Here run-on from other areas is not important and only local factors affect soil loss. Data in Table 3 show significant differences between concave (GNmc) and convex (GNmv) midslope positions. Concave positions receive more run-on, which increases the local transport capacity and therefore show higher soil losses. The downslope position is associated with an active eroding channel since soil and AW losses are extremely high, the latter more than 6.5 mm.

Differences among the different scenarios are closely linked to on-site and off-site effects. Under Scenario D, the upslope position shows a decreased on-site erosion and AW (Table 3). Consequently, the off-site effect is that at both midslope positions, there is more soil and AW losses (0.02–0.03 mm) because the transport capacities are not completely used upslope. These kinds of patterns on the concave and convex slope positions are also found by Gessler et al. (2000) for soil properties such as soil C and soil bulk density. Even though their study is much more quantitative and more detailed, this shows that within a dynamic model a coarser resolution does not necessarily mean loss of information.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
At the landscape scale, different spatial inherent properties and physical processes can contribute to the changing run-on, run-off, erosion, and sedimentation patterns. In this study, we distinguish four major effects causing these changing patterns: (i) the differences in generation of run-on and run-off from different flow-routing algorithms; (ii) the effect of changing infiltration patterns associated with initial soil depth; (iii) the effect of soil erodibility; and (iv) the changing soil depths and topography during simulation. The first and last effects are more related to modeling techniques, while the other two effects have to do with the implemented boundary conditions. However, they all imply both on-site and off-site effects in the landscape and different impacts for erosion and for resedimentation. Realistic modeling of processes in dynamic landscapes should at least include these effects.

When discussing and evaluating effects of soil redistribution one should always consider the landscape context. A net soil loss from a watershed does not reveal important spatial variation and implications. Erosion can become critical in one area while other areas can benefit from the net sedimentation. In the case of available soil water, it is important to know the interplay of soil depth, storage, and infiltration at the soil profile level, and run-on and run-off, slope and parent material in their specific position in the landscape. Therefore, soil profile, catena, or hillslope investigations will have to be combined into usable data for landscape analysis on the higher aggregation levels as discussed in this paper. Since quantitative data are difficult to obtain (by existing measuring techniques) and as a consequence calibration and validation are hampered on coarser spatial and temporal scales, the use of scenarios and modeling can provide a spatially explicit background for qualitative evaluation.


   ACKNOWLEDGMENTS
 
This research used the database of the interdisciplinary practical Sustainable Land Use in the Álora region, Dr. Willem Wielemaker and all staff and participants are greatly acknowledged. We sincerely thank several anonymous referees for their valuable comments and suggestions.

Received for publication February 27, 2001.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (9)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Schoorl, J. M.
Right arrow Articles by Bouma, J.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Schoorl, J. M.
Right arrow Articles by Bouma, J.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Schoorl, J. M.
Right arrow Articles by Bouma, J.
Related Collections
Right arrow Watershed and Landscape Processes
Right arrow Soil Erosion
Right arrow Water Content

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Journal of Natural Resources
and Life Sciences Education		 Vadose Zone Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome