Victor Jetten

Evaluation of field-scale and catchment-scale soil erosion models

Abstract One of the tasks of the International Geosphere–Biosphere Programme–Global Change and Terrestrial Ecosystems (IGBP–GCTE) Soil Erosion Network is to determine the suitability of modelling approaches for the estimation of soil erosion under global change. To achieve this, current erosion models are being evaluated in a series of GCTE meetings. This paper presents a synthesis of results from the first two meetings, which focused upon models for soil erosion by water at field and catchment scales, respectively. Apart from this comparison, discussions were held on model use and quality. The main conclusions from these discussions are included here as well. For both sets of evaluations, common datasets which had been split into a `training set and a `testing set were prepared and distributed to the participating modellers. For the `testing set data, measured values for runoff and sediment yield (field-scale models) or for erosion only (catchment-scale models) were withheld from the modellers. The data used for the field-scale evaluation represented 73 site-years from seven sites in three countries: six field-scale erosion models took part in the evaluation. For the catchment-scale evaluation, data for 10 events on a 40 ha catchment in the Netherlands was used to evaluate seven catchment models. Conclusions from both field-scale and catchment-scale exercises include the following: • calibration is desirable for many models, and necessary for some. Calibration is most effective if the event(s) to be estimated lie inside the range of calibration events; • total discharge is generally better predicted than peak discharge and both are better predicted than sediment discharge; • for continuous-simulation models, long-term average results are better simulated than results for individual time periods. In general, results are less good for shorter time periods, although there are exceptions; • while for certain events models may not perform well (absolute results), correlation coefficients between observed and predicted values are acceptable (relative results); • at a catchment scale, the predicted spatial runoff pattern is as important as a correct prediction of the net output; • from the discussions it was clear that additional `soft information, in particular regarding the change in soil structure as a result of agricultural activities and/or climate, greatly improves the quality of input data and model results.

Calibrating and validating the LISEM model for two data sets from the Netherlands and South Africa

Abstract Within the framework of the GCTE Soil Erosion Network the Limburg Soil Erosion Model (LISEM) has been tested and validated in two catchments in South-Limburg (the Netherlands) and Zululand (South Africa). The calibration and validation of the Green–Ampt version of the LISEM model using 10 storms from the Catsop catchment in the Netherlands shows that differences in measured and simulated hydrographs and sediment loads can be large. These differences may be caused by the sensitivity of the model to some of the input variables, such as saturated hydraulic conductivity and the initial soil moisture content. Given the uncertainty in the input maps of these variables and a limited number of point data that is used to create these maps, a large part of the differences between measured and simulated data can be explained by these uncertainties. Thus, it is clear that detailed process-based models such as LISEM require very detailed and high-resolution input data in order to produce quantitative reliable results. The runoff processes in the Zululand catchment appeared to be dominated by slow throughflow and groundwater flow, which are processes that are not incorporated in LISEM.

Effects of tillage on runoff and erosion patterns

Historically, research on water runoff and soil erosion has been conducted mainly at the plot scale, with slope gradient being one of the dominant controlling factors. However, tillage-induced roughness can also significantly impact runoff and erosion, and therefore runoff patterns on a field scale can be very different from the runoff pattern that would be predicted from topography alone. A model and methodology were developed to account for oriented tillage effects on runoff patterns. Results suggests that the prediction of soil erosion and deposition within a field, as well as total soil loss from a field, can be significantly improved by combining the effects of tillage-induced runoff patterns with detailed topographic data.

Interception of tropical rain forest: performance of a canopy water balance model

The throughfall in dry evergreen forest and mixed forest in central Guyana (South America) was measured over a period of 17 months using both a moving and a fixed stratified random sampling scheme. The throughfall, stemflow and canopy cover fractions of the two forest types were not significantly different. The data set was used to test the canopy water balance model designed by Rutter et al. (1971). Compared with both forest types, the model underestimated the total interception fraction: 13.4% for the Rutter model as opposed to 17.3 and 16.0% of the rainfall for dry evergreen forest and mixed forest. This was caused by a severe underestimation of interception on days with a rainfall of more than approximately 10 mm. Compared with the rainfall frequency distribution of the area, accurate model results were obtained for 20% of the rainfall only. Sensitivity analysis shows that an increase of saturation storage capacity or potential evaporation by 150% will give an acceptable accumulated interception, but that this is caused by large over-estimation on days with a low amount of rainfall. The Rutter model was extended with a layered representation of the canopy, with simulation of the microclimate, but using basically the same input data. This resulted in a higher total interception fraction (16.3% of the rainfall), especially in the range 5-20 mm/day, because of a larger storage at the start of each rainstorm and an increased surface available for evaporation.

The effect of tillage-induced roughness on runoff and erosion patterns

In agricultural areas, tillage-induced roughness may have large impacts on runoff patterns and therefore on the effective slope, local contributing areas and erosion patterns. A model was developed to create a runoff pattern with flow in the direction of the plough-lines for all tilled fields within a catchment. The model needs a digital elevation model (DEM), a landuse map and the major tillage orientation per tilled field as input. Optionally, the flow direction of roads and/or channels can be superimposed on the drainage network. The model calculates flow directions along parcel borders and creates headlands along borders that are not parallel to the tillage orientation. The model also calculates the slope gradient in flow direction. The created tillage-controlled runoff pattern can be combined with the topographic runoff pattern, if decision rules are available to choose between topographic or tillage direction. The tillage runoff pattern can be very different from the topographic pattern, leading to totally different patterns of slope and contributing area. The use of the tillage-controlled runoff pattern rather than the topographically controlled runoff pattern in a deterministic, event-based model (LISEM), results in a much better agreement of the predicted runoff and erosion pattern with field observations. However, accurate model predictions can only be obtained if the input DEM is sufficiently accurate to represent local topographic variations in the fields.

Calibration of the LISEM model for a small loess plateau catchment

The Limburg Soil Erosion Model (LISEM) soil erosion model was calibrated for a 2-km2 catchment on the Chinese Loess Plateau. The most important calibration factors were saturated conductivity and Mannings n. Calibration on catchment discharge was done by using the discharge peak (timing and discharge) followed by an adjustment for the total discharge to obtain the correct amount of sediment output. The results showed that LISEM can be successfully calibrated for a Loess Plateau catchment, and that small runoff events need to be calibrated separately from large runoff events. A separate calibration might even be needed for each event. The model performance was also evaluated using catchment wide spatially distributed data on rill erosion. Rill erosion intensity was mapped in the field and compared to spatial patterns of erosion predicted by LISEM. The simulated erosion patterns do show some resemblance with mapped erosion patterns in a general sense but they are very different in detail. The cause for this can be found in the extremely steep slopes and abrupt slope changes in the catchment. Some of the process descriptions in LISEM are not intended for such an environment, while the grid based kinematic wave routing cannot cope with the abrupt changes in flow conditions. The effects of this are amplified by inaccuracies in the input data and the DEM. For topographically complex catchments it will be very difficult to obtain data that are good enough for an accurate simulation of erosion patterns. This limits the use of a model such as LISEM as a predictive tool for future events. Simulation of different land use scenarios is less problematic, if a known event is used for all scenario simulations.

Spatial Analysis of Erosion Conservation Measures with LISEM

Runoff and erosion models are generally used to assess environmental problems such as soil erosion problems with loss of fertile soil and damage to crops, off-site damage to property and infrastructure by “mud-flows,” and pollution of surface water by sediment with agricultural chemicals and nutrients. These problems occur frequently in the loess zone in Western Europe (Boardman et al., 1994) of which Limburg, the southern province of the Netherlands, forms a small part. The Limburg Soil Erosion Model LISEM, (De Roo et al. 1996a, 1996b; LISEM, 2000) is a physically-based hydrological and soil erosion model, operating at the catchment scale, that was designed to assess these problems. The model simulates runoff and erosion with single rainstorms in agricultural catchments of a size ranging from 1 hectare up to approximately 10 km2. The upper limit size is determined by the fact that in LISEM a stream channel cannot be larger than one pixel; larger catchments with floodplains and river systems cannot be simulated.

Can infrared spectroscopy be used to measure change in potassium nitrate concentration as a proxy for soil particle movement

Displacement of soil particles caused by erosion influences soil condition and fertility. To date, the cesium 137 isotope (137Cs) technique is most commonly used for soil particle tracing. However when large areas are considered, the expensive soil sampling and analysis present an obstacle. Infrared spectral measurements would provide a solution, however the small concentrations of the isotope do not influence the spectral signal sufficiently. Potassium (K) has similar electrical, chemical and physical properties as Cs. Our hypothesis is that it can be used as possible replacement in soil particle tracing. Soils differing in texture were sampled for the study. Laboratory soil chemical analyses and spectral sensitivity analyses were carried out to identify the wavelength range related to K concentration. Different concentrations of K fertilizer were added to soils with varying texture properties in order to establish spectral characteristics of the absorption feature associated with the element. Changes in position of absorption feature center were observed at wavelengths between 2,450 and 2,470 nm, depending on the amount of fertilizer applied. Other absorption feature parameters (absorption band depth, width and area) were also found to change with K concentration with coefficient of determination between 0.85 and 0.99. Tracing soil particles using K fertilizer and infrared spectral response is considered suitable for soils with sandy and sandy silt texture. It is a new approach that can potentially grow to a technique for rapid monitoring of soil particle movement over large areas.