Modelling soil erosion in a small catchment on the Chinese Loess Plateau : Applying LISEM to extreme conditions

The Chinese Loess Plateau suffers some of the highest soil erosion rates on earth. This is caused by the fact that it can be characterized as a semi-arid area with low vegetation cover, erodible soils, steep slopes and occasional high intensity summer storms. In this study, the process based distributed soil erosion model LISEM was applied to a small catchment on the Chinese Loess Plateau. Several characteristics specific to the Loess Plateau needed to be taken into account to allow successful simulation with such a model. The most important were the occurrence of very high sediment concentrations, the occurrence of steep slopes and the presence of large gullies. High concentrations will affect fluid properties, steep slopes have effects on flow velocity and transport capacity, while in the gullies erosion processes (like soil fall) operate that are usually not considered in soil erosion models. A number of changes to LISEM were proposed to overcome the modelling problems posed by the steep slopes, high concentrations and permanent gullies of the Danangou catchment. These changes included a slope angle correction, the use of slope dependent Manning's n, the introduction of a concentration dependent settling velocity, the introduction of a loose material map and the use of sine instead of tangent. These changes were implemented in LISEM and evaluated on the hydrograph and sedigraph at the catchment outlet. Predicted discharge decreased by about 50% from applying a slope correction for the calculation of overland flow, but was only marginally affected by using a slope dependent value of Manning's n. Predicted concentration increased by applying a concentration dependent fall velocity, but was hardly changed by introducing a map with loose material or by using sine instead of tangent in the transport equations. However, the choice of transport equation had large effect on the simulation results. After recalibration the LISEM model simulated measured discharge and sediment yield only slightly better than before. LISEM was also found to be sensitive to changes in time step length and grid size. Calibration of the LISEM model for the Danangou catchment showed that LISEM can in principle be applied to the Chinese Loess Plateau. The results, however, also showed that a separate calibration was needed for low-magnitude and high-magnitude events and probably even for each event. Small events could not be calibrated well. A comparison of mapped and simulated erosion patterns showed that these patterns were very different in detail. The most likely causes are: process descriptions that are unsuitable for steep slopes, inaccurate data and effects caused by the raster-based approach itself. The effect of a number of land use and land management scenarios was also simulated with LISEM. The simulations predicted that implementing conservation methods would result in decreases of runoff and erosion of 5-20%. Changing the land use itself was predicted to have a much larger effect; discharge decreased by 40-50%, while erosion decreased by 50-70%. These large effects of land use change were mainly caused by extension of the woodland area. More research is needed before we can say to what degree the simulation results of LISEM reflect reality