Hillslope dynamics modeled with increasing complexity

Few studies have investigated how much model complexity is needed to simulate both the hillslope outflow and the internal hillslope dynamics. We studied the influence of model complexity on simulations for the Panola Mountain trenched hillslope. We analyzed the influences of the inclusion of bedrock permeability, variable soil depth and preferential flow on modeled hillslope responses. We found that without the inclusion of bedrock leakage the long-term subsurface flow response measured at the trenchface and the threshold relation between total precipitation and total subsurface flow could not be simulated adequately. Individual events could still be represented acceptably, showing the importance of long time series for model calibration and validation. The use of spatially constant bedrock conductivity allowed us to simulate spatially variable bedrock leakage rates because of the spatially variable depths of saturation. Without variable soil depth the spatial variability of subsurface flow along the trenchface and its temporal dynamics during events could not be represented. In addition the spatial patterns of saturation at the soil-bedrock interface did not agree with the observed patterns and responses to smaller events were underestimated. Inclusion of preferential flow mainly influenced the distribution of the maximum saturation depths at the soil-bedrock interface and increased peak flows and recessions. Soil moisture measurements were less useful for model validation for the Panola hillslope than measurements of the spatial patterns of saturation and subsurface flow. We plea for a new blue print for the set-up of hillslope experiments such that their data is useful for studies on hillslope model complexity and for model validation and rejection.