Improving the estimation of erosion-related suspended solid yields in mountainous, non-alpine river catchments

Sediment yields in river catchments may be modelled using a range of digital elevation models (DEM) and data algorithms. We assessed the impact of alternative choices on the evaluation of empirical modelling approaches. The model framework in this study consisted of the universal soil loss equation, a published local sediment delivery ratio (SDR), and commonly available input data. The study area comprised the catchments of 31 monitoring gauges for which daily data of suspended solids (SS) was available. For these stations, we studied the effects of two interpolation schemes for daily SS concentrations and of two approaches to separate erosion-related ("critical") fractions from total SS yields (SY) on model evaluation. Despite a good agreement between modelled and critical SY, the unexplained spatial variability was considerable. For additional 109 catchments, we quantified impacts of two DEM resolutions, two slope algorithms, and three slope length algorithms on modelled SY. DEM resolution and slope algorithm proved to be most relevant for the model uncertainty. High correlation coefficients between the respective alternatives revealed the minor relevance of data or algorithm choices for model quality. Correlation analyses showed that the SDR model lacked a hydrological parameter. Adjusting the modelled SDR accordingly significantly increased the explained variability of SY.