Potentials and limitations of modelling spatio-temporal patterns of soil moisture in a high mountain catchment using WaSiM-ETH

Soil moisture is a highly dynamic ecological variable. This dynamic is especially distinct in high mountains areas, where topography, snow cover, micro-climate, vegetation, soil properties and land cover vary across short distances. For such areas, soil moisture modelling on a fine scale is therefore particularly challenging. We evaluated the distributed, physically based model WaSiM-ETH with regard to its ability to model soil moisture in high mountain catchments, including spatio-temporal patterns. The study was performed in the alpine catchment area of the Lotschen valley (160 km(2)) in Switzerland, an area ranging from 600 to 4000 m a.s.l. The model was run for the period of 2003-2007 with high temporal (hourly) and spatial (50 m) resolution. Extensive TDR- and FD-measurements conducted in 2006 and 2007 enabled the validation of the simulated soil moisture. A local sensitivity analysis indicated a dependence of soil moisture on skeleton fraction and temperature. The validation results showed that WaSiM-ETH is able to simulate absolute soil moisture variability across different altitudes and land cover types with only a moderate accuracy (R = 0.47), whereas the soil moisture dynamic is reproduced well (R = 0.69, IoA = 0.66). High spatial variability of skeleton fraction and coarse meteorological data are the main causes for the limitation in soil moisture modelling. Nevertheless, analysing the spatio-temporal patterns of soil moisture revealed a clear seasonal pattern that is in line with literature. We found soil moisture to be determined by liquid precipitation, snow-melt and evapotranspiration. Greatest efforts should be laid on the exact determination of fine-scale meteorological data and spatial skeleton estimation to improve soil moisture modelling in high mountain areas in the future. Copyright (C) 2010 John Wiley & Sons, Ltd.