Articles | Volume 18, issue 11
Research article
28 Nov 2014
Research article |  | 28 Nov 2014

Effect of meteorological forcing and snow model complexity on hydrological simulations in the Sieber catchment (Harz Mountains, Germany)

K. Förster, G. Meon, T. Marke, and U. Strasser

Abstract. Detailed physically based snow models using energy balance approaches are spatially and temporally transferable and hence regarded as particularly suited for scenario applications including changing climate or land use. However, these snow models place high demands on meteorological input data at the model scale. Besides precipitation and temperature, time series of humidity, wind speed, and radiation have to be provided. In many catchments these time series are rarely available or provided by a few meteorological stations only. This study analyzes the effect of improved meteorological input on the results of four snow models with different complexity for the Sieber catchment (44.4 km2) in the Harz Mountains, Germany. The Weather Research and Forecast model (WRF) is applied to derive spatial and temporal fields of meteorological surface variables at hourly temporal resolution for a regular grid of 1.1 km × 1.1 km. All snow models are evaluated at the point and the catchment scale. For catchment-scale simulations, all snow models were integrated into the hydrological modeling system PANTA RHEI. The model results achieved with a simple temperature-index model using observed precipitation and temperature time series as input are compared to those achieved with WRF input. Due to a mismatch between modeled and observed precipitation, the observed melt runoff as provided by a snow lysimeter and the observed streamflow are better reproduced by application of observed meteorological input data. In total, precipitation is simulated statistically reasonably at the seasonal scale but some single precipitation events are not captured by the WRF data set. Regarding the model efficiencies achieved for all simulations using WRF data, energy balance approaches generally perform similarly compared to the temperature-index approach and partially outperform the latter.

Short summary
Four snow models of different complexity (temperature-index vs. energy balance models) are compared using observed and dynamically downscaled atmospheric analysis data as input. Biases in simulated precipitation lead to lower model performance. However, simulated meteorological conditions are proven to be a valuable meteorological data source as they provide model input in regions with limited availability of observations and allow the application of energy balance approaches.