Evaluating the snow component of a flood forecasting model

The objective of this study is to evaluate the snow routine of a semi-distributed conceptual water balance model calibrated to streamflow data alone. The model is used for operational flood forecasting in 57 catchments in Austria and southern Germany with elevations ranging 200–3,800 m a.s.l. We compared snow water equivalents (SWE) simulated by the hydrologic model with snow covered area (SCA) derived from a combined product of MODIS (version 5) Terra and Aqua satellite data for the period 2003–2009 using efficiency measures and a spatial analysis. In the comparison, thresholds for percent catchment snow cover and a cut-off water equivalent need to be chosen with care as they affect the snow model efficiency. Results indicate that the model has a tendency to underestimate snow cover in prealpine areas and forested areas while it performs better in alpine catchments and open land. The spatial analysis shows that for 88% of the analysed model area snow cover is modelled correctly on more than 80% of the days. The space borne snow cover data proved to be very useful for evaluating the snow model. We therefore suggest that the snow data will be similarly useful for data assimilation in real time flood forecasting.