Simulation of Discharge using Different Methods of Meteorological Data Distribution, Basin Discretization and Snow Modelling

Abstract

This contribution investigates the performance of discharge modelling when proceeding from rather simple methods of snowmelt modelling to more sophisticated, distributed approaches. In all cases, the conceptual HBV runoff model is used for the calculation of discharge from the Romanche River basin (French Alps, area = 224 km²,12.5 % glacierized), feeding the Lac de Chambon reservoir used for hydroelectric power production. The results show that when using temperature-index approaches, no gain in performance is achieved by the more detailed basin discretization. However, a clear gain is observed when the precipitation and air temperature distribution of the elaborate meteorological data analysis system “SAFRAN” is used as input to the conceptual snow model. When using the original version of the physically-based “CROCUS” snow model, a strong overestimation of discharge resulted during the main snow melt season. A revised SAFRAN model version employing a new radiation routine, and the use of a realistic distribution of snow cover in each sub-unit considered by CROCUS resulted in a discharge simulation performance which is as good as the one from the calibrated conceptual models. Sensitivity analyses of the various snow models investigating the hydrological consequences of climatic change show surprisingly similar results.