In mountain areas, and especially in the northern Alps, the improving of flood forecasting models is an important scientific and operational issue, due to environmental and climatic changes which modify hydrological system. Snow melt contributions are a specific difficulty in the hydrological modelling in the Alps. This thesis is concerning in improving snow melt modelling for hydrological forecasting. In a first step a snow melt model at a local scale is formulated, using a minimum of meteorological variables. For this the “degree day” model, which allows predicting the water release by snow function of air temperature, is used. Thanks to data of the Centre d'Etude de la Neige on the experimental site of Col de Porte (Chartreuse), several formulations have been tested with some interesting concepts, like the cold content of snow and the effect of seasons with a melt factor given with a sinusoidal expression. In a second step, we are interested in the regionalization of the most important parameter for the snow melt model: air temperature. The spatial interpolation of temperature in mountainous areas is done with geostatistical tools, focusing on a better understanding of the elevation lapse rate of temperature. The goal is mapping air temperatures at a sub daily time step and to propose a simple and relevant model. Prospects about these results of snowmelt modelling and temperatures interpolation are given for using them in mountainous catchments.
