In this letter, we propose the calibration procedure for a Snow Water Equivalent (SWE) forecasting model, using Moderate-Resolution Imaging Spectroradiometer (MODIS) multi-temporal snow cover maps and in situ measurements. The presented study refers to one of the largest artificial lakes in the Western Europe - the Serre-Ponçon reservoir, on the Durance river, in the region of the French Alps. The SWE model, an integral part of the MORDOR (MOdèle à Réservoirs de Détermination Objective du Ruissellement) hydrological model, provides SWE as a function of local precipitation and temperature, as well as of accumulation and melting correction coefficients. The principal motivation for the proposed calibration method comes from the significant model sensitivity with respect to these two coefficients, which, given that they account for the influences of topology and mountain winds, ought to vary spatially. Three different optimization procedures are compared using the set of in situ measurements acquired by the EDF (Eléctricité de france) cosmic-ray snow sensors for 4 out of 36 ground stations in the regions of interest. The appropriate optimization method is selected and the corresponding representative optimal coefficients are derived for these four stations. Further, by combining the selected optimization algorithm and the continuous activation function, we propose a new method for deriving the spatially varying coefficients characterizing the entire region, using multi-temporal MODIS snow cover binary maps. When analysed with respect to the mean square error (MSE) criterion, the SWE model, calibrated in this manner, appears to be significantly more accurate than the original version (using a priori estimated, spatially fixed coefficients). Furthermore, the calibration procedure based on MODIS data is comparable and, for some ground stations, exhibits even better performances than the one based on the in situ measurements.