Utilisation de la télédétection pour améliorer la précision des crues de fonte de neige simulées par le modèle CEQUEAU

Use of remotely sensed data to produce more accurate simulations of snowmelt floods with the CEQUEAU model ABSTRACT Simulations of the 1974 and 1975 snowmelt floods with the CEQUEAU model, on a 629 km2 basin in Quebec , show that simulation errors are not only due to models, but also to available climatological and snow course data. A first simulation, based on climatological data alone, is characterized by a daily snowmelt slightly too early for both years, together with an overestimated snowpack in 1975. The daily snowmelt is first corrected, by comparing maps of the water equivalents of the snowpack printed by the model, to visually interpreted Landsat or NOAA-4 imagery, for specific dates. A reduction of the water equivalent of the snowpack, using snow course data, had to be done only in 1975, just before the beginning of the snowmelt period, so as to obtain a better hydrological budget. The potential of using remotely sensed data in hydrological modelling is confirmed.