Vol. 4, No. 1, 106-118, 2005

Abstract
This work comes within the scope of a multidisciplinary study aimed at validating the hydrological simulations of the Canadian regional climate model over Quebec (Canada). Snow cover is a key factor in the modelling process. Because of their low density, conventional local observation networks do not provide enough accurate data to map snow cover on a large scale and with an adequate spatial resolution for regional climate modelling. Alternatively, this is easily feasible using visible and infrared satellite imagery. However, available satellite snow cover products are unusable for our special needs, because they either have an inadequate spatial resolution or too short observation series.

The objective of this study was therefore to develop an automatic algorithm for snow cover extent mapping using data from the AVHRR sensor on board NOAA satellite series, which allows monitoring the space-time evolution of snow cover extent over a long period of time and with a "fine" spatial resolution (1x1 km²). Snow cover extent mapping results were validated against in situ snow occurrence observations. The algorithm was tested over the province of Quebec (Canada) for three specific periods: 1998-1999, 1991-1992 and 1986-1987. The algorithm identifies surface class (snow/no-snow) with an average total success rate of 87%. The algorithm performances were higher in snow detection (90%) than they were for no-snow surfaces (82%).

ISSN 1729-3782