Caractérisation de la température de la neige par télédétection micro-onde passive au Canada

Abstract

The understanding of the dynamics of the climatic variables is critical to model and predict climatic and environmental changes. Traditional measures collected by the meteorological stations network are dispersedly located throughout the territory in the northern high latitudes and errors associated to these variations can be considerable. Our goal is to evaluate the contribution of the remote sensing by passive microwaves compared to the ground measurements for better characterizing the variations in the surface temperature during the winter. In the presence of snow, extraction of the surface parameters by microwave measurements is a complex and ill-conditioned problem. We evaluated a semi-empirical relation based on a theoretical analysis to estimate the surface temperatures with the measured brightness temperatures at 19 and 37 GHz (vertical polarization). Simulations of emissivity made with the Helsinki Technology University (HUT) model, the knowledge of the land cover, and the forest biomass enable us to define this relationship.The results of the comparison between the calculated surface temperatures and the air temperatures for the entire Canada (137 stations) and for two winters, 1992-93 and 2002-03 (16359 measurements) show a significant correlation with an estimation error ranging between 4 and 7[degrees Celsius] according to the regions considered. These results are discussed according to region and type of land cover (grassland, forest, tundra).