Use of ground observations to simulate the seasonal changes in the backscattering coefficient of the subarctic forest

Abstract

RADARSAT synthetic aperture radar (SAR) data acquired at C Band, HH polarization, and for the 20/spl deg/-27/spl deg/ and 45/spl deg/-49/spl deg/ incidence angle ranges were available over northern Quebec, Canada, (54/spl deg/N, 72/spl deg/12'W), in the fall of 1996, the winter of 1997, and the spring of 1997. The main land occupation of this area is sparse black spruce (Picea mariana) forests. Vegetation characteristics are jointly used with snow and soil observations coinciding with the satellite overpasses to simulate the seasonal changes in the backscattering coefficient of the subarctic forest. The aim of this study is twofold. First to evaluate the effects of the seasonal changes in vegetation on the RADARSAT SAR data, and second to use backscattering models as a tool for a better interpretation and understanding of the RADARSAT SAR data over snow-covered forested areas. Simulations show the importance of the surface-vegetation interaction term and the wet snow surface roughness on the discrimination between open forest and denser forest, and on the contrast between wet snow and dry snow covers. When comparing the simulations to the RADARSAT SAR data, the poorest results are obtained in the spring for a rough wet snow. It is shown that they are mainly due to a crude evaluation of the vegetation dielectric constant rather than to uncertainties introduced by the spatial variability in the wet snow surface roughness.