This study aims at assessing the quality of the altimetric elevation models (DEM) generated by interferometric SAR (Synthetic Aperture Radar) bands X and P in different polarizations for forested areas. This study was carried out in Pindamonhangaba/SP region, in an area featured by a 6 year old Eucalyptus saligna reforestation, whose forest inventory and topographic survey were conducted in the same air-surveying period. It was found out that canopy DEM using interferometry in the X-band was less precise than expected, with 2.7 meter standard deviation. By using a regression model that combined interferometric coherence and DEM, made up for the vegetation height measurement errors improved the quality of the cartographic product for standard deviation about 1.4 meters, reaching the desired resolution. The ground elevation models in P band, HH polarization, for forested areas showed similarity to the surveyed topographic data, with a standard deviation of around 1.97 meters, due to a greater radar beam penetration at Eucalyptus forest and therefore a greater interaction with the soil than that of with the trees. The DEM in X-band for pasture areas had a 0.6 meter standard deviation, and the mapping scale was compatible with the resolution, whereas elevation model in P-band showed a standard deviation value in the order of 6 to 20 meters, according to polarization, due to the specular scattering causing a low signal/noise ratio.
SAR; interferometry; Eucalyptus; DEM
