The obtention of soil quantitative information by non destructive techniques, like remote sensing and digital elevation model (DEM) open a possibility of faster and cheaper evaluation on soil studies. This study aims at verifying the possibility of quantifying soil atributtes through ASTER orbital sensor data and topographic features. The study area, with 184 ha, located in Rafard County, São Paulo State, Brazil, was sampled in a grid way collecting 1 sample/ha. Reflectance data was obtained from ASTER image and topographic features from DEM. Then, using orbital an topographic data were generated a multiple regression equation for soil attributes. The dependent variables were selected by a forward stepwise method. After that, the models with R² > 0.50 were tested with unknown soil samples and the estimated values (VE) and the determined values (VD) were evaluated through scatterplots. SiO2, Fe2O3, TiO2 and clay attributes were tested and, Fe2O3 and clay obtained higher values of determination coefficient. It is conclude that the models generated with spectral data and topographic information allowed the quantification of the above cited soil components.
digital elevation model (DEM); ASTER image; soil attributes
