The Diffuse Reflectance Spectroscopy (DRS) technique is an alternative method for soil attributes evaluation with a proven efficiency, low cost and environmental impact. Several studies, using different statistical methods, have shown that the selection of spectral bands or regions can help the performance of the models for soil attributes quantification. Hence, the objective of this study was to evaluate the spectral regions that allow discrimination between geomorphic surfaces (GSs) and soil attributes in mapping purposes for minimum areas of specific management. Soil samples were collected at transect every 25 m (100 samples) and by its sides (100 samples). Geomorphic surfaces (GSs) were mapped within the entire field area. Soil samples (collected from 0.00 to 0.20 m depth) were taken to the laboratory for physical, chemical, mineralogical, magnetic susceptibility and diffuse reflectance spectroscopy analyzes. Data were analyzed by statistical and multivariate geostatistics methods. The PC 1 and 2 scores of the studied spectral regions presented spatial dependence and the PC 2 spatial distribution maps of the entire spectrum and of PC 1 VIS region showed similar limits to those of the GSs. The attributes that have most correlated to PC 2 from the entire spectrum and PC 1 VIS region were clay and iron oxide extracted by dithionite-citrate-bicarbonate. The best spectral regions that allow to discriminate between geomorphic surfaces and soil attributes for minimum area of specific management mapping are the visible region and the entire spectrum.