Spectral Mixture Modeling of an ASTER Bare Soil Synthetic Image Using a Representative Spectral Library to Map Soils in Central-Brazil

Pedological maps in suitable scales are scarce most countries due to the high costs involved soil surveying. Therefore, methods for surveying and mapping must be developed overpass cartographic material obtention. In this sense, work aims at assessing a digital map (DSM) built by multispectral data extrapolation from source area target using ASTER time series modeling technique. For that process, eight representative toposequences were established two contiguous micro-watersheds, with total of 42 profiles analyses classification. We found Ferralsols, Plinthosols, Regosols, few Cambisols, Arenosols, Gleisols, Histosols, typical tropical regions. laboratory, surface samples submitted spectral readings 0.40 µm 2.50 µm. The spectra morphologically interpreted, identifying shapes main features soils. Soil texture grouped curves cluster analysis, forming library (SL). parallel, an (2001, 2004, 2006) was processed, generating bare synthetic image (SySI) covering 39.7% area. Multiple Endmember Spectral Mixture Analysis modeled SL on SySI DSM 73% Kappa index, which identified about 77% is covered rhodic Ferralsols. Besides overestimation, represented study area’s pedodiversity. Given discussion raised, we consider including subsoil other sensors operations machine learning algorithms improve results.