Areas with deep regolith profiles are widely distributed in the Amazonian region. Research on regolith geology in Brazil has concentrated, mainly, in geochemical and mineralogical aspects and where possible utilizing these in exploration. Although airborne geophysics and spatial data integration methods are being used increasingly for mineral resource assessments, the application in regolith-dominated terrains is new. The objective of this paper is to show the potential of a systematic approach, including the use of regional gamma-ray spectrometry, digital elevation model and image digital processing techniques for the identification of deep regolith areas characterized by potential economic interest in less well explored region of the eastern portion of Amazonas State. The application of multivariate statistics techniques (Principal Components Analysis and K-means) in airborne gamma-ray data allows the individualization of regoliths in comparison to the geologic units of this region. How exploratory guide, the following patterns were observed: low values of K (easily lixivied during the weathering) and K/eU ratio; high values of eTh, eTh/K ration and medium values of eU, besides the higher altitude (plateau). Based on these images, the hyperspectral algorithm named Spectral Angle Mapper (SAM) and the boolean and fuzzy logic spatial modeling techniques were applied to produce regolith potential maps that were compared with previous geological data in 1:250.000 scale. The quantitative relationships between geophysical and altimetric datasets have shown great capability for linking geological and geophysical interpretations, indicating potential areas that can direct new ways for mineral prospecting at the study area.
airborne gamma-ray spectrometry data; SRTM; Spectral Angle Mapper; spatial data modeling; Principal Components Analysis; K-means
