Boolean and fuzzy logic operators and multivariate linear regression applied to airborne gamma-ray spectrometry data for regolith mapping in granite-greenstone terrain in Midwest Brazil

Abstract

Recent studies have proposed several possibilities of combining K, Th and U airborne gamma-ray spectrometry channels to generate predictive algorithms maps. These algorithms can be used for mapping regolith in deeply weathered terrains with residual, erosion, and deposition surfaces helping to developed strategies to better understand the regolith landscape and to improve geomorphology interpretation, and to identify mineral exploration target sites for primary (bedrock or saprolite) or supergene (hosted in lateritic duricrust) ore deposits. With the goal to easily map the regolith, two mathematical procedures were used on airborne gamma-ray spectrometry data in GIS software, validated by fieldwork on granite-greenstone belts in Midwest Brazil: 1. airborne gamma-ray spectrometry and altimetric data integrated in Boolean and fuzzy logic allowed segregating the areas with ferruginous and manganese residual lateritic duricrusts from erosional surface with rocks and saprolite with 90% of accuracy (κBoolean = 0.69 and κFAPO = 0.66) and 2. airborne gamma-ray spectrometry and altimetric data integrated with weathering stages in multivariate linear regression (basic statistic) helped establish the regional weathering intensity index, with acceptable error (r2 adjusted >0.6 and p-value < 5%). These two modeling techniques provide useful, accurate, rapidly and complementary regolith maps and can be applied in large regions for preliminary interpretations.