Application of gamma-ray spectrometer data for lithological mapping in a cordilleran environment, Sekwi Region, NWT

Abstract

Gamma-ray spectrometry, magnetic, and Landsat data are evaluated for producing a predictive lithological map of a cordilleran environment (Sekwi region) in the Northwest Territories using supervised classification. Two approaches for defining training areas were used to conduct a maximum likelihood classification of the various datasets separately and in combination. The first approach involved defining training areas on an existing geological map in concert with evaluation of spectral, radiometric, and magnetic signatures. The second uses field locations as training areas. Validation of the classified maps was performed by comparing them to check training areas and existing geological maps. Both training methods produced similar predictive maps (classifications), and in this geologic environment from an individual perspective the gamma-ray spectrometry data produced more accurate results than the magnetic and Landsat data. However, the “best” map was produced using all the gamma-ray data, together with the residual magnetic total field data, and the Landsat data, supporting the notion that a variety of geoscience data, each responsive to different characteristics of rocks (spectral reflectance, radioelement concentrations, and magnetic susceptibility), provides the most accurate predictive map.