Image analysis of airborne geophysical data from the Salcha River – Pogo area, Alaska

Abstract

An airborne geophysical survey was conducted by the State of Alaska in the summer of 1999 in the Salcha River - Pogo area to support exploration for economically viable mineral deposits. We processed and analyzed a subset of the airborne geophysical data, including magnetic, apparent resistivity, K, equivalent Th (eTh), and equivalent U (eU) data, to further investigate this potentially resource rich area. We used unsupervised classification to analyze a data stack of geophysical data bands because that approach allows the data to self-organize into classes that may represent geologic features, with minimal a priori investigator-imposed class definition constraints. The image data were rescaled to a common range of values and kept in floating-point format, and pixels corresponding to valley areas were masked to reduce topographic and soil-moisture effects. Classification results were interpreted by comparison to published maps of the area. The final classification successfully differentiates conductive and nonconductive lithologies, reliably classifying carbonaceous phyllite and meta-argillite units; delineates mapped and potentially unmapped mafic to ultramafic and meta-mafic rock assemblages; discriminates major regional high-angle fault zones; and identifies what we infer to be evidence of previously unrecognized major low-angle geologic structures. Although unsupervised classification of the geophysical spectral data was unable to resolve the geology of large areas of the survey tract, it does generate potentially useful classes that can be interpreted. Hypotheses formed from these interpretations suggest new geologic features and mineral prospects in the survey tract that are amenable to testing in the field.