SOIL MINERALOGY ANOMALY DETECTION IN DIXIE VALLEY, NEVADA USING HYPERSPECTRAL DATA

Abstract

Anomalous soil mineralogy can indicate buried geologic structures and zones of permeability. Hyperspectral data can be used to map these anomalies in geothermal exploration efforts. This paper describes work done at Dixie Valley, Nevada, in this regard. Dixie Valley, which lies in the Great Basin, west-central Nevada, is host to a structurally controlled deep-circulation geothermal system. Caithness Energy, LLC operates a 65 Megawatt geothermal power plant in the northwest corner of the valley. This study took advantage of archive AVIRIS (Advanced Visible and Infrared Imaging Spectrometer) airborne hyperspectral imagery of the area adjacent to the power plant. Two spectral unmixing methods were employed to separate minerals from other components in a given pixel (ground unit in computer terms). The first was a supervised learning method (which requires the use of a training data set of known mineral spectra). The second method, polytopic vector analysis (PVA), is an unsupervised classification method, which allows spectral fingerprints to be derived based on analysis of ambient data. Both methods produced useful spectral end-members. The supervised methodology facilitated mapping of a strong calcium carbonate anomaly that exists in an area where fumaroles appeared shortly after the image was acquired.