Analysis of airborne visible-infrared imaging spectrometer (AVIRIS) data of the Iron Hill, Colorado, carbonatite-alkalic igneous complex

Abstract

The Iron Hill, Colorado, carbonatite-alkalic igneous rock complex consists mainly of pyroxenite, uncompahgrite, fenite and nepheline syenite, and carbonatite. The study area has been explored intermittently for several decades for thorium, niobium, rare earth elements, titanium, and vermiculite. Airborne visible-infrared imaging spectrometer (AVIRIS) data were acquired for the Iron Hill Complex on August 28, 1992, under clear sky conditions. The data were calibrated using a laboratory reflectance spectrum of a tuff-sagebrush mixture to represent the spectral response of a topographically flat, homogeneous image area within the ash-flow tuff. Analysis of the data was conducted using the spectral angle mapper (SAM) and linear spectral unmixing algorithms in the spectral image processing system (SIPS). Image spectra in the 2.1- to 2.45-mu m wavelength region were used in the SAM method to map dolomite in the Iron Hill carbonatite (rauhaugite) stock and in several smaller bodies, calcite in a small carbonatite (soevite) plug and in travertine spring deposits, illite mainly in sericitized felsite and granite, and kaolinite in hydrothermally altered ash-flow tuff. The general distribution of the main alkalic igneous rocks (pyroxenite, uncompahgrite, and ijolite) was mapped using a full AVIRIS wavelength range (0.40-2.45 mu m) image spectrum of pyroxenite, excluding channels affected by strong atmospheric absorption. However, some areas consisting of felsite and fenite were not spectrally distinguished from the alkalic rocks. Linear spectral unmixing, using full wavelength image spectra, was used to improve the compositional mapping achieved using the SAM algorithm. The fractional abundance images obtained for dolomite and calcite were similar to the SAM results, but the distinction between illite and kaolinite was significantly improved, especially where scattered coniferous trees are present. Hematite and goethite, which showed widely overlapping patterns in the SAM similarity images, were distinguished much better in the fractional abundance images. Biotite was used instead of pyroxenite as a reference end member in the unmixing analysis, and the results were better than those obtained in the SAM analysis...