Hyperspectral Data Processing For Mineral Mapping Using AVIRIS 1995 Data in Alto Paraíso de Goiás, Central Brazil

Abstract. Hyperspectral remote sensing data were acquired in some areas of Brazil in 1995, using NASA's Advanced Visible and Near-Infrared Imaging Spectrometer (AVIRIS). This paper examines the performance of image processing techniques specifically developed for hyperspectral data in the context of mineral mapping applications. The area chosen, located in Goiás State, Central Brazil, near the locality of Alto Paraíso de Goiás, had relatively sparse vegetation at the time of imagery acquisition (dry season). The regional geology comprises a variety of fluvial and marine Precambrian metasediments. Image processing was carried out in two steps: (i) data conversion from radiance units to reflectance using a MODTRAN-based radiative transfer method and; (ii) application of the Tricorder mapping algorithm, specifically designed for identifying superficial materials based on similarities between image pixels and spectra from a spectral library. The results show that hyperspectral remote sensing can add significantly to the amount information that can be extracted from multispectral sensors, such as the Landsat TM, and is a valuable tool for mapping and exploration in sub-tropical regions