MAPPING GEOLOGICAL AND GEOPHYSICAL FEATURES OF THE EARTH FROM SPACE MULTISPECTRAL DATA

MINERAL MAPPING USING MULTISPECTRAL REMOTE SENSING TECHNIQUES HAS ATTRACTED MANY INVESTIGATIONS SINCE THE EARLY WORK OF MULTISPECTRAL LUNAR MAPPING BY HUNT AND HIS COLLEAGUES AT THE AF GEOPHYSICS LABORATORY IN CAMBRIDGE, MASSACHUSETTS. MOST NOTABLE ARE THE ACTIVITIES OF LYON (STANFORD UNIVERSITY), SALISBURY, ROWAN AND BAILEY (USGS), GOETZ AND KAHLE (JPL), AND THE GEOSAT COMMITTEE INVESTIGATIONS. THE WORK OF THESE INVESTIGATORS ADDRESSED FEATURE IDENTIFICATION BY SPATIAL PATTERN RECOGNITION (TOPOGRAPHY, DRAINAGE, FRACTURES, ETC.), SPECTRAL PATTERN RECOGNITION (RESTORATION, BROADBAND REFLECTANCE, GEOBOTANICAL PHENOMENA, AND ALTERATIONS), AND TEMPORAL VARIATIONS (SEASONAL AND HEAT CAPACITY). SPECTRAL PATTERN RECOGNITION INVESTIGATIONS EMPLOY IMAGE PROCESSING TECHNIQUES SUCH AS BAND RATIOING, SUPERVISED AND UNSUPERVISED CLUSTERING AND CLASSIFICATION, PRINCIPAL COMPONENT ANALYSIS, AND CONTRAST ENHANCEMENTS. THIS STUDY BUILDS UPON THESE PAST ACTIVITIES AND WILL INVESTIGATE THE UTILITY OF THREE NEW ADDITIONAL APPROACHES TO GEOLOGICAL FEATURE ENHANCEMENT; USE OF TWO NOVEL TECHNIQUES FOR SUBTRACTING SPECTRAL RADIANCE ASSOCIATED WITH SPARSE VEGETATION COVER (VEGETATION REFLECTANCE AND SCATTERING), MERGING LANDSAT THERMAL WITH REFLECTANCE DATA, AND MAXIMIZING SPECTRAL CONTRAST THROUGH A UNIQUE METHOD EMPLOYING A BRIGHTNESS/GREENESS TRANSFORMATION. THE RESULTING IMAGERY WILL BE EVALUATED AGAINST A WELL DOCUMENTED TEST SITE IN THE WESTERN UNITED STATES.