Gypsum classification based on ASTER images in the Principal Cordillera of Mendoza

Abstract

A combination of two methodologies is presented for detection and mapping of gypsum using ASTER L3A imagery. One of the methodologies uses the Quartz index defined for the ASTER TIR subsystem, which can be used for gypsum detection given its low response in Qi. The other consists in the combination of two band ratios of the ASTER SWIR subsystem, (4/5)/(7/5), which allows the identification of gypsum highlighting its high response in 4/5 and low response in 7/5. Two areas in the Cordillera Principal in the province of Mendoza were selected as case studies, and a field survey was conducted in order to evaluate the results. Both techniques are proved successful, yet classify erroneously some pixels as gypsum. Errors by excess are different for each method, which allows for these two techniques to be combined using a "decision tree" classifier to solve the misclassifications.