The identification of different ice facies on the north patagonian icefield (Chile) from ETM+ and ASTER images using a rule-based classification

Abstract

The identification of land cover in glaciated areas from satellite imagery has many uses including mapping changes in extent, and monitoring glacial hazards, but interpretation of that land cover is subject to challenges including deep shadows, lakes of varying turbidity, and debris-covered glaciers. Here, a rule-based classification in Definians eCognition is used to identify different ice facies from ASTER and Landsat ETM+ imagery of the North Patagonian Icefield in Chile using visible, near infrared, short wave infrared, and thermal infrared bands. The North Patagonian Icefield is a complex area to classify because it contains a variety of features including snow covered areas, bare ice, heavily crevassed areas, and debris-covered ice. The rule-based classification was successfully applied to both ASTER and Landsat ETM+ imagery. A rule-based classification is shown to be an accurate way to identify ice extent on both ASTER and Landsat ETM+.A greater variety of different ice facies are identified from ASTER than from Landsat ETM+.This is attributed to the higher ground spatial resolution of the ASTER visible and near infrared bands resulting in fewer mixed segments in the first stage of the classification process. The map showing the Landsat ETM+ classification is at a scale of 1:1,000,000 while the map showing the ASTER classification is at a scale of 1:700,000.