Rock Glacier Mapping: A Method Utilizing Enhanced TM Data and GIS Modeling Techniques

Abstract

Rock glaciers, an important component to alpine systems, have been identified as an indicator of permafrost distribution as well as a resource for humans, serving as backdrops for residential areas, dam abutments, and a water source for urban areas. As a result, mapping is imperative; however, a simple method for rock glacier detection has not been developed. Traditionally, rock glaciers have been mapped using field techniques or aerial photography. Spaceborne Landsat MSS data has proven to be ineffective for identifying rock glaciers. The purpose of this research is to establish a modeling procedure to identify rock glaciers of the Blanca Massif of Colorado utilizing Landsat TM data, image enhancements, and ancillary DEM data. I used a topographic normalizing function to reduce shadows, band ratios and principal components to emphasize rock glaciers, edge enhancements to detect all distinct rock glacier borders, and a 1:24,000 DEM to restrict the study area to criteria common to all rock glaciers: elevations greater than 11,000 feet and a slope between 0° and 21°. Each of the enhancement techniques was tested individually and in combination with each other. The combination image, which utilized all of the enhancements, provided the greatest classification accuracy. The overall classification accuracy of rock glaciers in the Blanca Massif was 81.0%. Tests revealed that rock glaciers greater than 3.65 hectares can be mapped. These findings suggest that Landsat TM data, when used in conjunction with a combination of ratios, principal components, edge enhancements, and DEM data, can be used to accurately map most rock glaciers.