Improving semi-automated glacial mapping with a multi-method approach: areal changes in Central Asia

Abstract

Abstract. Central Asia has been strongly impacted by climate change, and will continue to be impacted by diverse climate stressors in the coming decades. This study aims to decipher the impact of climate change on glaciers in the central Tien Shan Mountain Range, a large and understudied region located northeast from the Pamir Knot. To address glacier characteristics over a wide swath of Central Asia, the authors designed and implemented a glacial mapping algorithm which delineates both clean glacial ice – methods which are well documented – and glacial debris tongues, which often require extensive manual digitization. This research improves upon methods developed to automatically delineate glacial areas using spectral, topographic, velocity, and spatial relationships. The authors found that the algorithm misclassifies between 2 and 10% of glacial areas, as compared to a ~750 glacier control dataset. After validating the algorithm against multiple manually digitized control datasets, the authors applied it to a study area encompassing eight Landsat scene footprints stretching from the central Pamir through the central Tien Shan. A statistically significant, though minor, gradient in glacier area loss was found, where glaciers in the west of the study area have shrunk less than those glaciers in the east. This gradient is explained by differences in regional climate, where extratropical cyclones propagating from the west weaken and disband under continued topographic influence, as well as differences in topography, where high-elevation glaciers are thermally insulated from some of the impacts of changing temperatures in the region.