Extracting Outcrop Extents in Antarctica: Current problems, new solutions and a complete dataset for the Antarctic continent

Abstract:

Differentiating rock outcrop from ice is a particular problem in Antarctica where extensive cloud cover and widespread shaded regions produced by the combination of steep, mountainous topography and high solar elevation angles lead to classification errors. The existing outcrop dataset has significant georeferencing issues including overestimation and generalisation of areas of rock exposure. The existing method for automated rock and snow differentiation, the Normalised Difference Snow Index (NDSI) has trouble delineating rock and ice due to misclassification in the shaded regions of images and cannot delineate illuminated rock from clouds. This study presents a new method for identifying rock exposures from Landsat 8 data that excludes areas of snow (both illuminated and shaded), clouds and liquid water, achieving much higher accuracies than existing data and methods. The new methodology has been applied to the entire Landsat 8 coverage of the Antarctic continent (North of 82°40’) to produce a new rock outcrop dataset for Antarctica.