Object-based analysis of unmanned aerial vehicle imagery to map and characterise surface features on a debris-covered glacier

Abstract

Debris-covered glaciers in the Himalaya may have spatially-averaged rates of surface height change that are similar to those observed on bare-ice glaciers, despite the insulating effects of thick debris. Spatially heterogeneous melt patterns caused by the development and evolution of ice cliffs and supraglacial pond systems result in substantial mass losses over time. However, mechanisms controlling the formation and survival of cliffs and ponds remain largely unknown. To study the distribution and characteristics of these surface features we deploy an unmanned aerial vehicle (UAV) over a stretch of the debris-covered Langtang Glacier, Nepal. Acquired images are processed into high-resolution orthomosaics and elevation models with the Structure from Motion (SfM) photogrammetry algorithm. Ice cliffs and ponds are classified using object-based image analysis (OBIA) and their morphology and spatial distribution are analysed and evaluated using object, pixel and point cloud approaches. Results show that ice cliffs are predominantly north-facing, and larger ice cliffs are generally coupled with supraglacial ponds, which may affect their evolution considerably. The spatial distribution of ice cliffs indicates that they are more likely to form in areas where high strain rates are expected. The spatial configuration of ponds over the entire tongue reveals high pond density near confluences, possibly due to closure of conduits via transverse compression. We conclude that the combination of OBIA and UAV imagery is a valuable tool in the semi-automatic and objective analysis of surface features on debris-covered glaciers. The technique may also have potential for upscaling to the use of spaceborne imagery, and the use of UAV-derived point clouds to analyse ice cliff undercuts is promising.