An automated method for mapping geomorphological expressions of former subglacial meltwater pathways (hummock corridors) from high resolution digital elevation data

Elongated tracts of hummocks or ‘hummock corridors’, exposed on palaeo-ice sheet beds, are believed to represent former subglacial meltwater pathways. Here, we present a method, coded in MATLAB, for automatically detecting and mapping hummock corridors from high-resolution digital elevation models (DEMs). Initially the DEM is filtered to remove bed roughness outside the size range of hummocks. A Fast Fourier Transform is then performed to determine the dominant orientation of hummock corridors and remove misaligned features. Finally, image segmentation is used to isolate and extract the hummock corridors as a binary mask. We tested this automated approach visually and statistically against detailed manual mapping in three areas of Canada and northern Scandinavia. Results show that while the automated method does not perfectly reproduce the manual mapping, it successfully captures the general configuration, morphometry (length, width) and location of hummock corridors, despite variation in expression across and between sites. This technique is ideally suited to take advantage of newly available high-resolution digital elevation data (e.g. the ArcticDEM), whose enormous volume makes large-scale manual mapping prohibitively time consuming. Its application will enable efficient and comprehensive mapping of the spatial distribution of hummock corridors across palaeo-beds that is necessary for deriving insights into their formation and the organisation of subglacial meltwater flow beneath ice sheets.