Automated Terminus Detection for Greenland's Peripheral Marine-Terminating Glaciers

Abstract

Changes in the length of marine-terminating glaciers strongly influence the mass balance of glaciers, ice caps, and ice sheets. Currently, quantification of glacier length change through measurement of terminus position relies on time-consuming and subjective manual mapping techniques, limiting our ability to understand the dynamics controlling glacier terminus changes. I developed an automated method of mapping glacier terminus positions in satellite imagery using observations from a representative sample of Greenlands peripheral glaciers. The method is adapted from the 2D Wavelet Transform Modulus Maxima (WTMM) segmentation method, which has been used previously for image segmentation in biomedical and other applied science fields. The gradient-based method places edge detection lines along regions with the greatest gradient in intensity in the image, such as the contrast between glacier ice and water or glacier ice and sea ice. I quantified the accuracy of the automated method with reference to a validation dataset consisting of over 500 manual delineations and determined that the automated method is capable of mapping glacier termini over a wide range of image conditions (light to intermediate cloud cover, uniformly dim or bright lighting, etc.) within 1-pixel uncertainty. These time series generated automatically from Landsat images (which have a frequent repeat interval and a long record of images) are capable of resolving sub-seasonal to multiannual temporal patterns as well as regional patterns in terminus change for these glaciers. The terminus position time series generated from this automated method indicate that the marine-terminating peripheral glaciers in southeast Greenland undergo synchronous terminus retreat in 2016-17. Initial exploration of regional atmospheric and ocean conditions links this synchronous retreat to subsurface ocean warming and increased surface runoff.