Unsupervised glacier terminus detection in satellite imagery using the 2D Wavelet Transform Modulus Maxima (WTMM) segmentation method

We adapt a program used for delineating mouse chromosome territories in fluorescence microscopy images using the automated multi-scale 2D Wavelet Transform Modulus Maxima (WTMM) segmentation method to map glacier termini in satellite images. The goal of the project is to develop an unsupervised approach to efficiently quantify glacier terminus changes from satellite images so that we can investigate the spatial and temporal patterns in terminus

change and explore the environmental factors driving them. Here we describe the adapted method, which we have applied to a subset of glaciers spanning Greenland’s periphery for method development and error quantification. Although we focus here on only a small subset of the 641 marine-terminating glaciers around Greenland’s periphery and panchromatic imagery from the Landsat 8 satellite, the unsupervised method will be used to construct terminus position time series for all of Greenland’s peripheral marine-terminating glaciers using the full ~40-year Landsat satellite imagery archive.