Glacier Movement Prediction through Computer Vision and Satellite Imagery

Over the last decade, climate change has impacted Earths' atmosphere and environment more than anytime before. Glaciers are the most sensitive indicators of its impact. In this work, we model a glacier's evolution by applying computer vision algorithms on high-resolution satellite imagery. We detect changes in the ice coverage movement by applying a dense optical flow algorithm over an image time series covering a particular scene (region) and processed to extract the NDSI. We perform tests on the Jungfrau-Aletsch-Bietschhorn (JAB) glacier in the Swiss Alps. Our results show that we are able to obtain relevant information by computing motion vectors across time. Furthermore, we observe small differences between our predicted NDSI and the observed values demonstrating the efficiency of the approach.