A Machine Learning Framework to Automate the Classification of Surge‐Type Glaciers in Svalbard

Abstract Surge‐type glaciers are present in many cold environments the world. These experience a dramatic increase velocity over short time periods, surge, followed by an extended period of slow movement, quiescence. This study aims at understanding why only few exhibit transient behavior. We develop machine learning framework to classify surge‐type glaciers, based on their location, exposure, geometry, climatic mass balance and runoff. apply this approach Svalbard archipelago, region with relatively homogeneous climate. compare performance logistic regression, random forest, extreme gradient boosting (XGBoost) models that we newly combined database Svalbard. Based most accurate model, XGBoost, compute surge probabilities along glacier centerlines quantify relative importance several controlling features. Results show surface bed slopes, ice thickness, width, balance, runoff significant features explaining probability for A thicker wider low slope has higher be classified as surge‐type, which is good agreement existing theories surging. Finally, build map The shows robustness classifying were not previously such inventories but have been observed Our methodology could other areas