Assessment of factors leading to Glacial Lake outburst floods using ensemble learning: a case study of Shishper Glacier, Himalayas

The examination of spatio-temporal variations in the geography and climate of a region is crucial for understanding the underlying causes of the possible occurrences of Glacial Lake Outburst Floods (GLOFs) and enhancing the ability to predict these catastrophic events. This study presents a comprehensive methodology for assessing GLOF-related factors by integrating topographic and climatic variables with advanced statistical and machine-learning techniques. The analysis focuses on spatial and temporal variations of the considered topographic and climatic factors, supported by statistical validations that reinforce the robustness of the findings. Specifically, the research investigates the Shishper glacier, identifying elevation and surface air temperature (SAT) as the most significant contributors to GLOF occurrence, based on vulnerability analysis. The role of black carbon in accelerating snow cover degradation is elucidated through spatial semi-variogram modelling. Additionally, SAT is projected 3–4 days in advance using Gradient Boosting, achieving a validation accuracy exceeding 97%. The study also explores changes in snow cover and lake area through the application of the Mann–Kendall trend test and Sen’s Slope estimator, with results at a 95% confidence level or higher. This integrative approach not only quantifies the interrelated dynamics between snow cover and glacial lakes but also offers a robust framework for the prediction of future GLOF events–filling a critical gap in the existing literature.