Modeling Cascading Hazards in High Mountain Environments: Challenges and Approaches from the Thame Case Study, Nepal

High mountain environments often experience hazards that do not occur in isolation but as interconnected processes. A typical setting may involve a steep rock face, sometimes topped by glacier ice, where failures can trigger rock-ice or mixed avalanches depending on seasonal conditions. When such events occur above a glacial lake, as is common in many regions, the impact can initiate secondary processes such as glacial lake outburst floods, with significant downstream consequences.

Numerical models are valuable tools for estimating the runout of individual processes; however, simulating entire hazard cascades involving multiple material types remains challenging—particularly for forward modeling. In this study, we explore methods for modeling cascading processes, either through integrated physical models or suites of specialized models, and assess which approaches are most suitable at different spatial scales (local, basin, regional, national).

At least two GLOFs in the recent five years in Nepal were caused by a cascade of a mass flow impacting the lake and causing dam failure or overtopping, followed by a downstream flood with significant impacts. Permafrost thaw induced slope instability as well as excessive snow melt in source areas contributed to the initial release and a variety of subsequent erosional processes further downstream exacerbated impacts. Previous modelling has been largely focused on the flood from the lake exit, not considering the multiple aspects contributing to the complexity of the cascade.

Our analysis focuses on the Thame area in the Everest region of Nepal, where a rock-ice avalanche impacted Thyanbo Lake in August 2024, triggering a glacial lake outburst flood that caused severe damage downstream. This is done in light of producing risk maps for the wider Dudh Kosi Basin, where a number of upstream processes can potentially exacerbate impacts for communities much lower than the periglacial terrain. We discuss the advantages and limitations of various modeling strategies, the challenges of representing full process chains, and potential ways to combine approaches to improve physical realism and predictive capability.