Machine learning for cryospheric mass movements: challenges and pathways

Cryospheric mass movements (CMMs) are accelerating under rapid Arctic and high-mountain warming, yet understanding and predicting them remains challenging due to sparse observations, complex multi-scale processes, and shifting environmental baselines. This Perspective argues that machine learning should advance CMM science not only through prediction, but through transferable and physics-aware inference. We outline a roadmap centered on benchmark datasets, cross-scale generalization, and transdisciplinary collaboration for scientifically grounded hazard assessment.