Ocean-induced glacier retreat drives mass loss in Svalbard 

Arctic Amplification is making Svalbard one of the most climatically sensitive regions in the world and it has been undergoing accelerated mass loss over the past several decades. A major uncertainty in predicting the future sea-level rise contribution from marine-terminating glaciers is ice dynamics, which can be driven by non-linear calving processes. However, the relationship between calving and ice dynamics is not well understood in Svalbard, in part due to the lack of high-resolution calving front observations. To improve our understanding of the glacier calving dynamics and its relation to dynamic mass loss, here we use a novel fully automated deep learning framework to produce a new calving front dataset of 149 marine-terminating glaciers in Svalbard. This dataset, which includes 124919 glacier calving front positions from 1985 to 2023, has high spatial and temporal resolutions and is derived from multiple optical and SAR satellite images. We then use this new calving front dataset to systematically quantify the calving front change variabilities at different temporal scales, and identify the key climate drivers controlling the calving dynamics. We show that ocean forcing plays a central role in controlling the glacier calving front changes and mass imbalance. Our study highlights the importance of including ice-ocean interaction in projecting future glacier mass loss from Svalbard.