| Summary: | Temporally and spatially comprehensive data products of calving front variation are essential for a better understanding and modelling of tidewater glaciers. However, most current calving front records are limited in temporal resolution as they rely on manual delineation which is a laborious and time-consuming, hence ineffective process. In this contribution, we address this issue by applying an automated method to delineate calving front positions from optical satellite imagery. The technique is based on recent developments in deep learning while taking full advantage of multi-spectral input data. After evaluating the method utilizing three independent test datasets, we apply it to Landsat imagery generating 9243 calving front positions across 23 Greenland outlet glaciers from 2013 to 2021. Resulting time series are analysed using a rectilinear box method. In this way we are able to resolve not only long-term and seasonal terminus changes but also sub-seasonal fluctuations. This allows us to classify different calving patterns and accurately identify pattern changes within our time series. We discuss different glaciological applications of our results, in particular their implications for associated glacier modelling efforts.Our method and inferred results form a significant advancement towards establishing intelligent processing strategies for glacier monitoring tasks. We create new opportunities to study and model the dynamics of tidewater glaciers. These include the advance towards constructing a digital twin of the Greenland ice sheet, which will enhance our understanding of its evolution and its role within the broader Earth’s climate system.
|
|---|