Antarctica is surrounded by ice shelves holding back the glacier flow to the ocean in many parts of the coastline. As soon as iceberg calving happens and buttressing parts of the ice shelf are lost, the ice shelf front retreats and the buttressing effect weakens. This results in increased glacier flow velocities enhancing the contribution of the Antarctic ice sheet to global sea level rise.
For the first time, the recently created IceLines dataset allows the continuous monitoring of Antarctic ice shelf front dynamics. This allows conclusions on iceberg calving mechanisms and to observe changes in the Antarctic coastline. On the basis of Sentinel-1 data, the fronts of the largest ice shelves of Antarctica are continuously mapped every 6 to 12 days (depending on the availability of Sentinel-1 data). A deep learning approach was envisaged to automatically extract ice shelf front positions by training a neural network (HED-Unet). Only with the help of artificial intelligence it was possible to develop a temporal and spatial transferable algorithm which can be applied on the entire Antarctic continent. The generated dataset includes front positions for 36 ice shelves with a daily, monthly, seasonal and annual temporal resolution since the launch of Sentinel-1 in 2014. IceLines is updated on a monthly basis and can be explored or downloaded from the EOC GeoService. Moreover, it is possible to integrate the dataset in existing applications and projects via the Web Map Services (WMS).
The IceLines dataset for four exemplary ice shelves.