Sentinel-2 image courtesy of
Sentinel-1 image mosaic - www.seaice.dk
Manual ice charting from multi-sensor satellite data analysis has for many years been the method at National Ice Centers around the world for producing sea ice information for maritime safety. Also at the DMI Ice Service ice charts are today made by ice analysts that use all relevant imagery available; primarily Copernicus Sentinel-1 radar imagery (SAR), due to their high resolution and capability to see through clouds and in polar darkness.
Manual ice charting is a time-consuming method that is increasingly challenged by the vast amount of available and free satellite imagery these years. Along with a growing user group accessing wider parts of the Arctic due to the retreat and thinning of the Arctic sea ice, this calls for a more effective way of producing detailed and timely ice information to the users. Ultimately, users want reliable ice forecasts for safe navigation and planning. To be able to provide these, there is a need for future automated, detailed and standardized ice observations from satellite data for integration in forecast models.
The ASIP vision for designing an automatic and robust sea ice classification scheme is to merge the Sentinel-1 imagery with other satellite sensor data that have complementary capabilities, such as passive microwave data from AMSR2, to better resolve the ambiguities that can occur in SAR imagery eg. at high wind speeds. ASIP use Convolutional Neural Networks (CNN) that are trained with a vast data set of ice charts, to merge the different satellite sensor data and generate ice maps.
It is a project goal to test an assimilation of the future automated ice maps in the DMI sea ice model and to do demonstration services to users.