Development of Operational Iceberg Detection and Forecasting for the Barents Sea
Norwegian Ice Service, NORWAY
The Norwegian Ice Service utilizes around 150 dual-polarization Radarsat-2 ScanSAR scenes per month, supplied through the GMES MyOcean project, as part of its sea ice charting for the European Arctic. With increased economic activity in the Barents Sea, both in fisheries and the oil and gas sectors, there is an urgent requirement by marine users for improved mapping of iceberg concentrations and drift patterns. Existing historical observation data is typically over 20 years old, has sporadic temporal and spatial coverage, and therefore cannot be used to validate detection techniques using modern satellite sensors.
This paper describes efforts by the Ice Service to utilize its archive of Radarsat-2 scenes in generating an up-to-date iceberg climatology for the Barents Sea. The standard Constant False Alarm Rate (CFAR) method is used to generate possible targets in each image. These results are then filtered by comparison between sets of image targets using the target shapes and sizes, multiple-polarization backscatter statistics, and model predictions of iceberg velocity. This allows reduction in false alarms due to detection of vessels, and allows tracking of distinct targets to generate drift tracks.
The analysis covers the period 2009 to 2012, when Radarsat-2 data has been used by the Ice Service, and the results applied to the ongoing acquisition of Radarsat-2 images allows iceberg target warnings to be supplied to marine users in the area. The product will be further developed with the regular coverage of Sentinel-1 data when that becomes available.