SAR-based estimation of the baltic sea ice motion

The ice season in the northern part of the Baltic Sea varies from a few weeks up to 4-5 months depending on the location and the winter. As a long-term average the Baltic Sea is ice covered for about 45% of its surface area at the annual maximum. Due to the relatively narrow basins, the deformation rate remains rather high throughout the winter. The ice thickness measurements show that in the Bay of Bothnia over one third of the ice cover is thicker than 1 m in the later stages of the ice season during a typical winter. Modeling the state and development of the Baltic ice cover has been an active research area which have resulted in operative ice models. The current operative ice model run at the Finnish Ice Service (FIS) models the following parameters: ice motion and concentration, mean ice thickness, ridged ice thickness, ridged ice concentration, compressive region, and deformed ice fraction. Some case studies have been made to validate the results. However, a more systematic forecast evaluation is needed, the final goal being the data assimilation to integrate the EO data into the numerical ice model. In this paper we present an algorithm which estimates the ice motion field from two successive RADARSAT-1 SAR images. A large number of RADARSAT-1 ScanSAR Wide mode SAR images, usually over 100, is received at FIS every winter. The same sea areas are visible in the images with a time interval of 1-3 days.