Evaluation of a sea ice algorithm for SAR data from the Bay of Bothnia

Övrigt konferensbidrag, 2010

A sea ice drift algorithm published by M. Thomas et. al. (2008) has been implemented and evaluated. Input to the algorithm is Synthetic Aperture Radar (SAR) images, which are processed using phase correlation in a multi-resolution processing system. The algorithm has been tested with horizontally co-polarized (HH) ENVISAT ASAR Wide Swath images and RADARSAT-2 ScanSAR images. The possible benefits of using cross-polarized (HV) RADARSAT-2 ScanSAR data are investigated, and initial testing of the algorithm for L-band SAR data from ALOS PALSAR has been done. The validity of the produced motion fields has been tested in three different ways. Most of the SAR images in this analysis were acquired over the Bay of Bothnia. Five meteorological stations located in this region have been collecting wind data with a temporal resolution of three hours. It is confirmed that the wind data correlates with the derived sea ice motion. Another source of validation data that has been used is the daily ice charts published by the Swedish Meteorological and Hydrological Institute (SMHI). The third method used for validation is straightforward, visual tracking of sea ice features. The algorithm is facing some difficulties when it comes to ice tracking close to the shoreline, in archipelagoes etc, since the motionless solution will be favoured. It can however be suppressed by filtering areas of land. This addition gives a more robust algorithm.