Use of sequential SAR images for detecting ice and water in view of data assimilation

Abstract

In this study, we present a technique for automated detection of ice and open water based on ice motion information derived from sequential RADARSAT-2 images and an ice probability model applied to both SAR images. We investigate how the use of sequential synthetic aperture radar (SAR) images could increase the number of ice/water retrievals compared to the ice/water detection applied to a single SAR image only. The proposed technique was run for 736 image pairs and the ice/water retrieval results were verified against Canadian Ice Service Image Analysis products. Our results suggest that the fraction of water samples classified correctly has significantly increased from 65% (in the case of using a single SAR image) to 81% (in the case of using sequential SAR), while the detection accuracy stayed at approximately the same high level exceeding 99%. The developed approach is recommended to be implemented as part of the data assimilation component of the operational Environment and Climate Change Canada Regional Ice-Ocean Prediction System. The results are particularly important in light of the upcoming Canadian RADARSAT constellation mission which will significantly increase the amount and frequency of SAR observations over the Arctic region.