Improvement of Sea Ice Drift Extraction Based on Feature Tracking from C-SAR/01 Imagery

Abstract

In this study, the extraction of sea ice drift from imagery captured by the 1-meter C-SAR 01 satellite (C-SAR/01) was facilitated utilizing the oriented fast and rotated brief algorithm within the feature tracking procedure, thus addressing the previously unexplored area of sea ice drift extraction using C-SAR/01 imagery. The retained keypoints and nearest neighbor distance ratio test for sea ice drift extracted from C-SAR/01 imagery were compared, indicating high reliability with 300 000 and 0.75, respectively. In addition, the local outlier factor algorithm is proposed in this article, which can effectively remove erroneous sea ice drift vectors. The sea ice drift extracted from C-SAR/01 was validated against manually extracted sea ice drift, revealing an uncertainty of 0.271 cm/s in speed and 8.331° in direction. Furthermore, the sea ice drift obtained from the algorithm in this study, when compared with sea ice drift from IABP buoys, exhibits high accuracy, reflecting the robustness of the algorithm.

Keywords

• 1-meter C-SAR 01(C-SAR/01)
• Arctic
• feature tracking (FT)
• oriented fast and rotated brief (ORB)
• sea ice drift
• synthetic aperture radar (SAR)