SAR Algorithm for Sea Ice Concentration-Evaluation for the Baltic Sea

• Published in: IEEE geoscience and remote sensing letters Vol. 9; no. 5; pp. 938 - 942
• Main Authors: Berg, A., Eriksson, L. E. B.
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.09.2012; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Accuracy; Algorithms; Autocorrelation; C-band; Charts; Classification; Correlation; ice concentration; Marine; neural network; Neural networks; Pixels; SAR; Sea ice; Sea surface; Statistics; Synthetic aperture radar; synthetic aperture radar (SAR); Training; ANE, Baltic Sea
• ISSN: 1545-598X; 1558-0571; 1558-0571
• DOI: 10.1109/LGRS.2012.2186280

A new sea ice concentration algorithm has been developed for C-band synthetic aperture radar data. Detailed autocorrelation statistics are derived and adopted by the algorithm, and a neural network is utilized for training against 41 sea ice charts. The charts are produced by ice analysts at the Swedish Ice Service and cover the Baltic Sea. The classification of open water pixels is accurate to 94% on average, and the classification of sea ice pixels has an accuracy of 87%. This results in a root-mean-square error of 6.7 percentage points in estimating the sea ice concentration.