Synthetic Aperture Radar (SAR) is a high-resolution remote sensing imagery technique where a satellite sends microwave radiation down to earth and then receives the backscatter on the same antenna. The antenna measures amplitude and phase of the backscatter, and this can be used to create satellite imagery that cuts through weather events and illumination levels. In the Arctic ocean, scientists use SAR to manually chart sea ice. The conditions of the Arctic ocean necessitates SAR imagery use for environmental monitoring. Scientists have been searching for a strictly automatic segmentation of the sea ice, and this paper provides an overview of the systems that multiple national organizations have produced to solve this task.
Key Takeaways
The paper first takes us through the development of SAR. In 1978 the USA launched a satellite named SEASAT, which produced daily images of the sea ice in the Beaufort Sea using SAR. Kosmos-1870 was launched in 1987 and Almaz-1 was launched in 1991, and these two satellites demonstrated the capabilities of SAR for a variety of environmental monitoring tasks. The paper continues to describe the satellites that incrementally progressed SAR forward, ERS-1,2 , RADARSAT-1, Envisat, RADARSAT-2, TerraSAR-x, TanDEM-X, ALOS-2, Sentinel-1A/B.
SAR imagery uses Geographic Information System (GIS) mapping technology to be analyzed, and many of the national sea ice organization developed applications which suited their needs.
The paper goes into the several systems that have been developed for classification. Characteristics of SAR imagery, like tone or brightness, are commonly used to classify ice across a handful of labels such as multiyear ice (MYI), rough first year ice (FYI), smooth first year ice (FYI) and open water (OW). These labels seem to be generally standardized across techniques, with the addition or exclusion of certain categories. The paper discusses classification systems and their processes from the University of Kansas, the Alaska Satellite Facility Synthetic Aperture Radar Data Center at the University of Alaska, the University of Colorado, the Finnish Meteorological Institute (FMI), the University of Waterloo, and the Norwegian Meteorological Institute.
The paper discusses the technical aspects of SAR and the methods for ice classification. Some methods for ice classification fall under the category of supervised learning algorithms. Algorithms such as neural networks, support vector machines, bayesian classification, and linear discriminant analysis have all been applied to SAR data processing. Improvements in the technical and methodological aspects of SAR data processing will make significant improvements to the task of sea ice classification.
Citation
Zakhvatkina N, Smirnov V, Bychkova I. Satellite SAR Data-based Sea Ice Classification: An Overview. Geosciences. 2019; 9(4):152. https://doi.org/10.3390/geosciences9040152