Sea-ice motion derived from two (partially) overlapping ESA Sentinel1 [S1]A or B scenes. Satellites S1A/B carry C-band (5.405 GHz) Synthetic Aperture Radar [SAR] sensors. For this data set images from the Extra Wide swath (EW) mode of operation (swath width 410 km) have been used. EW mode data are available as a medium-resolution ground range detected (GRD) product, i.e., resolution of 93 × 87 m and pixel size 40 × 40 m. Approximately two-thirds of the EW mode data recorded over the Antarctic area are dual-polarisation (HH + HV) products. The remainder are mainly single-polarisation (HH) products. For further detail, see ESA's Copernicus web portal.
Ice motion is derived from suitable SAR image pairs with sufficient spatial overlap but relatively short time separation, i.e. ideally 6 days or less. Image-crosscorrelation analysis is employed to identify displacement vectors within the image pair.
A NetCDF-format database of high-resolution(:SAR) Antarctic sea-ice motion vectors.
With regards to the geolocation accuracy of Sentinel-1 EW high-resolution ground range detected (GRD) product, it has been found that the 3σ accuracy specification is be approximately ±14.4 m (or ±0.36 pixels). This correlation uncertainty is sub-pixel, in agreement with that of Landsat data with an uncertainty of about 0.8 pixel.
The accuracy of the image crosscorrelation has a standard deviation of σ = 0.4 pixels (or 16 m). With the geocoding accuracy of 4.8 m (about 0.1 pixels), then this suggests the image-correlation uncertainty is the dominant error component at around ±0.3 pixels.
These data will be presented in a manuscript currently in preparation:
Improved Antarctic sea-ice motion fields from Sentinel-1 SAR imagery
Glenn Hyland, Petra Heil, and others
In preparation, 2020