Author(s): Jorg Haarpaintner
Linked Author(s):
Keywords: No Keywords
Abstract: The Storfjorden sea ice cover has been observed during two winters, 1997 to 1999, using ERS-2 synthetic aperture radar (SAR) imagery. Under northerly winds this region presents high sea ice dynamics resulting in an ice cover composed of fast ice, pack ice and polynyas. Using the ice tracking method of nested correlation between sequential images, a high-resolution displacement vector field of the sea ice cover is obtained. Every node of a grid superposed on the initial SAR image A has its correspondent in a node of a deformed grid of image B obtained by maximum correlation. Hence, a correlation coefficient c can be attributed to each grid cell. An absolute velocity field, shear zones and a divergence field can be extracted as enhanced images from the displacement vector field. Together with the original SAR image and an image corresponding to the correlation coefficients for each grid node, a supervised sea ice classification into fast ice, pack ice and polynya is performed and gives first promising results that are compared to a manual segmentation. Guidance for future algorithm improvement is discussed and a future combination of this method with other algorithms is suggested.
Year: 2002