Ice cover discrimination in the Greenland waters using first-order texture parameters of ERS SAR images
Abstract. The first order texture parameters: power-to-mean ratio (PMR), skewness and kurtosis were computed for four surface types of an ice covered ocean, namely ( a ) open water, ( b ) ice areas of concentrations between 40 and 90%, ( c ) fast ice, and ( d ) thin ice or ice of low concentration (<20%) from the ERS SAR precision images of the waters around Greenland. The different regions were identified by making near-simultaneous observations using aerial reconnaissance. It was found that using the PMR values it is possible to discriminate between regions of low concentration of ice or thin ice, on one hand, and open water, fast ice and ice of concentration between 40 and 90% on the other. Discrimination between open water and ice of concentration between 40 and 90% is, in many cases, also possible. No discrimination is possible between open water and fast ice. The higher order parameters: skewness and kurtosis, gave similar results (but no improvement). These parameters were also compared with those given by the most commonly used theoretical probability distribution functions (pdf's) such as the log-normal, Rayleigh, gamma and the k -distribution. The comparison showed that the values for open water, fast ice and ice regions were in good agreement with those given by the k-pdf with the values for water approaching those of the gamma pdf (limiting case of k -pdf). Regions of thin ice or ice of low concentration did not fit any of the above distributions. The shape parameter of the gamma function used to model the textural variations in the k -pdf model was also used to discriminate between the different regions. Compared to the PMR parameter it also contained no additional information.
No Reference information available - sign in for access.
No Citation information available - sign in for access.
No Supplementary Data.
No Article Media