Balanced inversion of simulated bistatic radar cross-section data for remote sensing of snow-covered sea ice

Abstract

The reconstruction of the dielectric profile of snow-covered sea ice from simulated bistatic radar cross-section data is investigated. To this end, a cost function representing the discrepancy between the true radar cross-section data and the calculated radar cross-section data due to a predicted dielectric profile is formed. This cost function is iteratively minimized using the differential evolution algorithm so as to reconstruct the unknown dielectric profile. Herein, four different formulations for this cost function are considered. These formulations are distinguished by their utilized weighting factors for each data point within the bistatic radar cross-section data set. Although these cost functions use the same simulated bistatic normalized radar cross-section data set, it will be shown that the reconstruction accuracy achievable from each of them will be different. Finally, an appropriate weighting scheme for bistatic data sets, which emphasizes the scattering data points collected at the vicinity of the specular angle, is presented. It will be shown that the use of this weighting scheme results in enhanced reconstruction accuracy as compared to the other formulations. It is speculated that the proposed weighting scheme balances the individual data points more properly, thus, resulting in enhanced accuracy.