Improved satellite ground resolution for sea-ice observation using an inversion method and a priori information

Abstract

The proposals made for improving ground resolution in satellite microwave radiometry primarily seek to sharpen the main beam and refine hardware. However, deformation and changes in electromagnetic characteristics with scanning still significantly affect the antenna patterns of large-aperture antennas. This paper proposes an advanced method with two improvements. The first improvement is related to the method used to obtain a point spreading function (PSF) filter at every spot of interest in the scanning footprint. The method estimates only a typical PSF of the footprint just below the satellite orbit and applies rotation by the angular change that occurs as the footprint moves and rotates with scanning. The second improvement is to introduce a projection onto convex sets (POCS) method into the iterative calculation to solve the inverse problem and to find the most likely estimate of ground information. Two pieces of a priori information are also introduced in the iteration to achieve rapid convergence and reasonable estimation: The brightness temperature of the ground surface is 1) smooth and 2) nonnegative. Correlations between estimated results are also taken between neighboring footprints after retrieval, and convergence is easily attained in the iteration. As no practical satellite data were available, an observational simulation was used as a case study at the lowest frequency (6.9 GHz), which has the poorest resolution. In this case study, ice margins were clearly retrieved by the algorithm when ice floes were 50 km or more apart. Irregularities like ringing, unreasonable undulation, or unreasonable values have occasionally appeared with the current method, but are absent with the proposed improved inversion method used in combination with a POCS method. © 2006 IEEE.