An FFT-Based CLEAN Deconvolution Method for Interferometric Microwave Radiometers With Spatially Variable Beam Pattern

A modification to the extended CLEAN deconvolution method for passive microwave interferometric radiometers is presented to account for the interferometer beam patterns that vary depending on the source position within the interferometer field of view. This method is developed for the newly proposed multisatellite rotating microwave interferometric radiometer concept for geostationary atmospheric sounding, which is found to have a spatially varying beam pattern. The proposed method is implemented by the piecewise application of Fast Fourier Transform (FFT)-based convolution, i.e., the image is divided into discrete segments, where the beam pattern is assumed to be constant within each segment. This approach allows the proposed method to exploit the speed advantage of the FFT. Applying this method to the proposed multisatellite interferometer shows a twofold improvement in the radiometric accuracy of the restored brightness temperature map, as compared with the constant beam pattern CLEAN.