Reconstruction of Ice Sheet Temperature Maps using a Sparsity-based Image Deconvolution Method

This paper explores the application of modern image processing techniques in retrieving high-resolution passive microwave images of the polar ice regions on Earth from sparsely sampled interferometric array measurements. Such observations, sensitive to ice sheet temperature, would be valuable benchmark measurements for ice process models. In this paper, we propose to use a total variation-based method that addresses the challenges associated with large sidelobes and blurry maps resulting from long baseline interferometry. We present a robust algorithm that employs total variation (TV) minimization and the split Bregman optimization. This technique effectively deconvolves images, preserves edges, and minimizes noise amplification without introducing artifacts. To evaluate the algorithm's performance, we performed tests on a simulated image and a real satellite image of Antarctica. Additionally, we assessed the algorithm's performance using different interferometric array configurations, including both dense and sparse arrays with varying numbers of elements.