Multilook SAR processing and array optimization applied to radio echo ice sounding data

Several outlet glaciers in Greenland are retreating and speeding up, and Greenland ice sheet's contribution to sea level is increasing over the last decade. Information on ice thickness and bed conditions is needed in order to understand why these outlet glaciers are retreating and speeding up, and to develop both diagnostic and prognostic models. These outlet glaciers contain very lossy ice and have a very-rough crevassed surface. Airborne sounding of these fast-moving outlet glaciers is very difficult because extremely weak bed echoes are often buried in off-nadir clutter - off-nadir scattered signals from the rough surface. We processed data from each element of the antenna-array with a multilook time-domain back-projection (MLTDBP) synthetic aperture radar (SAR) processor and used these data in an adaptive array processing algorithm we developed to extract weak bed echoes buried in clutter and noise. The algorithm is based on the gradient descent method. It estimates and minimizes phase and amplitude errors introduced by aircraft roll, the radar system, and aircraft wing flexure. We will show a few examples of recovered bed echoes from data collected over glaciers near Nuuk, Greenland during the 2009 field season to demonstrate the performance of the MLTDBP processor and array optimization algorithm. We also included an example from data collected on the interior ice during 2006.