In recent years, large number of Interferometric SAR applications in earthquake, cryosphere and tectonic geodesy show the great needs for ionospheric effect correction [1-4]. Several approaches, which mainly include the split spectrum InSAR technique, the Faraday rotation based method and the azimuth shift based method (the Multiple Aperture Interferometric and conventional azimuth shift) have been proposed to meet this end [5-11]. These techniques exploit different aspects - phase delay, polarimetric property and geometric distortion of the ionospheric effect on SAR data to retrieve the differential ionosphere in InSAR. These techniques work well in certain conditions, however, their performance might be limited in cases due to noise effect and the weakness of each technique itself. In this study, we are proposing a combined estimator by joining the Faraday rotation based method and the split spectrum technique of InSAR ionospheric effect correction, which is supposed to achieve a more robust and higher accuracy performance comparing to individual technique itself.

Summary form only given. Monitoring ice motion is of great importance for determining ice mass balance and its contribution to sea level rise. InSAR has become a reliable and important tool for monitoring the ice dynamic due to its known allweather and all-day capabilities. Since 1990s, with the emergence of various sensors of different wavelengths, such as the C-band ERS1/2, Envisat ASAR, Radarsat-1/2, X band TerraSAR-X, Sentinel-1 and the recent L-band ALOS 1&2 PALSAR SAR data, InSAR has become a major technique for monitoring large scale ice motion with unprecedented resolution. Recently the first comprehensive ice motion of the Greenland and the Antarctica have been generated with this technique. Using InSAR to detect the ice motion, the ionospheric phase delay has always been a problem and this is still not properly resolved. This is especially true for low-frequency SAR data. Recent publications of comprehensive ice motion mapping with L-band data shows ionospheric errors are about 17 m/yr near magnetic pole, which can be larger than the actual signal in some areas. Filter-based methods and empirical methods have typically been used to mitigate ionospheric effects, however, without a proper consideration of the physical origin of the ionospheric phase, these methods are prone to error and may introduce spurious biases. In this study, we are going to assess the benefits of ionospheric correction on ice motion mapping using our newly developed split spectrum InSAR-based ionospheric correction. The split spectrum technique forms two or more interferograms from non-overlapping range frequency sub-bands, and estimates the differential ionospheric phase signal by exploiting the dispersive nature of ionospheric delay. Robust co-registration techniques, automatic phase unwrapping error correction, and adaptive filter techniques were developed to enable ionospheric correction with high accuracy. In this paper, we will first present an outline of our split-spectrum InSAR-based ionospheric correction approach, including our advanced error correction and filtering algorithms. We will apply this algorithm to a large number of ionosphere-affected dataset over the large ice sheets (Antarctica and Greenland) to evaluate correction performance and to estimate the benefit of split-spectrumbased ionospheric correction for ice motion analysis. For validation purposes, we will compare the corrected interferometric phase to stable feature like ice islands, emergent mountains and exposed rock out crops. We will also compare the corrected phase data to known ice velocity fields for the analyzed areas. These velocity fields are available to us through cooperation with the lead U.S. scientists for the respective ice sheets (Greenland: Ian Joughin; Antarctic: Eric Rignot). A preliminary result is displayed below. Panel (a) shows the ionosphere contaminated interferogram, and panel (b) is the estimated ionosphere phase which is wrapped, and Panel (c) is the ionosphere corrected interferogram. More analysis about the residue signal will be discussed in detailed in the full paper.