Ionosphere effect correction in InSAR using improved split spectrum processing

Abstract:

During last decade, many researchers have indicated the presence of ionospheric artifacts in SAR and InSAR data, especially in low frequency data. Different methods, such as Faraday rotation-based, azimuth shift-based (pixel offset and Multiple aperture InSAR) and split-spectrum InSAR-based techniques, have been developed to mitigate these ionosphere effect. While the efficacy of these methods was shown in test cases, all existing methods have limitations that reduce their performance in operational applications. Split-spectrum InSAR approaches, which split the image’s range spectrum into sub-bands and measure phase difference between these sub-bands that are introduced by the dispersive ionosphere, are the most promising candidates for operational implementation. While tests indicated their global applicability, they still suffer from low signal-to-noise ratio and from sensitivities to processing errors related to phase unwrapping and low coherence. Therefore, more research is needed to make this approach more reliable and stable.

To achieve this purpose, a cascading multiple sub-band InSAR processing combined with quasi-coherent scatter (QCS) approach is put forward. A cascading multiple sub-bands processing is first implemented in order to avoid the largest error source from phase unwrapping. To further increase the signal-to-noise ration for ionospheric phase screen estimation, a quasi-coherent scatter method is implemented. A quasi-coherent scatter was firstly selected based on coherence analysis. Multiple sub-bands processing for ionosphere correction will then only be based on the selected coherent scatters. After this, a geostatistical interpolation method is used to reconstruct the ionospheric phase screen for the whole image.

To test the effectiveness of this method, both real data contaminated by ionosphere and ionosphere free data with simulated ionosphere phase screens from WBMOD will be used. For real data, a cross validation with results from Faraday roation is conducted.