Multilayer Ionosphere Correction in Low-Frequency Interferometric SAR: Modeling, Inversion, and Application to BIOMASS

Early analysis of BIOMASS interferometric data acquired during the commissioning phase at Boreal and Antarctic latitudes revealed the presence of strong ionospheric disturbances that cause severe coherence losses. This phenomenon was investigated by studying the variations of the interferometric phase with respect to the squint angle. The observed phase variations turned out to be consistent with the assumption of a complex vertical structure of the ionospheric phase, characterized by fast azimuth variations occurring at km or sub-km scale at multiple layers between the ground and the satellite orbit. To the best of our knowledge, this phenomenon has never been considered in synthetic aperture radar (SAR) literature. This article is therefore intended to present and discuss the results of our investigation on suitable correction methods needed to guarantee the generation of accurate geophysical products from BIOMASS interferometric data. The starting point of our analysis is a multilayered model of the ionospheric phase formulated through multisquint interferometry (MS-InSAR), that is, by varying the radar line of sight (LoS) across different squint angles. The model is analyzed from a theoretical perspective to discuss the sensitivity of BIOMASS to ionospheric layering. Next, an inversion procedure is presented to estimate the ionospheric phase at an arbitrary number of layers directly from the complex coherence. The third element of our research consists of the formulation of correction methods to restore data quality given the estimated single- or multilayer ionospheric phase. All aspects of the research are consistently supported by analysis of BIOMASS interferometric data acquired at different latitudes and affected by mild to very strong ionospheric disturbances. The capability of the developed methods to correct BIOMASS interferometric data was quantitatively evaluated by comparison against interferograms generated by data-based coregistration. In the case of strong ionospheric disturbances occurring at Boreal and Antarctic latitudes, the developed methods are observed to largely outperform data-based coregistration. Specifically, coherence magnitude over the Antarctica dataset was observed to be about 0.41 on average with data-based coregistration, and over 0.86 with the proposed multilayer correction.