InSAR phase errors induced by Faraday rotation

Abstract:

Microwaves propagating through the ionosphere are subject to Faraday rotation. In single- or dual-pol data, there is insufficient information to rigorously correct for Faraday rotation, and usually no attempt is made to correct for it. Uncompensated Faraday rotation distorts interferometric observations, but in contrast to polarimetric information, the associated phase errors have received scant scrutiny.

Here, we focus on the errors in repeat-pass interferometric observations over distributed targets. We attribute the Faraday-induced interferometric phase errors to the leakage of polarimetric phases into the interferometric phase and to interferometric phase diversity. To assess their magnitude and nature, we combine theoretical analyses with an empirical quantification for a range of land covers using L-band observations with simulated Faraday rotation. We determine the errors in the observed and in the split-spectrum ionospherically corrected phase for the co-pol and the cross-pol channels.
These results show that the typical error magnitude is up to 1 mm in the co-pol channels, but it may exceed 5 mm for intense solar maxima and surfaces with adverse scattering characteristics. The cross-pol channel is much more prone to severe errors, which can exceed several centimeters.

These errors are systematic, as they can add up and persist over time. Their temporal characteristics, such as pronounced seasonal and quasi-decadal variability, are similar to those of common deformation processes. They are further strongly associated with the topography and land cover because they result from the modulation of ionospheric signals by surface scattering characteristics. Even when these errors remain subtle (say, 1 mm at L band), their systematic nature makes the spurious Faraday-induced patterns prone to being misinterpreted as deformation.

Because these systematic errors cannot be removed rigorously, they deserve to be accounted for in quantitative analyses.