Modeling InSAR closure phase and soil moisture inversion

Abstract

Closure phases in SAR interferometry have the potential to carry information about the scatterer and changes in its dielectric properties. We are modeling the effects of scatterer changes on the interferometric phases. In this contribution we present some result and open questions. We start by using the interferometric model in De Zan et al. 2014 to explain the closure phases observed in InSAR with moisture variations and invert them. This model seems to work rather well in L-band and we report inversion results for the CanEX-SM10 campaign by UAVSAR (NASA/JPL) over an agricultural site in Saskatchewan, Canada. Here the field-based correlation between our inversion and moisture probes are higher than 0.7 for most of the fields. The scaling factor between the inversion and the probe signal is, however, highly variable and we do not have a satisfactory model for it yet. We report also encouraging first results with C-band modeling over southern Italy observed with Sentinel-1. The closure phases obey a similar model as for L-band, but they are typically downscaled, i.e. we observed values smaller than the model would predict. There is a possibility that the moisture model is not explaining the totality of the observed closure phase. This idea is suggested also by processing interferometrically stacks of SAR images and limiting the temporal separation to a few months. The retrieved deformation exhibits a spurious drift that amounts to a few millimeters / year. The drift is rather constant and does not seem to be related to the moisture cycles. By adding just a few (3) parameters to the interferometric models it is possible to explain significantly better both the observed closure phases and the drifts when reconstructing a deformation series by limiting the temporal separation.