Soil moisture effects on InSAR - A correction approach and example from a hyper-arid region

We present Interferometric Synthetic Aperture data spanning a series of precipitation events that impacted the southern edge of the Arabian Peninsula in 2017-2018. The arid climate, sparse vegetation and low topographic relief result in very high interferometric coherence magnitude between most pairs of dates, even for those separated by multiple years. For pairs of dates with differing soil moisture conditions, such as a “dry” date and a date immediately following one of the precipitation events, the interferometric coherence magnitude is much lower. However, pairs spanning the same event, but with a longer time interval, have high interferometric coherence magnitude. This observation suggests that the phase changes that result in lower coherence for some pairs are not permanent, such as those that would result from erosion or deposition of material, but are due to the variations in soil moisture. In support of this view, when we compare the phase of individual pixels to their neighbors, we observe similar phase change trends for each precipitation event. We present a simple statistical model of the relationship between soil moisture and phase, and show that it predicts the observed coherence and phase histories within this particular SAR time series. We also show how the parameters of this relationship can be inferred from the InSAR observables, and can be used to reduce the soil moisture effects on coherence and phase even for pairs of dates that were not used in that parameter estimation. For the test data considered here, the noise associated with soil moisture is reduced by 40%. We present results for synthetic time series, including a demonstration of the widely-observed phenomenon that displacement rates inferred from InSAR time series depend on the choice of interferometric pairs used in the analysis.