Pixel-to-pixel Displacement Rate Heterogeneity and Effects on InSAR Time Series Near Laguna Salada in Baja California, Mexico

Interferometric Synthetic Aperture Radar (InSAR) can provide high precision estimates of displacement due to a wide variety of natural hazards and human activity. Many of the processes of interest occur well beneath the surface (e.g., fault slip or groundwater withdrawal) and are spatially smooth at scales larger than a pixel. However, surface and soil processes can affect individual pixels to varying degrees, and can introduce biases over an entire region. We observe this phenomenon on alluvial fans near Laguna Salada, a dry salt lake in northern Baja California, Mexico. We infer displacement rates over $\sim$4.5 years between 2017 and 2022 using unfiltered, full resolution interferograms that are typically very coherent. Throughout this region, individual pixels often steadily diverge in position from their neighbors in a manner that varies smoothly over time. We hypothesize that this variability indicates the existence of soil/surface processes, such as evaporite crystal growth, that may bias InSAR studies focused on deeper deformation sources. We explore independent metrics for evaluating the contribution of surface processes to the time series at each pixel. We show that the inferred rates differ by $\sim$2 mm/yr depending on which subset of pixels is selected based on these metrics. We also observe a convergence of the inferred rates for pixels with increasingly “stable” characteristics, suggesting that, for studies that require $\sim$mm/yr precision on displacement rates, it is worthwhile to attempt to mask the pixels most strongly affected by surface processes before averaging of full resolution interferograms.