Evaluation of Vegetation Bias in InSAR Time Series for Agricultural Areas Within the San Joaquin Valley, CA

Agricultural regions present a particularly difficult set of challenges during interferometric synthetic aperture radar (InSAR) displacement time series analyses due to the existence of abrupt transitions in land use over short spatial scales and rapid temporal changes associated with different stages of the agricultural cycle. Plant growth and soil moisture changes can introduce phase biases within interferograms that could be misinterpreted as displacement. We analyze a full-resolution, multi-year SAR time series over California's San Joaquin Valley, an intensively cultivated region producing a wide variety of crops. Using independent information about land cover and crop type, we isolate the effects of individual crops on backscatter amplitude, interferometric phase change, and interferometric coherence over space and time. We determine the temporal behavior of the phase changes associated with several key crop types by isolating the difference between the phase of pixels averaged over each agricultural field and the phase values of pixels in nearby roads and developed areas. We find that some fields are associated with a bias of ∼2–4 cm/yr of apparent subsidence, with strong seasonal variability in the degree of bias. When InSAR imagery is spatially averaged or filtered, these biases also impact the inferred phase in nearby roads and other land cover types. We show that even a simple approach, where pixels associated with agricultural fields are removed or masked out before further processing, can mitigate the crop-related biases that we observe in the study area.