Vegitation, smegitation: How InSAR research on tectonics and land surface change has expanded from the deserts to the rain forests and beyond.

Abstract:

The upcoming avalanche of openly available SAR imagery is expanding the horizons of what can and cannot be accomplished with InSAR. Historically, InSAR performed best in arid regions - an observation that became a bit of a self-fulfilling prophecy as satellite agencies acquired less data in regions with vegetation and agricultural activity. Here we present an overview of the development of InSAR research in tectonics and land surface change over the past 2+ decades, with a view towards some of the new advances in modeling and data management that will be necessary to fully take advantage of SAR data in the near future. One of the most basic breakthroughs will be that issues with decorrelation and the need for sophisticated time series analysis just to enable phase unwrapping will no longer be a major problem over many areas of the globe. New challenges will be the interpretation of coherent signal related to vegetation, soil moisture and time-variable phase scattering height in regions that previously would have just been flagged as “noise”. We present results based on the ingestion of independent optical and radar observation types into SAR time series analysis, with applications to deformation sources in the Central and Eastern United States.

The land surface properties in the Central and Eastern United States differ from those in the arid regions where InSAR has often been used, both in the presence of vegetation and the often very rapid changes in surface scattering characteristics that occur seasonally and during single events (snowfall, flooding, etc.). In the past, observations were so sparse that these changes resulted in decorrelation, rendering the data unusable. However, shorter acquisition times and a wider range of radar wavelengths allow the extraction of coherent signals from these areas, even spanning large snow storms. The resulting data contain signals that were often disregarded during InSAR time series analysis, but that must be either accounted for or masked when no sufficient model exists. We present examples associated with agricultural and urban areas, where the addition of independent data types can help classify the surface and improve the extraction of true ground motion.