Modeling water vapor using GPS with application to mitigating InSAR atmospheric distortions

Here we present a new approach for correcting the phase signatures using measurements of neutral atmospheric delay from GPS receivers within, or close to, regions imaged by InSAR. Atmospheric water vapor corrupts InSAR phase measurements of ground deformation, often precluding their use for geophysical studies.; Our method for estimating maps of water-vapor-induced signal delay, which is proportional to total column water vapor, uses a time series of measurements from a sparse network of GPS receivers. We estimate separately components of atmospheric phase distortions caused by (1) vertical-stratification of water vapor and (2) turbulent mixing of water vapor in the atmosphere. Solution and correction for the altitude dependence of GPS delay acquired at the radar observation times reduces InSAR phase distortions due to a layered atmosphere by 46%, in one image acquired over Los Angeles. To correct the turbulently-mixed component, we spatially-interpolate the GPS measurements, and find that the resulting correction is limited by the spatial sparsity of GPS observations.; We have developed two techniques to overcome the sparseness of GPS measurements by using additional GPS measurements before and after the radar observation times. The first method is based on the "frozen-flow" hypothesis which posits a flow-driven, advecting slab of atmosphere with homogeneous spatial statistics. We estimate mean flow from covariance analysis of the GPS delay time series, which we use to infer denser networks of virtual control points. We interpolate new networks and remove short-scale variations not reproducible by the GPS data, thereby obtaining a delay map that reduces turbulently-mixed atmospheric phase error by an additional 31%.; In the second method, we describe the spatio-temporal variation of water-vapor-induced delay fields measured by the GPS network using a deterministic transport model. This model is derived by assuming the conservation of mean humidity in atmospheric flow consisting of laminar and turbulent components. Using this model, we develop an algorithm to estimate spatially-variable flow fields and derive a delay map from GPS measurements acquired around the radar observation times. We find that this correction reduces the turbulently-mixed atmospheric phase by an additional 18% compared to using data only at the radar observation times.