Advanced InSAR Tropospheric Correction from Global Atmospheric Models while Considering Spatial Stochastic Interpolation Strategies

Abstract:

Thanks to the continuous improvement of global atmospheric models (GAMs) in both quality and spatio-temporal resolution, the GAMs have shown great potential in correcting InSAR tropospheric delay signals. However, the weighting strategy of reconstructing high-resolution (~10s meters) InSAR tropospheric delay maps from coarser (~10s kilometers) GAM outputs has been largely overlooked in previous GAM correction studies. We present here an advanced GAM-based tropospheric correction method, towards further improving InSAR geodesy, by incorporating spatial stochastic models of the troposphere in the correction calculations. We interpolate the tropospheric parameters (temperature, pressure, and partial pressure of water vapor) according to the correlation between a pixel of interest and GAM grid locations (3D) flexibly based on the spatial variabilities of the tropospheric random field, instead of subjectively using an inverse distance method or using a local spline function. We also estimate the integral of the tropospheric delays along the satellite line-of-sight (LOS) direction directly, instead of calculating the projected zenith-delays, because the troposphere is not uniformly stratified. The new method can be generalized to all GAMs, and we implement it with the latest ECMWF (European Center for Medium-Range Weather Forecasts) ERA5 reanalysis data in this study. We validate the new method by using hundreds of Sentinel-1 images with tropospheric corrections for both interferograms and time-series analysis products (deformation velocities and time-series solutions). We also compare the new method with other GMA-corrections, including PyAPS, d-LOS, and GACOS, to demonstrate (1) the importance of considering the tropospheric stochastic models in GAM-corrections, and (2) the importance of considering the horizontal heterogeneities when estimating the LOS delays.