Iterative Weighted Least Squares Solution to InSAR Stratified Tropospheric Phase Delay

Phase-elevation regression has demonstrated the ability to mitigate the impacts of vertically stratified tropospheric phase delay on InSAR phase observations. To estimate stratified tropospheric phase delay, a reasonable stochastic model is required, in addition to a high-precision digital elevation model, because the observation noise in the interferogram typically is spatially correlated when corrections, such as turbulence, deformation, and ionospheric delay, are not fully considered, particularly in mountainous areas with abundant water vapour. This work proposes an iteratively weighted least squares approach for stratified tropospheric delay estimation, in which the weight matrix is the inverse of the variance-covariance matrix constructed from the residuals of the last iteration. The Sentinel-1 data over Taihang Mountain, China, encompassing severe tropospheric delay signals, are processed using the proposed approach. The results show that this approach can effectively reduce the bias of the stratified component estimator associated with other spatially correlated phases, and it can achieve robust correction for the stratified tropospheric phase delay in the interferogram. The average standard deviation of 46 consecutive interferograms estimated by the proposed approach was reduced by 22.9%, from 1.66 to 1.28 cm in the global correction, and by 20.9%, from 1.34 to 1.06 cm in the local correction, smaller than the values obtained via the GACOS (1.4 cm) and the raw phase (2.06 cm). This demonstrates the potential of our weighted approach to improve InSAR monitoring in areas with significant terrain variations, such as landslides, mining subsidence, and infrastructure stability assessment.