Evaluation of Time-Series InSAR Tropospheric Delay Correction Methods over Northwestern Margin of the Qinghai-Tibet Plateau

  Objectives  Time-series analysis of interferometric synthetic aperture radar (InSAR) plays an important role in monitoring large-scale surface slow deformation. However, the impact of tropospheric delay limits the accuracy of mapping deformation. Tropospheric delay correction is crucial to InSAR deformation rate inversion objective.

  Methods  We focus on the northwestern margin of the Qinghai-Tibet Plateau and use three methods which are the empirical model linear correction, generic atmospheric correction online service for InSAR (GACOS) and ERA5 dataset released by the European Centre for Medium-Range Weather Forecasts (ECMWF) in the time-series InSAR inversion of deformation rate to study their applicability in this region.

  Results  By calculating the standard deviation in the non-deformation region with the mask technique, analyzing the correlation between deformation rate and topography, and comparing with the GPS data, the results show that three study areas from west to east, in the range of D136 (76.5°E—79.7°E), D165 (80.5°E—83.7°E) and D19 (84.9°E—88.1°E), the standard deviations of linear corrected are decreased by 41.05%, 59.21% and 25.13%. The standard deviations of GACOS corrected are reduced by 38.76%, 55.97% and 30.73%, and those of ERA5 corrected are decreased by 10.05%, 30.11% and 20.15%, respectively. Besides, the correlation coefficients between deformation rate and elevation of the three profiles are 0.68, 0.93 and 0.60, respectively. With the increase of correlation coefficients, the advantage of linear correction becomes obvious. In addition, the comparison of LOS deformation rates between InSAR and GPS projected shows that the root mean square error of linear correction, GACOS correction and ERA5 correction in the three study areas are reduced by 46.07%, 51.28% and 35.51% respectively.

  Conclusions  The northwestern margin of the Qinghai-Tibet Plateau, all three methods can mitigate tropospheric effects, and the linear correction and GACOS correction perform better and have higher applicability. While the performance of ERA5 is slightly poor due to the density of ground monitoring stations. Considering that terrain, altitude, geographical location and other factors will affect the atmospheric delay, the atmospheric correction methods and correction effects applicable to different regions need to be comprehensively studied and analyzed based on the specific situation.