Xiao, R., Yu, C., Li Z., & He, X.
The atmospheric effect represents one of the major error sources in Interferometric Synthetic Aperture Radar (InSAR), and its mitigation is found crucial for high-precision InSAR applications. Numerous studies on InSAR atmospheric correction methods and applications covering a wide range of regions worldwide have been reported with varying degrees of success. However, more efforts on the performance assessment are needed, and the conclusions may lack statistical significance due to the limited interferograms involved in most of these analyses. To optimally utilise different InSAR atmospheric correction methods and avoid potential uncertainty caused, appropriate statistical metrics to assess the correction performance must be set up. In this work, we provide a general guideline for statistical assessment of InSAR atmospheric correction. Based on the physical properties of the atmosphere, three metrics are applied: (i) the phase standard deviation which assesses the overall performance of the correction; (ii) the spatial structure function which evaluates the reduction of the long-wavelength atmospheric effect; and (iii) the phase-elevation correlation coefficient which measures the reduction of the stratified component of the atmospheric delay. The performance of the Generic Atmospheric Correction Online Service for InSAR (GACOS) products for two typical terrains in Eastern China is evaluated. Statistical results of the 1250 Sentinel-1 interferograms covering the Yellow River Delta and Shandong hilly region show that (i) GACOS reduces the interferometric phase standard deviation in 84.6% of the interferograms by an average of 36.4%; (ii) the phase decorrelation distance decreases from 321 km to 225 km on average after correction; and (iii) the mean phase-elevation correlation declines by 33.3% for the areas with considerable topographic variations. The results verify the effectiveness of GACOS products in Eastern China for the first time, and the three proposed metrics further characterise the sources of improvement after correction.