Spatial statistics of atmospheric signal in repeat-pass InSAR

Spatial statistics of atmospheric effects in InSAR measurements are of great importance for understanding and mitigating the effects, which provide reliable tools to assess the InSAR data quality and the derived parameters (e.g., elevations and deformations). Besides, a comprehensive cognition of the spatial statistics of atmospheric effects could be conducive to the development of InSAR meteorology. Traditionally, the estimation of the spatial statistics of atmospheric effects is generally based on the hypothesis of isotropy, which is however not realistic since the spatial variation of the atmospheric signal is always anisotropic. In this paper, the geostatistics method is adopted to test the anisotropy ratio and model the spatial variation of the effects. Atmospheric effects in InSAR measurements over Shanghai region are analyzed, where two pairs of ERS-Tandem images are used to carry out the experiment. The results showed that, for the random fields of InSAR atmospheric effects, the correlation lengths range from 25 km to 55 km. The anisotropy ratios all exceed 1.4, which indicates that the anisotropy of InSAR atmospheric effects should not be ignored and the spatial variation model of atmospheric delay should not be simply assumed as isotropy. Therefore, considering the anisotropic model of atmospheric effects in the underlying processes of InSAR data is very essential.