A Multiple-Regression Model Considering Deformation Information for Atmospheric Phase Screen Compensation in Ground-Based SAR

The accuracy of ground-based synthetic aperture radar (GB-SAR) interferometry is highly affected by the atmospheric phase screen (APS). A multiple-regression model (MRM) for APS estimation with two parameters (line of sight (LOS) distance and the height difference between the targets and the GB-SAR sensor) was proposed. However, this MRM is inapplicable if the LOS deformation of the observed objects has similar behavior to that of the APS. For example, both the APS and the LOS deformation of observed objects can have an increasing trend along the LOS direction. In this article, an improved MRM (IMRM) that considers the spatial pattern of the LOS deformation of observed objects was proposed. The spatial pattern can be expressed by the location information of the observed objects. The applicability of this improved model was demonstrated in a deformation-measured campaign carried out on a high loess cut slope using 285 images obtained by the GB-SAR sensor during a three-day measurement. The similarity and dissimilarity between the APS and the behavior of the LOS deformation of the slope were illustrated and used to estimate the APS. The accuracy of this IMRM was verified using deformation data measured by a displacement meter. This IMRM will be helpful in APS estimation and compensation in GB-SAR interferometry in case studies where the APS and the LOS deformation of observed objects have similar behavior.