Deformation Retrieval Using the Spatially Constrained MTInSAR Method

Abstract

The observation model of multitemporal inteferometric synthetic aperture radar (MTInSAR) is an underdetermined system. To obtain a unique solution, the traditional techniques impose the temporal constraints by assuming the deformation pattern typically as, e.g., a linear or polynomial model. However, these temporal constraints are not necessarily compatible with the realistic deformation, especially for complex deformations of, e.g., landslides and permafrost. Such discrepancy will bias the retrieval of MTInSAR parameters, thus producing inaccurate deformation results. In this letter, we propose a method for directly solving the deformation sequence by imposing the spatial similarity constraints instead of temporal constraints. The underlying rationale is that the spatially closer points share more similar deformation patterns for most motion events. The capability of the proposed method for retrieving displacement is initially demonstrated by using both simulated and real data experiments.