Deformation Estimation for Time Series InSAR Using Simulated Annealing Algorithm

Abstract

Time series interferometric synthetic aperture radar SAR (TSInSAR) is one of the most important surface deformation monitoring techniques, and has been widely used in geodesy. Deformation estimation is one of the main steps of TSInSAR processing, so an effective and efficient algorithm is necessary. Present algorithms have some limitations such as computing c osts or errors caused by local extremums. In this work, a novel deformation estimation method based on the simulated annealing (SA) algorithm is proposed to handle this problem. The SA algorithm uses a random search to avoid local extremums and thus can be more likely to get the global optimal solution of deformation. By adopting a better annealing method, this algorithm gets high precision deformation results in less time than most present algorithms. In addition, it can estimate complex nonlinear deformation without adding any computing costs. The results, tested on the real SAR data, confirm the reliability and effectiveness of the SA-based deformation estimation algorithm.