The investigation of slow displacement in urban areas using the multi-baseline DInSAR technique has been a hot research topic in the field of DInSAR. The basic flow of this technique includes several steps such as the combination of interferometric image pairs, generation of differential interferograms, selection of high coherent points, generation of the Delaunay triangular network, calculation and integration of increments in network, unwrapping and calibration of the residual phase, and the estimation of both atmospheric and nonlinear displacement phase. Among these steps, the calculation of increments is the key to retrieve linear displacement, while unwrapping and calibration of the residual phase are the keys to retrieve nonlinear displacement. In order to improve the performance of these two steps, this paper proposes a modified model coherence function for increments estimation, and a triangular "circle" algorithm to deal with phase unwrapping and calibration. Based on the above algorithms, the subsidence of Suzhou City is investigated using 24 ERS scenes from February 1993 to December 2000. The results show that the linear subsidence velocity of the most urban area is about -20 to -30 mm/a during the time, with a yearly decrease in velocity. The displacement seems to be stable after 2000. Leveling data validate our results and demonstrate the reliability of the algorithm.
