Human society has entered the big data era given the exponential growth of remote sensing data due to the emergence of higher resolution, frequent revisits, and multi-platform image acquisitions. This phenomenon raised challenges for Interferometric SAR (InSAR) and Multitemporal InSAR (MTInSAR) data processing in near real time. For instance, the traditional PSInSAR algorithm can no longer satisfy a fast response monitoring due to delay in deformation time series updating.To address the aforementioned technical limitations, we proposed a PSInSAR sequential processing algorithm characterized by the optimized searching-space to achieve the performance improvement of Differential InSAR (DInSAR) data preprocessing and MTInSAR parameter estimation. In this approach, new SAR acquisitions were seamlessly integrated into the reconstructed spatiotemporal baselines of interferograms. Then, a triple-level Delaunay network was established using the temporal coherence value (high, low, and decorrelated) on network edges. On the basis of the value inheritance from previous PSInSAR estimations, the unknown parameter estimation on PS candidates for the sequential PSInSAR was accelerated owing to the proposed searching strategy adopted to the triple-level coherence network edges. That is, the solution space was first sampled using a large searching step (for example, 10 times the measurement accuracy of unknown parameters) to determine the potential interval of the optimal solution. Then, the choice of network edge was determined on the basis of the maximum value of the temporal coherence, followed by a dedicated fine searching (with the step equivalent to the predetermined accuracy of unknown parameters) concentrating on the potential interval for the optimal inversion. Owing to the applied global-local searching strategy, the optimization of calculation efficiency and estimation accuracy can be achieved.We conducted a comparative investigation for the deformation estimation and performance assessment between the current PSInSAR and the proposed sequential PSInSAR methods using 32 scenes Cosmo SkyMed Stripmap images (in descending orbits and acquired from January 2015 to February 2018) covering the Nanjing Ming Dynasty City wall. Results indicate that a high efficiency of unknown parameter estimation (height, deformation, and thermal dilation) was obtained using the sequential PSInSAR with the adopted optimized searching-space approach, with the computation acceleration with approximately an order of 10 times. The cross comparison of the deformation velocity rates from both approaches reveals a consistent estimation as presented by the overall dispersion values ranging from 0 to 1 mm/a, which validates the feasibility and reliability of sequential PSInSAR in the deformation estimation.The driving force of detected deformation anomalies along three sections of the city wall was further exploited, providing new insights for the sustainable conservation of the heritage properties. This study implies the potential of the sequential PSInSAR method in the accurate, near real-time deformation monitoring, and preventive conservation of large-scale cultural heritage sites (i.e., Nanjing Dynasty City Wall), particularly on the emergence of big data.

remote sensing;PSInSAR;sequential processing;optimized searching-space;cultural heritage