Persistent scatterer interferometric synthetic aperture radar (PS-InSAR) is a widely used technique for local ground deformation estimation due to its millimetric accuracy and full-resolution results. However, with the enhancement of data acquisition capability of the SAR sensors, larger coverage and higher temporal-spatial resolution of SAR images can cause explosive increase and uneven distribution of PS points. PS network constructed between PSs with different qualities may cause spatial-error propagation. Large PS points list with large amount of data also have high requirements on computing performance and storage space. Those problems can bring big challenge for PS-InSAR technique in processing capacity, deformation monitoring accuracy and algorithm efficiency. In order to effectively overcome those limitations, this paper proposed a novel block PS-InSAR method that uses the approach of partitioning a study area into regular blocks with overlapping regions, selecting a high-quality reference point for each block and eliminating the discontinuity of the parameter calculating results in adjacent blocks by the weighted least square method. We used this approach to analyze ground subsidence in over 13,000 km² of southern California, and it performed successfully when 69 Sentinel-1 images were used and 21,029,968 PSs were selected. Only approximately 10 h was needed for this experiment. Comparing the deformation rate and time series of the InSAR results with 29 GPS observations, the mean and standard deviation of difference of deformation rate were − 0.63 mm/yr and 1.53 mm/yr, respectively, and the average root mean squared error of deformation time series was 3.7 mm. The experiment demonstrated that this method could achieve high accuracy results without spatial discontinuity and increase the computational efficiency significantly.

持久散射体干涉合成孔径雷达（PS-InSAR）由于其毫米级精度和全分辨率结果，是一种广泛用于局部地面变形估计的技术。然而，随着SAR传感器数据采集能力的增强，SAR图像更大的覆盖范围和更高的时空分辨率会导致PS点的爆炸性增加和分布不均。在不同质量的 PS 之间构建的 PS 网络可能会导致空间误差传播。数据量大的大型PS点列表对计算性能和存储空间也有很高的要求。这些问题都给PS-InSAR技术在处理能力、变形监测精度和算法效率方面带来了巨大挑战。为了有效地克服这些限制，本文提出了一种新的块 PS-InSAR 方法，该方法利用将研究区域划分为具有重叠区域的规则块，为每个块选择一个高质量的参考点，并通过算法消除相邻块中参数计算结果的不连续性。加权最小二乘法。我们使用这种方法分析了超过 13,000 公里的地面沉降²南加州，当使用 69 个 Sentinel-1 图像并选择 21,029,968 个 PS 时，它成功执行。这个实验只需要大约 10 小时。将 InSAR 结果的变形率和时间序列与 29 个 GPS 观测值进行比较，变形率差异的平均值和标准差分别为 − 0.63 mm/yr 和 1.53 mm/yr，变形时间的平均均方根误差系列为 3.7 毫米。实验表明，该方法可以在没有空间不连续性的情况下获得高精度的结果，并显着提高了计算效率。