In Beijing, there has been serious land subsidence due to overexploitation of ground water in recent years. It's becoming crucial to evaluate the spatial and temporal heterogeneity for hazard control. In this study, we derived spatial-temporal evolution of land subsidence in Beijing plain area using 48 scenes of ENVISAT-ASAR data by Permanent Scatterer (PS) InSAR techniques. The monitoring results agreed well with leveling measurements. To better understand the uneven characteristics of settlement, Self Organizing Mapping (SOM) was then used to classify the PS points based on the time-series deformation. Six categories were produced and represented different patterns of time-series settlements.