Landslides are destructive to property, infrastructure and people in potential landslide zones. Identifying potential landslides is an important step in landslide preparedness and will help develop sustainable landslide risk management. Interferometric synthetic aperture radar (InSAR) and landslide susceptibility assessment (LSA) have poor reliability in individually identifying potential landslide zones. This study proposes a threshold model for identifying potential landslide zones that fuses the InSAR two-dimensional (vertical direction and east-west direction) deformation rates and LSA results. The deformation rate threshold for this threshold model is |D_U| or |D_E|>10 mm/year (D_U and D_E are the vertical and the east-west deformation rates, respectively), with threshold levels of LSA set to high and very high susceptibility. The criterion of potential landslide zones is ((|D_U| or |D_E|>10 mm/year) ∩ (high or very high susceptibility of LSA)), and points with similar deformation and susceptibility are clustered by the K-means algorithm, and the potential landslide zones are obtained by elimination, smooth and speckle removal operations. The results showed that the InSAR two-dimensional deformation rates D_U and D_E were −32.71 − 12.72 mm/year and −14.88 − 24.81 mm/year, respectively, during 2015–2020 in Lanzhou city. The LSA showed that very low, low, medium, high, and very high susceptibility accounted for 55.36%, 10.54%, 21.37%, 9.63%, 3.1% of the total area, respectively. Using the proposed threshold model, 117 potential landslide zones were identified in Lanzhou. The overlap rate between potential landslide zones and the landslide inventory was 40.17%, indicating that about 40% of the potential landslide zones overlapped with the landslide inventory and that about 60% were new potential landslide zones in Lanzhou. The feasibility of the threshold model in identifying potential landslides was confirmed by field research and time-series InSAR analysis on typical areas (L1, L2, L3, and L4), which had large deformation variables and landslide features. The proposed method can quickly determine the spatial location of potential landslides, providing targeting data for landslide field investigations, technical support for rapid early landslide identification, and data support for landslide management and prevention in Lanzhou.

While surface deformations and their impact on buildings have been observed in the city of Lanzhou, it is difficult to find studies of surface deformation and the influential factors in the recent decades. This study was designed to detect the spatial position of these surface deformations and to understand the mechanism behind them. Sentinel-1A TOPS model image data acquired in descending orbits between March 2015 and May 2019 were processed by using Small Baseline Subsets Interferometry (SBAS)-Interferometric Synthetic Aperture Radar (InSAR) technology, and then compared with geology, tectonical aspects of the study area and land cover types in Lanzhou. The results reveal that the land surface deformation is uneven in this city, and seven surface deformation regions were detected in the following areas: the north freight yard, Jiuzhou, Country garden, Donggang, Yanjiaping, Zhongxinping and Liuquan town. The land surface deformation rate in Lanzhou ranges from −82.13 mm/year to 19.31 mm/year. Time-series land surface deformation analysis showed that deformation increased over time in major deformation regions. Surface deformation expansion was significant after June 2017and it continued to expand. The surface deformation of Lanzhou is affected by natural factors (geology and geological faults) and human activities (land cover types/land cover changes). Local geological conditions control the location of the surface deformation process. These findings provide compelling data and theoretical support for disaster prevention and reduction in Lanzhou.