Settlement monitoring and prediction using network model and time-series InSAR in large-scale land creation areas: A case study of yan’an New Area, China

Abstract

At present, there are fewer studies on the monitoring of surface settlement caused by large-scale land creation projects and the detection and identification of potential engineering landslides, and the traditional ground settlement simulation and prediction model based on the need for a large amount of hydrogeological data and measured data, which is difficult to simulate and predict the deformation of the landslides of the land creation projects affected by a variety of factors. In this paper, we obtain the ground deformation rate field information of Yan’an New Area, Shaanxi Province, based on the Small Baseline Subsets-Interferometric Synthetic Aperture Radar (SBAS-InSAR) technology, and analyze the causes of subsidence in the study area using multi-source remote sensing data; combining the optical image and exponential model, we detect and identify the engineering landslides in Yan’an New Area. In addition, we considered the effects of stratigraphic lithology, filling depth, surface temperature, precipitation and soil moisture on landslide deformation, and combined the Particle Swarm Optimization-Back Propagation (PSO-BP) neural network model to predict the time-series deformation values of landslides. The study shows that: (1) the ground settlement in Yan’an New Area after the mountain filling and city building project is mainly distributed in the filling area, and the radar line-of-sight deformation rate of Yan’an New Area from 2019 to 2022 is −44.92 ∼ 19.24 mm/a, mainly distributed in the center of Qiaoergou, Gaojiagou, and Tanyaogou area. (2) There is a high correlation between the ground settlement in Yan’an New Area and the project filling area, with an overlap of 92.44 %, in addition to the change in land use classification in the study area and the building loads also have some influence. (3) Fifteen potentially hazardous subsidence landslides in Yan’an New Area were accurately identified during the study period, and the correlation between the rate of landslide subsidence and the depth of fill reached 0.89. In addition to precipitation, changes in soil moisture and surface temperature can accelerate the subsidence of engineered landslides. (4) The PSO-BP models developed using a combination of stratigraphic lithology, filling depth, precipitation, surface temperature and soil moisture data outperform PSO-BP models using only precipitation data, and the correlation coefficient (R²) obtained is 0.96, the mean absolute value of the error (MAE) is 0.66 mm, and the root-mean-square error (RMSE) is 0.76 mm, which can predict the settlement of the landslides of the long-time sequence project effectively. This study can provide reliable technology support for the prevention and control of surface and landslides settlement in large-scale land creation areas.