The Research On Monitoring And Prediction Of Ground Subsidence In Zhuhai City Based On PS-InSAR

As the only city in China's mainland directly connected with Hong Kong and Macao by land,Zhuhai has become a transportation hub connecting southwest China with Hong Kong and Macao.In recent years,with the acceleration of urbanization and the increase of human engineering activities in Zhuhai,geological disasters are more frequent,among which the direct economic loss caused by land subsidence is the most serious,which greatly limits the rapid and stable development of social economy in Zhuhai.At present,the monitoring methods of surface subsidence in Zhuhai are mainly conventional geodetic observation based on discrete points,such as leveling,GNSS monitoring,etc.It is limited by the problem of small monitoring area and high cost,and cannot meet the needs of rapid urban development.Time-series In SAR technology can provide all-weather,high-precision and high-resolution surface deformation information,and has the advantage of large-scale land subsidence monitoring.In addition,it is of great reference significance for urban planning,construction,decision-making and management to predict the spatial-temporal evolution trend of urban surface subsidence based on the current observed deformation.Therefore,based on the time-series In SAR technology,this paper firstly monitored the land surface subsidence in Zhuhai City by using the 55-scene Sentinel-1A data from March 2017 to August 2019,and studied the spatial distribution characteristics and time evolution of the land subsidence in Zhuhai City.On this basis,the main influencing factors of ground subsidence in zhuhai city are analyzed by statistical analysis method and combining the factors of soil geological type,climate and hydrological environment and human engineering activities.Finally,the prediction model of land subsidence is established based on the momentum BP neural network with variable learning rate,and the accuracy of the model is verified.This study can provide reference for monitoring and forecasting methods of urban land surface subsidence and provide basic data for urban disaster prevention and mitigation.The main work and conclusions of this paper are as follows:(1)The high-precision surface deformation field in Zhuhai City is obtained.The Sentinel-1A data of the study area from March 2017 to August 2019 are selected to invert the average annual surface deformation rate and the sequence variable of Zhuhai City by using PS-In SAR technology.The accuracy of the In SAR monitoring results and the measured leveling data are compared and analyzed.The correlation coefficient R~2 between the two is 0.6420.The root mean square error(RMSE)was 3.8114mm/y,which verified the reliability of PS-In SAR monitoring results.(2)The temporal and spatial evolution characteristics of surface subsidence in Zhuhai City are studied.PS-In SAR monitoring results were used to analyze the linear and planar surface features of Zhuhai City.The results show that the surface deformation rate in most areas of Zhuhai City is between-10mm/y and 0mm/y,the northwest area is relatively stable,and there are obvious surface subsidence in central,eastern and southern parts of the city.The main sedimentation centers are concentrated in Gaolan Port Economic Zone,aquaculture areas on both sides of Jiti Men River,Baijiao Town and Hengqin New Area.The deformation of some sections of main highways in Zhuhai City is obvious,and the deformation rate is mostly concentrated at-30mm/y,and the maximum deformation rate exceeds-50mm/y.In addition,the deformation rate of Zhuhai City was reclassified,and the area of the surface subsidence areas of different classification grades were calculated,so as to master the severity of the land subsidence in Zhuhai City as a whole and the affected area of each grade subsidence area.(3)The main influencing factors of ground surface subsidence in Zhuhai City are analyzed.Combined with the factors of soil geological type,meteorological and hydrological environment and human engineering activities,the inducement of main subsidence centers in Zhuhai City was analyzed.According to the analysis results,the main causes of surface subsidence in Zhuhai are soft soil consolidation,upper load and groundwater exploitation.PS-In SAR monitoring results further verify the spatial and temporal evolution details of these subsidence areas,which can provide a scientific basis for urban construction and disaster prevention in Zhuhai City.(4)The prediction model of ground subsidence in Zhuhai City is constructed.Based on the deformation data of the first 45 periods of PS-In SAR monitoring results,the momentum BP neural network model with variable learning rate was established to predict the surface subsidence in Zhuhai City,and the accuracy was analyzed by using the deformation data of the last 10 periods.The results show that the model has high accuracy in predicting the surface subsidence of zhuhai city,and can well reflect the overall subsidence trend in the study area.In order to make the model more practical,by comparing and analyzing the accuracy of the prediction results of different time periods,the best prediction time of the model is determined to be 3 months,which provides a reference for further improving the disaster prevention and reduction and early warning ability of cities.