Monitoring And Predicting The Ground Subsidence By Using Time Series In Sar Technique

In order to relieve the pressure of land use during the urban development,many countries have taken steps to increase the land such as land reclamation and valley filling.These steps expand the area of urban land but may cause severe land subsidence.In this context,it is necessary to monitor and predict the land subsidence due to land reclamation with purpose of ensuring the safety of construction.Interferometric Synthetic Aperture Radar(InSAR)technology has been widely used in land subsidence monitoring due to its wide coverage,allday,all-weather,high precision,high spatial resolution and low cost.However,there are still some problems when using InSAR technology to monitor land subsidence due to the land reclamation,such as digital elevation model(DEM)errors caused by the land excavation and filling as well as the quantitative reasons analysis.In addition,the subsidence prediction of high-filled land based on InSAR observations is still lacking.Based on the above background,this paper takes the Yan'an New District,a typical land reclamation and valley filling region,as experimental area to study the subsidence monitoring and prediction with time-series InSAR technology.During this procedure,improved multibaseline InSAR DEM reconstruction technology was involved in the InSAR time series deformation calculation to correct the DEM errors caused by land excavation and filling.After obtaining the time series deformation of Yan'an New District,quantitative reasons analysis of subsidence is discussed through the fuzzy comprehensive evaluation and cluster analysis.Finally,three models are explored to predict the land subsidence over the study area.This research can provide sound management for the high-filled land.The main research contents and results of this paper are summarized as follows:(1)The time series deformation of Yan 'an New District spanning from October 2014 to December 2020 was obtained with stacking and SBAS-InSAR technologies.Two sets of Sentinel-1A data with ascending orbit were collected and processed to estimate the surface elevation changes and subsidence time series.Considering the DEM changes in Yan'an New Area,an improved multi-baseline InSAR DEM reconstruction technology was introduced to correct the DEM errors.The results show that the subsiding area of Yan'an New District is mainly concentrated in the filling area,where the maximum subsiding rate reaches 10cm/y and the maximum cumulated subsidence reaches 60 cm in the past 6 years.The reliability of InSAR results are verified through InSAR internal verification and comparison of InSAR results with traditional ground-based monitoring results.(2)The correlation and quantitative evaluation of subsiding reasons are discussed over the study area.Based on the InSAR observation and field investigation,fuzzy judgment and clustering analysis methods are comprehensively used to discuss the subsiding reasons.After constructing the hierarchical judgment matrix,it is showed that the proportion of land subsidence accounts for 64%,21%,10% and 5% by the land consolidation,building load,precipitation and engineering construction,respectively.(3)The short-term and long-term prediction of time series deformation was carried out over the study area.Three predicting models including hyperbolic model,Asaoka model and the BP neural network model are used to predict the subsidence.The result shows that the hyperbolic model and BP neural network model have better predicting results in a short time,where BP neural network model is better than hyperbolic model.The Asaoka model has poor short-term prediction results,while it is more stable and suitable for prediction of land subsidence in a long period of time.The Asaoka model is selected to predict the settlement development curve of three points with a land-filling depth of about 50 meters,80 meters,and100 meters in the next five years.The results show that the three points are basically in a stable state of deformation in the next five years,and the maximum subsidence will be 45 cm.