Research On SAR Offset-Tracking Method For Monitoring Large Gradient Surface Deformation In Mining Area

China is a big energy country with coal consumption as the main consumption,and its coal storage is also in the forefront of the world.Although coal mining has driven the development of China's economic construction,it also brings a series of environmental problems.Coal mining will directly cause the mining area surface subsidence,and then lead to subsidence water,building damage,surface cracks and other ecological environment disasters.Therefore,strengthening the monitoring of mining area surface deformation is an important basis for the prevention,control and governance of these problems.Synthetic aperture radar interferometry(InSAR),as a new earth observation technology,has developed rapidly in recent years and has been widely used in practical production.Although InSAR is not affected by cloud and rain,all weather,large-area monitoring and other characteristics,but because of the mining area surface subsidence is often very large,resulting in serious decoherence of synthetic aperture radar(SAR)interferometry,it is difficult to obtain accurate deformation results.In order to solve this problem,this paper will focus on the research of SAR image based offset tracking method to obtain large gradient deformation information.The main research contents are as follows:(1)This paper discusses the background and significance of the topic,summarizes the research status of InSAR technology applied in mining area surface deformation monitoring at home and abroad,and expounds the basic principles of differential synthetic aperture radar interferometry(D-InSAR),offset tracking and InSAR technology to obtain three-dimensional deformation.(2)Aiming at the problem that the traditional offset tracking method is easily affected by the size of cross-correlation window and the computational efficiency is low,an adaptive window selection method of offset tracking combined with CPU and GPU is studied.Firstly,the principle of adaptive window and the principle of CPU +GPU collaborative computing are briefly introduced.Then,the influence of different windows on offset tracking monitoring results is analyzed by using Terra SAR-X data,that is,small windows are used to monitor in areas with low noise and large shape variables,and large windows are used to suppress noise in areas with large noise.Finally,the adaptive window selection method is verified by examples The monitoring accuracy is higher,and the CPU + GPU collaborative computing speed is faster.(3)In order to improve the monitoring accuracy and applicability of offset tracking method in the areas with large topographic relief and high altitude,an offset tracking monitoring method for surface deformation in mining area considering topographic relief is studied.In this method,the DEM 30 m data is used to assist the refinement of the query table,and the terrain error caused by terrain fluctuation is removed.In this paper,the Terra SAR-X and alos-2 data are used to verify,and the Terra SAR-X image experimental results are compared with the measured data.The results show that the root mean square error of this method is 0.201 m,and the traditional method is 0.242 m.Compared with the traditional method,this method improves the monitoring accuracy by 41 mm,which verifies that this method can remove the terrain error caused by terrain fluctuation.(4)In order to weaken or eliminate the influence of gross errors on multi-source data to solve large gradient surface deformation in mining area,a method of weighted multi-source data to solve three-dimensional surface deformation in mining area is proposed.Firstly,the offset tracking method is used to obtain the range and azimuth two-dimensional deformation field,and then the probability integral method is used according to the relevant mining data of the working face Method,PIM)to simulate the three-dimensional deformation of the working face,and transform it into the range and azimuth deformation;finally,this paper uses the measured GPS data to build a fitting model,fitting the weight of each data,building a weight matrix,and solving the threedimensional deformation of the study area through the least square method.The experimental results show that the root mean square error of the method is 0.149 m in vertical direction,0.127 m in north-south direction and 0.164 m in east-west direction.Compared with the traditional method,it is increased by 49 mm,27mm and 56 mm respectively,equivalent to 33%,21% and 34%,which also proves that the weighted method has higher accuracy and is more conducive to the solution of 3D deformation.There are 47 figures and 11 tables and 96 references in this paper.