Multi-temporal SAR And Optical Remote Sensing For Identification Of Illegal Rare Earth Mining

Ganzhou city is rich in rare earth mineral resources.Due to the particularity of rare earth ore occurrence and the unitary supervision means of illegal mining,illegal mining has become increasingly serious,which has seriously affected the ecological environment of soil,surface and water in the surrounding area.Ganzhou city set no rare earth mine mining,the author of this paper,therefore,the related problems of the in-depth analysis and exploration,the comprehensive utilization of InSAR technology and the change detection method,Sentinel,a series of images as data source,around the "rare earth mining area of digital elevation model inversion,rare earth surface deformation monitoring and identification of illegal mining area of rare earth mining area" study,The research content mainly includes the following four aspects:(1)Based on the inversion of DEM by traditional InSAR technology,an improved inversion method using coherence coefficient is provided to obtain DEM of rare earth ore with finer texture information.The pixel points with low coherence were eliminated by setting the coherence coefficient threshold,and the SRTM DEM was used as the reference DEM for verification.The results show that,compared with the traditional InSAR inversion DEM method,the orbit error of this method is reduced by 52.11%,and the root mean square deviation and standard deviation of inversion results are reduced by 75.78%and 8.53%,respectively,indicating that the DEM obtained by this method is more accurate.At the same time,the error sources of coherence estimation are analyzed from the aspects of satellite sensor,polarization mode,season,ground object category and data processing process,and the correlation between coherence coefficient and deformation results and ground object category is explored.(2)To solve the problem that blind selection of control points may introduce errors due to the lack of deformation prior knowledge and ground control point information during orbital refining,an improved DInSAR technique was proposed.Dingnan County of Ganzhou City was taken as the research area.With the help of Sentinel-1A satellite data from 2015 to 2020,certain coherence coefficient threshold was set.Suitable highly coherent pixels are selected as control points for orbit refining.Compared with the conventional DInSAR method,the accuracy of control point selection was improved by72.9%.On this basis,DInSAR surface deformation information extraction method based on coherence coefficient is proposed by taking the DEM inversion results of improved by coherence coefficient as external DEM.The results show that compared with the improved DInSAR technology,DInSAR based on coherence coefficient can reduce the root mean square deviation and standard deviation by 20.68% and 15% respectively,and the deformation result has a higher resolution.(3)In view of the problem that change detection of medium and low resolution images by conventional classification methods cannot achieve ideal results,a change detection method of rare earth mining area based on ratio method is adopted.Integrated radar data and the advantages of optical image,to set the pingnan rare earth mining area as the research object,this method is illustrated by comparing difference method compared with single band ratio,principal component analysis method and principal component analysis difference method and so on several common advantages of the application of remote sensing change detection methods,summarizes the advantages and disadvantages of different methods.It is also used to detect the mining change of rare earth mining area.The results show that the ratio method has the best effect in mining change detection,and the accuracy rate,false alarm rate and Kappa coefficient are 90.91%,8.33% and 0.793,respectively.Compared with other change detection methods,ratio method not only improves accuracy,but also reduces false alarm rate.(4)In view of the phenomenon of illegal mining in rare earth mines,a recognition method for illegal mining of rare earth mines taking into account radar and optical remote sensing is proposed.Based on the coherence coefficient,intensity information,deformation information and spectral information of Sentinel series images,the change areas of radar and optical images were detected respectively,and six illegal mining areas were delineated according to typical features,change standards and existing mining rights information.Finally,according to the time series deformation diagram obtained by SBAS-InSAR technology,the suspected illegal mining area is analyzed.The results show that,compared with the identification of illegal mining using a single image,the identification effect of illegal mining using SAR and optical remote sensing is better,the accuracy is increased by 11.6% and 4.46% respectively,and the false alarm rate is reduced by 38.24% and 23.53% respectively,which verifies that the method can quickly and reliably identify illegal mining areas.At the same time,the variation trend of illegal rare earth mining area is highly consistent in the coherence coefficient diagram and the surface deformation diagram,and the coherence coefficient and surface deformation range from0.37 to 0.71 mm and-25 to 15 mm,respectively.