The Application Research Of Surface Subsidence Monitoring And Time-series Interpolation Prediction In Mining Areas Based On InSAR Technology

The current research hotspot in the field of surface subsidence monitoring in mining areas is the use of synthetic aperture radar interferometry technology to process remote sensing images to obtain surface deformation characteristics,which has the characteristics of all-day,all-weather,wide range and high monitoring accuracy,and can quickly obtain high-precision surface subsidence deformation information.Based on the Sentinel-1A satellite ascending and descending SAR image data from March 2019 to March 2021 in the Panzhihua mining area and the Baima mining area,this paper uses the SBAS-InSAR technology to obtain the line-of-sight deformation of the mining area’s ascending and descending orbit,Then,Multi-SBAS technology was used to integrate the data of the ascending and descending orbit to obtain the two-dimensional surface subsidence information of the Panzhihua mining area and the Baima mining area,and then conduct interpolation and prediction research on the acquired twodimensional subsidence time series.The main research results are as follows:(1)The SBAS-InSAR technique was used to process 62 scenes of Sentinel-1A ascending and descending orbit data respectively to obtain the annual average subsidence rate and cumulative subsidence of the ascending and descending orbits at the Panzhihua and Baima mines from March 2019 to March 2021.The settlement characteristics of the study area were analyzed in detail by drawing profile lines and the characteristic point analysis method,combined with the mining data,and the results showed that the settlement of the mine area has a high consistency with the spatial distribution of the working face of the mine area.Finally,the reliability of SBAS monitoring results was proved by calculating the correlation coefficients of ascending and descending orbits.(2)The multi-SBAS technology was used to integrate the results of the LOS direction deformation of the ascending and descending orbits to obtain the two-dimensional time series deformation of the Panzhihua mining area and the Baima mining area.Among them,the vertical maximum subsidence rate of Panzhihua mining area is-51mm/a,and the east-west maximum subsidence rate is-33mm/a;the vertical maximum subsidence rate of Baima mining area is-194mm/a,and the east-west maximum subsidence rate is-80mm/a.On this basis,the twodimensional evolution characteristics of the four feature points were quantitatively analyzed,the results indicate that the surface is shifting to the east while vertical subsidence is occurring on the surface as a whole.The stability region was selected as the deformation reference region of the study region,and the reliability of the Multi-SBAS results was verified by calculating the average subsidence rates and standard deviations in the vertical and east-west directions of the reference region.(3)To avoid the influence of missing time series values on the prediction accuracy in the process of time series prediction of surface subsidence,the Prophet model is proposed to be applied to the InSAR two-dimensional surface subsidence time series interpolation research.The simulation data and the actual monitoring data are used for interpolation research,and different random missing ratios and continuous missing lengths are designed to conduct experiments,and use cubic spline,linear,Lagrangian and Prophet interpolation to compare.The experimental results show that under different missing conditions,the Prophet model has better interpolation accuracy and better stability than the traditional interpolation method,and the accuracy advantage of this method becomes more obvious with the increase of the continuous missing length.Finally the method is used to interpolate the temporally missing ephemerides within the study time span.(4)The interpolated two-dimensional subsidence time series is the sample data for the prediction study,and for the subsidence time series deformation characteristics,the empirical wavelet transform(EWT)combined with the Prophet model is proposed to build a combined EWT-Prophet prediction method for surface subsidence time series prediction.EWT has the characteristics of multi-scale analysis,which can effectively weaken the influence of the Prophet model on the weak prediction ability of nonlinear data,and the combination method has high adaptive ability.The retrospective prediction study was conducted by simulating data and prediction sample data,and different prediction lengths were designed for experiments.The experimental results show that the method in this paper is overall better than the traditional method and the single Prophet method,and can obtain higher accuracy prediction results,which can provide some meaningful reference for early warning of subsidence hazards in mining areas.