A fine subsidence information extraction model based on multi-source inversion by integrating InSAR and leveling data

Subsidence information is of paramount significance for the restoration and treatment of geological destruction in mining areas. However, the traditional interferometric synthetic aperture radar (InSAR) technique is incapable of detecting mining subsidence with large gradient deformation. To overcome this limitation, this study proposes a fine subsidence information extraction model to improve the subsidence detection ability. The model consists of the data integration, multi-source parameter inversion, and subsidence extraction method. The integration of small baseline subset InSAR (SBAS-InSAR) and leveling data allows the parameters to be accurately inversed and predict the subsidence funnel based on the probability integration method (PIM). By fusing the subsidence funnel and SBAS-InSAR data based on geographically weighted regression, the fine subsidence information is extracted. Model validation with a certain working face of coal mine in China shows that the proposed model is significantly efficient in detecting large deformation in the goaf center, as well as acquiring more detailed information for goaf boundary areas, indicating its superiority compared with SBAS-InSAR and PIM. The proposed model is not only a low-cost and practical way of parameter acquisition, but also a promising method for obtaining more accurate and reliable subsidence information, and is expected to provide decision support for geological disasters in mining areas.