Abstract: Accurately extracting surface deformation is essential for the prevention and control of geological hazards caused by underground coal mining. By taking a working face in Guotun Coal Mine, Shandong Province, as the case study, this paper first obtains 18 Sentinel-1A satellite images during the extraction period of the working face (July 31, 2017, to May 3, 2018), and derives the surface deformation of the goaf area based on SBAS-InSAR technology. Then, driven by InSAR observations, the functional projection relationships for the three-dimensional parameters between the probability integration method (PIM) and line-of-sight (LOS) deformation derived by SBAS-InSAR are deduced, and a three-dimensional and full-parameter inversion model based on genetic algorithm with random error elimination (GAREE) is proposed. Based on this model, the subsidence parameters inside the study area are accurately retrieved with the deviation for each parameter less than 3% compared with the empirical parameters. Finally, by using the retrieved parameters, PIM is employed to predict the whole goad deformation with the predicted results highly consistent with the field leveling data. The root mean square errors (RMSE) on observation line A and line F are 0.083 m and 0.102 m, respectively, and the mean absolute errors (MAE) are 0.068 m and 0.089 m, respectively. Results show that the parameter inversion model proposed by this study can effectively obtain the subsidence information for the whole basin of a mining goaf in a low-cost way, providing scientific and significant importance for engineering application and potential disaster predictions in coal mining areas.