A mining-induced collapse is often characterized by large deformation gradient, spatial discontinuity, and temporal nonlinearity, resulting in the loss of interferometric SAR (InSAR) coherence and consequently subsidence information in areas of steep deformation gradients. In this study, we present different SAR deformation monitoring schemes to map the mining-induced subsidence and collapse. First, SAR data with different wavelengths, including C-band ERS-1, C-band Envisat ASAR, and X-band TerraSAR-X data, are used to highlight three mining subsidence stages and their temporal evolutions over Datong mine (China) in the past 20 years. Mining-induced subsidence over a region can be delimited from InSAR deformation maps, where InSAR coherence is lost over the area of peak subsidence. Second, in order to monitor the large-gradient surface deformation caused by underground mining activities, three SAR deformation monitoring schemes are proposed, including a full-resolution interferogram method, a “remove-restore” phase unwrapping method, and a fusion of phase and offset measurements. Then, taking the Datong coalfield as an example, we demonstrate the capabilities of these methods on mapping large-gradient deformation. Finally, we have found that over 80% of coalfields have deformed during the past 20 years. We conclude that the fusion of the InSAR phase and offset measurements can provide a reliable estimate of large-gradient mining-induced deformation.