We use the small baseline subset (SBAS) method and Sentinel-1B synthetic aperture radar data to obtain the cumulative surface deformation at highly coherent points and different times (12-d intervals from September 10, 2017 to June 1, 2018) following the sixth nuclear explosion conducted by the Democratic People’s Republic of Korea (DPRK). The study is conducted for a 17 km ×22 km area centered on the explosion. Measurement points are aggregated into 14 sets according to their spatial neighborhood. According to the average coherence of each point in the set, the point deformation is weighted and averaged to obtain the cumulative deformation of each set. The location of each set is also weighted and averaged according to the average coherence of each point in standing for all points in the set. The analysis and discussion in this paper are based on these 14 sets for 14 different regions. Results show that the deformation process in the thermal radiation aftereffect stage of the sixth nuclear test can be effectively observed using interferometric synthetic aperture radar. There was still surface uplift near the epicenter for ~10 d after the explosion, after which the surface began to sink. The sinking rate and total sinking amount varied by location. Meanwhile, the phenomenon of subsidence slowing or even the surface uplifting possibly due to the freeze–thaw cycle of water in underground rock in winter was observed. After May 24, 2018, deformation began to rise because the government of the DPRK bombed the entrance of the nuclear facilities. We indirectly demonstrate the reasonableness and consistency of the observation results via the coherence of high-resolution optical images, meteorological data, and interferograms for the area of the nuclear explosion. The spatial and temporal distributions of surface deformation and their causes are then modeled. The results of modeling analysis are as follows. (1) In the thermal radiation aftereffect stage of the DPRK’s sixth nuclear explosion, the surrounding rock softened under high temperature and high pressure, and the surrounding metamorphic rock then compressed under the action of its own gravity and began to sink. This time-varying process is fitted by a deformation prediction model based on the Weibull function, and four parameters—namely the acceleration factor of deformation, comprehensive influence factor of deformation, initial surface deformation, and prediction value of maximum deformation—are obtained. The average correlation coefficient of the fitting curve is about 0.97. Model fitting results show that the sinking deformation tended to stop around May 20. (2) A function model was proposed to analyze the genetic mechanism of surface deformation, where the thickness of the layer of metamorphic rock is taken as the independent variable and the maximum deformation is taken as the dependent variable. The thickness of the layer of metamorphic rock was calculated for explosion burial depths of 450 and 770 m. The vertical impact distance of the explosion from the epicenter was about 1800–2300 m while the deformation coefficient of the metamorphic rock was about 7×10−5–8×10−5. The statistical fitting degree R 2 is about 0.8 and the P value is close to zero. In conclusion, the scope of the impact of the explosion metamorphism is the intrinsic condition of deformation, and gravity is the driving force of surface subsidence in the explosion area. Differences in the deformation rate and amount in different geomorphological locations are due to different thicknesses of the compressible layer relating to the rock thickness and metamorphism. Such differences have directionality, possibly because the attenuation of high-temperature and high-pressure propagation in the direction of tunnel works is less than that in the direction of surrounding rock.