Abstract: A large-scale automatic surface monitoring method for highways built in soft soil areas based on the InSAR technique is proposed to overcome the deficiencies of labor-consuming and unavailability of large-scale deformation by the traditional monitoring method for ground subsidence of highways. Considering the deformation characteristics of soft soils, a nonlinear viscoelastic-plastic four-component combined rheological model is introduced into the InSAR deformation modeling. Then the physical soft soil highway subsidence model is proposed and the time-series InSAR functions are established. The parameters for subsidence are estimated to calculate the results of the large-scale surface deformation areas of highways. The simulated and field tests on a segment of Lungui Road in Guangdong Province are carried out. Compared to that of the traditional linear model, the modeling accuracy of the proposed method has an improvement of 24%. The RMSE for Lungui Road is estimated as ±5.6 mm, with a relative accuracy of 5% and a good consistency with its deformation tendency. A case study of a highway near Dongting Lake in Yueyang, Hunan Province is carried out to verify the capacity, and the 1.5-year time-series settlement results are obtained, with the subsidence rate following a fast-to-slow nonlinear capacity. The results show that the subsidence near Dongting Lake is significantly higher than that in the inland area, with the maximum subsidence accumulated to 46 mm. The proposed method may provide a reference for early subsidence detection and maintenance management of soft soil highways.