【Objective】 Soil organic carbon (SOC) forms the basis of soil fertility, food production, and soil health, and plays a key role in climate change via mediating greenhouse gas emissions. Consequently, accurate characterization of SOC spatiotemporal dynamics is extremely important for the sustainable management of soil resources, ecosystem stability maintenance, and mitigation and adaptation to climate change. 【Method】 A total of 399, 413, and 407 cropland topsoil (0 ~ 20 cm) SOC data in 1980, 2000, and 2015 were collected from the southern Jiangsu Province of China, respectively, and the microbial-explicit SOC model MIMICS (Microbial-Mineral Carbon Stabilization) was used to model the spatiotemporal dynamics of SOC. The Sobol global sensitivity analysis was applied to identify the sensitive parameters of the MIMICS model, and then, two-parameter optimization schemes, one batch (using all SOC observations in a batch mode to optimize the parameters) and site-by-site (using SOC observations at individual sites to optimize the parameters site by site), were used to optimize the sensitive parameters of the MIMICS model through Markov Chain Monte Carlo (MCMC) approach, respectively. The coefficient of determination(R²), root mean squared error (RMSE), and mean absolute error (MAE) that were calculated from the independent validation of SOC in 2000 and 2015 were used to compare the performance of different parameter optimization schemes.【Result】Results show: (1) The net increment of SOC density between 1980 and 2000 was 0.89 kg·m^-2, while the net decrement was 0.44 kg·m^-2 between 2000 and 2015, representing a net increment of 0.45 kg·m^-2 over the period of 1980-2015; (2) The MIMICS model with parameters optimized by either One batch or site-by-site method can represent the overall trends in topsoil SOC dynamics during the period of 1980-2015, but the model with parameters optimized by the site-by-site method presents more local details on the variability of the SOC change rate; (3) Compared with the default parameter values and the One-batch optimized parameter values, the MIMICS model with site-by-site optimized parameter values had the best performance in modeling the spatiotemporal dynamics of SOC in the study area, with the RMSE decreasing by 22.2% (the independent validation in 2000) and 14.7% (the independent validation in 2015) in comparison with the MIMICS model with default parameter values. Yet, its prediction accuracy in 2015 was still relatively low (R² = 0.13, RMSE = 1.22 kg·m^-2). 【Conclusion】The optimization of sensitive parameters can improve the space-time SOC prediction accuracy of the MIMICS model, and the representation of local details on the spatiotemporal patterns of SOC dynamics. Although the MIMICS model with the spatially heterogeneous parameter values optimized by the site-by-site method had the best performance, its prediction accuracy in 2015 was still relatively low, which indicated that the MIMICS model still has limitations in representing the responses of SOC to anthropogenic activities such as changes in land use and agricultural management practices. Thus, further improvement of the MIMICS model structure and enhancing the spatiotemporal resolution of model input data are still significant challenges for regional scale modeling of SOC spatiotemporal dynamics through microbial-explicit SOC models.