Abstract: Soil physicochemical properties serve as crucial driving factors for carbon storage. As a vital ecological barrier zone in China's southwestern border region, the Sichuan-Yunnan ecological barrier area has received limited research attention regarding the specific influence of soil physicochemical properties on carbon storage dynamics. This study integrates the InVEST model with the Optimal Parameters Geographical Detector (OPGD) and Multiscale Geographically Weighted Regression (MGWR) model, drawing on multi-source datasets—including land use/land cover data at 30 m resolution, soil properties at 1 km resolution, and climate variables—to systematically examine the spatiotemporal evolution patterns of carbon storage and the spatially heterogeneous driving factors in the Sichuan-Yunnan ecological barrier area from 2000 to 2020.The results reveal that: (1) Over the study period, total carbon storage exhibited an overall declining trend, primarily attributable to rapid urban expansion and intensive agricultural development in lowland river valley regions. These anthropogenic activities drove widespread conversion of woodland and grassland into construction land and cropland, resulting in substantial losses in aboveground and belowground biomass carbon pools, as well as accelerated soil organic carbon release due to tillage disturbance and erosion. In contrast, localized increases in carbon storage were observed in certain areas, largely benefiting from woodland expansion and large-scale ecological restoration projects such as the Grain for Green Program (returning farmland to forest), which enhanced vegetation cover, litter input, and soil organic matter accumulation. (2) Normalized Difference Vegetation Index (NDVI), soil organic matter content (OC), and elevation (DEM) emerged as the dominant driving factors, each accounting for explanatory power exceeding 50%. All multi-factor interactions were of the enhancement type, with the strongest synergistic effect observed between NDVI and OC (NDVI ∩ OC). This finding underscores the dominant role of the coupled ecology–terrain–soil nutrient mechanism in shaping the spatial distribution of carbon storage across the region. (3) The MGWR model successfully captured significant spatial heterogeneity in the effects of driving factors, confirming that soil physicochemical properties play a pivotal role in influencing both the distribution patterns and transfer processes of carbon storage, particularly in areas with strong topographic gradients. These findings provide practical guidance for future land-use optimization, ecological governance, and carbon sink management in the Sichuan-Yunnan ecological barrier area. By highlighting the central importance of soil properties, vegetation–soil interactions, and spatial heterogeneity, the study offers a scientific foundation and innovative multi-model analytical framework for sustainable carbon storage management in similar ecological barrier regions. Ultimately, the results contribute to climate change mitigation efforts, support enhanced regional ecological conservation, and promote ecosystem resilience under ongoing human and climatic pressures.