To examine regional carbon-stock dynamics within the broader context of global climate change and carbon-cycle processes, the PLUS (patch-generating land use simulation) model was coupled with the InVEST (integrated valuation of ecosystem services and trade-offs) model to quantify the spatiotemporal evolution of carbon storage in Weifang, a representative city within the Shandong Peninsula urban agglomeration, from 2010 to 2022. Carbon-storage patterns for 2030 were simulated under four scenarios—baseline development, cropland-prioritized development, urban-development-prioritized growth, and green-oriented development. The drivers shaping carbon-storage variation were further identified using XGBOOST (extreme gradient boosting) model integrated with SHAP (shapley additive explanations). Carbon storage in Weifang exhibited a persistent decline between 2010 and 2022, decreasing from 16.56×10⁶ t in 2010 to 15.98×10⁶ t in 2020. This reduction primarily resulted from the loss of cropland and the expansion of impervious surfaces. Among the simulated scenarios, the cropland-prioritized scenario produced the highest carbon-storage total in 2030, underscoring both the carbon-sink capacity of cropland and the critical importance of cropland-protection strategies. The green-oriented scenario performed slightly better than the baseline due to enhanced carbon sequestration associated with forest expansion, whereas the urban-development-prioritized scenario yielded the lowest carbon storage. Temperature and slope emerged as the dominant contributors to carbon-storage variation, while GDP and population density exerted negative influences, indicating that economic growth and urban expansion may further intensify the decline in carbon storage. Overall, the findings reveal complex interactions between anthropogenic activities and natural ecosystems within urban-agglomeration contexts and provide a robust scientific basis for formulating land-use planning and carbon-mitigation strategies.