Probabilistic Assessment of Carbon Storage Response to Reduction Planning for Sustainable Development in Shanghai: A Coupled Bidirectional CA and MC-InVEST Approach

China’s megacities are increasingly adopting land reduction planning, yet most land cover simulations remain expansion-oriented and InVEST assessments often overlook carbon density uncertainty. To address these limitations, this study developed an integrated framework coupling a bidirectional cellular automata (CA) model with a Monte Carlo-based InVEST (MC-InVEST) approach to evaluate the probabilistic response of Shanghai’s terrestrial carbon storage to simultaneous construction land expansion and reduction. Based on 2000 Monte Carlo iterations, Shanghai’s 2024 carbon storage was estimated at 60.00 Mt (95% UI: 51.09–69.84 Mt). By 2035, mean carbon storage declined to 53.95 Mt (95% UI: 44.07–65.15 Mt) under the Urban Development Priority (UDP) scenario, 56.73 Mt (95% UI: 47.90–66.44 Mt) under the Farmland Protection Priority (FPP) scenario, and 59.16 Mt (95% UI: 49.98–69.11 Mt) under the Ecological Protection Priority (EPP) scenario, respectively. Paired Monte Carlo comparisons showed that EPP retained 5.21 Mt (95% UI: 3.53–6.82 Mt) more carbon than UDP and 2.43 Mt (95% UI: 1.73–3.01 Mt) more than FPP, with both differences remaining positive across all 2000 paired iterations. However, restoration maturity sensitivity analysis substantially reduced the EPP–FPP difference. These results show that the carbon storage consequences of area-balanced construction land expansion and reduction depend on both the land cover types converted and the subsequent restoration pathways. The proposed framework provides an uncertainty-aware basis for evaluating land-sector carbon storage under alternative urban spatial planning strategies.