How does urban land use and transportation spatial configuration affect carbon exacerbation and mitigation?

Urban emissions remain a critical global challenge, yet the role of spatial planning in mitigation is poorly quantified. This study aims to clarify how urban form and transportation networks influence carbon emissions by introducing a new metric — the Bidirectional Carbon-Emission Discrepancy Coefficient (BCDC) — which captures the difference between ground-based and satellite-observed emissions to reveal spatial impacts. Focusing on the Wuhan Metropolitan Area, we combined a Bayesian-optimized LightGBM model with SHAP analysis to identify nonlinear effects and thresholds. We find that urban compactness improves emissions up to a threshold of 0.23, beyond which benefits diminish. A U-shaped relationship emerges between carbon efficiency and both ecological land share and average travel distance, indicating optimal ranges for mitigation. Spatial analysis shows distinct patterns: emission hotspots need targeted industrial relocation, while coldspots benefit most from expanding blue-green infrastructure (>30 %) and decentralizing transport networks (15–20 % centrality reduction). These results provide actionable, location-specific strategies for urban decarbonization. By linking carbon data with spatial analytics, this framework advances climate-responsive urban planning, offering a scalable toolkit for tailored, evidence-based interventions that move beyond one-size-fits-all approaches.