Multidimensional urban morphology and carbon intensity: Nonlinear drivers in dense cities

Area of Science:

• Urban morphology
• Environmental science
• Machine learning

Background:

• Urban morphology significantly impacts carbon emissions, but fine-scale 3D structure and natural substrate mechanisms are poorly understood.
• Existing 2D metrics inadequately capture complex urban form drivers of carbon emissions.

Purpose of the Study:

• To investigate fine-scale urban morphology mechanisms influencing carbon emissions.
• To identify key drivers and quantify their impact on carbon emissions in a high-density metropolis.

Main Methods:

• Utilized explainable machine learning (XGBoost-SHAP) for analysis.
• Analyzed 708 functionally distinct urban blocks in Wuhan using refined block delineation.
• Incorporated 3D structure, natural geographic elements, and human activity intensity.

Main Results:

• Natural geographic elements, 3D morphology, and human activity intensity are stronger drivers than 2D metrics.
• Identified geographic constraints (slope) and their interaction with floor area ratio (FAR) as key synergistic risks.
• Discovered nonlinear thresholds for FAR and sky view factor (SVF), and functional heterogeneity impacts.

Conclusions:

• Low-carbon planning requires multidimensional management beyond traditional 2D metrics.
• Incorporating geographic constraints, nonlinear thresholds, and functional zoning is crucial for effective urban planning.
• Urban morphology's impact on carbon emissions is complex and context-dependent.