(180ac) Clustering US Metro Areas By per-Capita Mobile Greenhouse Gas Emissions: Targeted Mitigation Strategies

We present a novel typology of 349 metropolitan statistical areas (MSAs) in the United States and demonstrate its implication for understanding and mitigating per-capita mobile greenhouse gas emissions (PMGE).
Via factor analysis of 55 indicators across 7 categories of road network, buildings, socioeconomics, car ownership, commute mode share, transit energy, and climate, we identified eight key drivers of mobile emissions: Auto + Scale, Transit Propensity, Density, Sustainable Transit, Warm Weather, Remote + Active, Network Connectivity, Network Sprawl.
Based on these, we clustered the metro areas via Gaussian mixture modeling to obtain seven metro types, namely Auto Large, Auto Small, Dense Active, Heavyweight Eco-Transit, Heavyweight Transit, Sparse Active, and Sparse Auto. We used an Extreme Gradient Boosting regression model to predict PMGE based on all the indicators, excluding population. By using SHapley Additive exPlanations (SHAP) values, we identified the most significant indicators for all MSAs and for each cluster type.
The model reveals that density and transit mitigate PMGE, while car use and climate exacerbate it.
The typology provides further insights into how these effects vary across network, mode share and development patterns.
Ultimately, the typology can serve as a framework to inform and initialize type-specific mitigation strategies.