hide
Free keywords:
climate change, typologies, causality, machine learning, spatially explicit, urban form, car travel
Abstract:
Urban transport is a major source of greenhouse gas emissions, making effective urban planning crucial for climate mitigation. Global typologies of cities can help to scale planning strategies, yet they hardly capture how interventions translate into local contexts. Big urban data, combined with artificial intelligence, holds great potential to facilitate scalable yet location-specific planning to reduce urban travel and related emissions. However, current research falls short in recognizing underlying variable dependencies, understanding neighborhood-specific differences, and comparing relationships across world regions. Here, we present a systematic quantitative analysis of how urban form influences car-based work trip distance and related emissions in six cities on three continents. We integrate causal discovery and explainable machine learning and apply it to a sample of 10 million mobility data points derived from GPS and call detail records. We find significant direct dependencies between urban form and car-based work trip distance, neglected in previous research. Across cities, geographic access to city center and employment matters more than density or connectivity, yet the magnitude of such effects and their spatial distribution vary depending on a city’s size and centrality. Compact central development appears most effective, while our approach identifies suburban corridors up to 40 km from the center where densification offers additional mitigation potential. In more polycentric cities, subcenter development provides further leverage. Our results contribute high-resolution comparative evidence to refine urban typologies for effective emission reduction in the context of car-based work travel.
