Ubiquitous Data-Driven Framework for Traffic Emission Estimation and Policy Evaluation

Quantifying urban traffic emission dynamics is critical for evaluating decarbonization policies, yet existing models often lack resolution, timeliness, or scalability. Ubiquitous urban data offer new opportunities for large-scale, fine-grained, and near-real-time emission estimation. Here we present a data-driven framework that integrates traffic camera footage with mobile phone data to estimate citywide vehicular emissions. Applied to Manhattan, New York, our method captures substantial spatiotemporal variation in emissions across hours, days, and road segments. Omitting fine-grained inputs, such as traffic signals, speed variation, or fleet heterogeneity, introduces average uncertainties of -49% to +25% in emission estimates. We further evaluate the early impacts of Manhattan’s congestion pricing program, finding that eight weeks after the implementation, traffic volumes declined by 10%, resulting in a 16–22% drop in emissions. Our approach enables timely, high-resolution policy assessment using widely available urban big data, offering a practical and transferable tool for supporting climate action.