Urban emissions hotspots: Quantifying vehicle congestion and air pollution using mobile phone GPS data

On-road emissions vary widely on time scales as short as minutes and length scales as short as tens of meters. Detailed data on emissions at these scales are a prerequisite to accurately quantifying ambient pollution concentrations and identifying hotspots of human exposure within urban areas. We construct a highly resolved inventory of hourly fluxes of CO, NO₂, NO_x, PM_2.5 and CO₂ from road vehicles on 280,000 road segments in eastern Massachusetts for the year 2012. Our inventory integrates a large database of hourly vehicle speeds derived from mobile phone and vehicle GPS data with multiple regional datasets of vehicle flows, fleet characteristics, and local meteorology. We quantify the 'excess' emissions from traffic congestion, finding modest congestion enhancement (3-6%) at regional scales, but hundreds of local hotspots with highly elevated annual emissions (up to 75% for individual roadways in key corridors). Congestion-driven reductions in vehicle fuel economy necessitated 'excess' consumption of 113 million gallons of motor fuel, worth ∼ $415M, but this accounted for only 3.5% of the total fuel consumed in Massachusetts, as over 80% of vehicle travel occurs in uncongested conditions. Across our study domain, emissions are highly spatially concentrated, with 70% of pollution originating from only 10% of the roads. The 2011 EPA National Emissions Inventory (NEI) understates our aggregate emissions of NO_x, PM_2.5, and CO₂ by 46%, 38%, and 18%, respectively. However, CO emissions agree within 5% for the two inventories, suggesting that the large biases in NO_x and PM_2.5 emissions arise from differences in estimates of diesel vehicle activity. By providing fine-scale information on local emission hotspots and regional emissions patterns, our inventory framework supports targeted traffic interventions, transparent benchmarking, and improvements in overall urban air quality.