Advancing high-resolution U.S. onroad fossil fuel carbon dioxide emissions estimates

Governments at multiple scales have been enacting policies to reduce onroad fossil fuel CO₂ emissions, one of the largest and rapidly increasing emitting sectors. Onroad GHG emissions at fine spatial resolution are critical to reduction policies. To answer this need in the U.S., two high-resolution onroad FFCO₂ emissions datasets, Vulcan and DARTE, have been constructed. Here, we contrast the methods and numerical results of these datasets to offer insights into methodological improvements and sensitivity to assumptions and methodological choices. While the total U.S. relative difference varies from +1.3% to +5.1%, larger differences emerge at fine scales. For example, urban and rural emission mean relative differences (MRD) are −37.7% and −7.8%, respectively. Apart from differences in input datasets, discrepancies stem from three methodological elements. The first is the choice of road basemap and classification schema. DARTE was found to have classified non-motorable roads as motorable, leading to FFCO₂ spatial pattern differences. Second is the procedure by which county-scale emissions are downscaled to individual roadways. For example, DARTE uses spatially invariant downscaling in contrast to Vulcan's use of traffic volume variations. The third is the method to estimate inter-annual onroad FFCO₂ emissions variation. The DARTE variation is driven by county-scale data in contrast to coarser state-scale Vulcan input data. A comparison to independently observed onroad FFCO₂ emissions from a sample of monitoring sites shows an MRD of −14.5% for Vulcan versus an MRD of −58.8% for DARTE. These differences and methodological insights lead to recommended best practices for advancing onroad GHG emissions estimation at fine scales.