Using near-road observations of CO, NOy, and CO2 to investigate emissions from vehicles: Evidence for an impact of ambient temperature and specific humidity

Abstract

Abstract Vehicles are a significant source of carbon monoxide (CO) and nitrogen oxides (NOx), two harmful pollutants and precursors to ozone formation. Previous studies have shown that emissions of NOx in the US EPA's National Emissions Inventory (NEI) are overestimated relative to observations in the summer and possibly for an annual average. Here we use measurements of CO, NOx, carbon dioxide (CO2), and meteorological variables collected at a near-road (NR) site along I-95 in Howard County, Maryland, during the cold months of 2016 and 2017, to infer ΔCO/ΔNOx, ΔCO2/ΔNOx, and ΔCO2/ΔCO emission ratios from vehicular running exhaust and their sensitivity to temperature and specific humidity. We also use aircraft observations of CO, NOx, and meteorological variables collected during the 2011 summertime DISCOVER-AQ campaign over the Baltimore-Washington region to analyze the impact of temperature and humidity on ΔCO/ΔNOx ratios, which integrate anthropogenic and biogenic sources in the urban area. Overall, we find a strong, statistically significant increase of 113% in ΔCO/ΔNOx and of 112% in ΔCO2/ΔNOx from −5 to 25 °C at the I-95 NR site, indicating a decrease of approximately 50% in emissions of NOx as air warms, linked primarily to diesel-powered trucks. Temperature sensitivity of pollution control equipment on diesel vehicles may contribute to this trend. Results are robust when using several different techniques for calculating emission ratios. The sensitivity of vehicular emissions of NOx to specific humidity is much weaker and cannot solely explain the trend with temperature. The aircraft data show a similar increase of 114% in ΔCO/ΔNOx from 25 °C to 34 °C, with a weaker sensitivity to specific humidity. In comparison to the NR observations, ΔCO/ΔNOx output from the MOtor Vehicle Emission Simulator (MOVES) with default settings, used to simulate mobile emissions for air quality models and in the NEI, showed a smaller increase for ΔCO/ΔNOx of 41% over the temperature range −5 to 25 °C. The increase in ΔCO/ΔNOx from MOVES is due to an increase in emissions of CO by 23% and a decrease in emissions of NOx by 11% over −5 to 25 °C, which is less than the observed decrease in NOx. Our study suggests that the overestimate in emissions of NOx in the NEI previously reported using summertime observations may be corrected in part by accounting for the temperature sensitivity of mobile NOx running emissions within MOVES. Future work will focus on improving MOVES by adjusting parameters controlling the impact of temperature and humidity on emissions to better represent the behavior of real-world vehicular emissions.