The National Emissions Inventory Significantly Overestimates NOx Emissions: Analysis of CMAQ and in situ observations from DISCOVER-AQ

NOx and CO not only adversely impact human health, but they, along with associated VOCs, are also important precursors for O3 formation. While ambient NOx and CO concentrations have decreased dramatically over the past 10-20 years, O3 has remained a more recalcitrant problem, particularly in the Baltimore/Washington region. Reduction of O3 production requires that emissions inventories, such as the National Emissions Inventory (NEI), accurately capture total emissions of CO and NOx while also correctly apportioning them among different sectors. Previous evaluations of the NEI paint different pictures of its accuracy, with assertions that it overestimates either one or both of CO and NOx from anywhere between 25 percent to a factor of 2. These conflicting claims warrant further investigation. In this study, measurements of NOx and CO taken aboard the NOAA P3B airplane during the 2011 DISCOVER-AQ field campaign were used to determine the NOx/CO emissions ratio at 6 locations in the Washington/Baltimore region. An average molar emissions ratio of 12.8 × 1.2 CO/NOx was found by calculating the change in CO over the change in NOx from vertical concentration profiles in the planetary boundary layer. Ratios showed little variation with location. Observed values were approximately a factor of 1.35 - 1.75 times greater than that predicted by the annual, countywide emissions ratio from the 2008 NEI. When compared to a temporalized, gridded version of the inventory processed by SMOKE, ratio observations were greater than that predicted by inventories by up to a factor of 2. Comparison of the in situ measurements and remotely sensed observations from MOPITT of CO to the Community Multiscale Air Quality (CMAQ) model agree within 10-35 percent, with the model higher on average. Measurements of NOy by two separate analytical techniques, on the other hand, show that CMAQ consistently and significantly overestimates NOy concentrations. Combined with the CO observations, this indicates that the NEI overestimates NOx emissions by approximately a factor of 2. Comparison of the temporalized NEI to continuous monitoring of NOx emissions from point sources shows that, on average, agreement between observations and the NEI were within 5 percent. In a region where the NEI estimates on-road emissions can account for 50-75 percent of total NOx, the most likely source of error in the NOx inventory is in the on-road sector. Assumptions about the lifetime and efficacy of catalytic converters in the MOVES model should be investigated as a possible source of this error.