The impact of satellite-adjusted NO_x emissions on simulated NO_x and O₃ discrepancies in the urban and outflow areas of the Pacific and Lower Middle US

Abstract. We analyze the simulation results from a CMAQ model and GOME-2 NO₂ retrievals over the United States for August 2009 to estimate the model-simulated biases of NO_x concentrations over six geological regions (Pacific Coast = PC, Rocky Mountains = RM, Lower Middle = LM, Upper Middle = UM, Southeast = SE, Northeast = NE). By comparing GOME-2 NO₂ columns to corresponding CMAQ NO₂ columns, we produced satellite-adjusted NO_x emission ("GOME2009") and compared baseline emission ("BASE2009") CMAQ simulations with GOME2009 CMAQ runs. We found that the latter exhibited decreases of −5.6%, −12.3%, −21.3%, and −15.9 % over the PC, RM, LM, and SE regions, respectively, and increases of +2.3% and +10.0% over the UM and NE regions. In addition, we found that changes in NO_x emissions generally mitigate discrepancies between the surface NO_x concentrations of baseline CMAQ and those of AQS at EPA AQS stations (mean bias of +19.8% to −13.7% over PC, −13.8% to −36.7% over RM, +149.7% to −1.8% over LM, +22.5% to −7.8% over UM, +31.3% to −7.9% over SE, and +11.6% to +0.7% over NE). The relatively high simulated NO_x biases from baseline CMAQ over LM (+149.7%) are likely the results of over-predictions of simulated NO_x emissions, which could shed light on those from global/regional Chemical Transport Models.

We also perform more detailed investigations on surface NO_x and O₃ concentrations in two urban and outflow areas, PC (e.g., Los Angeles, South Pasadena, Anaheim, La Habra and Riverside) and LM (e.g., Houston, Beaumont and Sulphur). From two case studies, we found that the GOME2009 emissions decreased surface NO_x concentrations significantly in the urban areas of PC (up to 30 ppbv) and in those of LM (up to 10 ppbv) during the daytime and that simulated NO_x concentrations from CMAQ with GOME2009 compare well to those of in-situ AQS observations. A significant reduction in NO_x concentrations resulted in a comparable increase in surface O₃ concentrations in the urban areas of PC (up to 30 ppbv) and the resulting simulated O₃ concentrations compare well with in-situ surface O₃ observations over South Pasadena, Anaheim, and Riverside. Over Houston, Beaumont, and Sulphur, large reductions in NO_x emissions from CMAQ with GOME2009 coincides with large reduced concentrations of simulated NO_x. These concentrations are similar to those of the EPA AQS NO_x observations. However, the resulting simulated increase in surface O₃ at the urban stations in Houston and Sulphur exacerbated preexisting high O₃ over-predictions of the baseline CMAQ. This study implies that simulated low O₃ biases in the urban areas of PC are likely caused by simulated high NO_x biases, but high O₃ biases in the urban areas of LM cannot be explained by simulated high NO_x biases over the region. This study also suggests that both in-situ surface NO_x and O₃ observations should be used simultaneously to resolve issues pertaining to simulated high/low O₃ bias and that remote-sensing data could be used as a constraint for bottom-up emissions. In addition, we also found that daytime O₃ reductions over the outflow regions of LM following large reductions in NO_x emissions in the urban areas are significantly larger than they are over outflow regions of PC. These findings provide policymakers in the two regions with information critical to establishing strategies for mitigating air pollution.