You are here:
Evaluating Long-Term Ozone and PM2.5 Simulations Over the United States
Citation:
Hogrefe, C., K. Foley, Keith Appel, S. Roselle, D. Schwede, J. Bash, AND R. Mathur. Evaluating Long-Term Ozone and PM2.5 Simulations Over the United States. ITM 2019 - 37th International Technical Meeting on Air Pollution Modelling and its Application, Hamburg, Hamburg, GERMANY, September 23 - 27, 2019.
Impact/Purpose:
By comparing observed and CMAQ-simulated long-term ozone variations and trends, the work in this study addresses the question of how well CMAQ performs when applied to simulate the effects of changing emissions and meteorological variability on ambient ozone. The study examines two sets of simulations that differ in two of the key input fields, i.e. anthropogenic emissions and boundary conditions. Results of this analysis can inform model development efforts by focusing such efforts on processes and input fields most directly connected to the model’s ability in reproducing observed variability and trends
Description:
Performing decadal regional air quality model simulations has become more feasible in recent years due to increased computational resources. Analyzing such long-term simulations can help to build confidence in the modeling systems’ use for air quality planning applications by evaluating its ability to represent changes in air quality resulting from changes and variations in emissions and meteorology. In this study, we analyze ozone and total and speciated PM2.5 fields from two sets of WRF/CMAQ simulations over the continental U.S. The simulations were performed for 1990 – 2010 and 2002 – 2014, respectively, and differed in terms of horizontal grid spacing, emissions, and chemical boundary conditions. Our analysis includes a comparison of these input files to investigate to which extent they may have caused differences in the simulated pollutant levels, variations, and trends. Results show that trends in ozone boundary conditions affect trends in modeled surface ozone concentrations. Moreover, observed summer ozone trends often are more negative than modeled trends and modeled interannual variability in daily maximum 8-hr ozone tends to be smaller than observed. For total PM2.5, modelled and observed seasonal cycles were found to differ in parts of the country, and assumptions about the relative abundance of primary vs. secondary organic aerosols as well as their volatility were found to be a key driver of these differences. Modeled PM2.5 concentration trends were found to be sensitive to the processing of primary PM2.5 emissions, especially primary organic carbon. Disagreements between observed and modeled organic carbon trends occurred across all percentiles, likely resulting both from assumptions in emissions processing and the treatment of organic aerosol processes in these WRF/CMAQ simulations. The results will be discussed in the context of designing and interpreting dynamic model evaluation studies and their implications for using model output fields for applications such as health impact studies.