An official website of the United States government.

This is not the current EPA website. To navigate to the current EPA website, please go to www.epa.gov. This website is historical material reflecting the EPA website as it existed on January 19, 2021. This website is no longer updated and links to external websites and some internal pages may not work. More information »

Community Multiscale Air Quality Modeling System (CMAQ)

Incremental Evaluation of CMAQv5.2

Over the last two decades, new versions of the CMAQ model have been periodically released. Each new version consists of multiple updates to the model’s scientific algorithms and input data quality. Since CMAQ is used by the USEPA, as well as State and Regional offices, to estimate criteria pollutant (e.g. ozone, O3HelpO3Ozone (O3) is a colorless gas with a pungent odor. It is found in two layers of the atmosphere, the stratosphere and the troposphere. In the stratosphere, ozone provides a protective layer shielding the Earth from ultraviolet radiation's potentially harmful health effects. At ground level (the troposphere), ozone is a pollutant that affects human health and the environment, and contributes to the formation of smog.; fine particulate matter, PM2.5) concentrations, each new release requires a comprehensive evaluation to establish model credibility for a wide range of applications. The CMAQ modeling system version 5.2 (v5.2) was tested and evaluated against observations and was publicly released in June 2017.

Major Updates to the CMAQv5.2 Modeling System

  • Transition from WRFv3.7 to WRFv3.8
  • WRFv3.8 with lightning assimilation
    • Assimilation of lightning strike data from the National Lightning Detection Network (NLDN)
    • Force deep convection where lightning is observed; only allow shallow convection where it is not
  • Science updates in CMAQv5.2
    • Increased O3 deposition to wetted surfaces (particularly important in the early morning hours)
    • Updated organic aerosol (OA) with semivolatile primary OA and empirical secondary OA formation from combustion sources
      • Large decreases in primary OA in both summer (75-95%) and winter (55-90%)
      • In summer, increase in OA due to the increased contribution from secondary OA formation pathways
    • Updated windblown dust scheme
    • Updated sub-grid cloud scheme
    • Revised the O3 dry deposition (M3Dry update) to correct an error in an update that was implemented in CMAQv5.1
  • Update from the 2011 NEI v2 eh to 2011 NEI v2 ek platform
    • Updated from MOVES2014 to MOVES2014a for mobile emissions (updates for TX, OH, NJ, CA, AZ)
    • Revised spatial surrogates for underway emissions from commercial marine
  • Update from the CB05e51 to CB6r3 chemical mechanism
    • Generally lower O3 and NO2 from lower production of alkyl nitrates (winter), lack of NOx recycling, lower HO2 and higher heterogeneous removal of nitrates
  • Transition from GEOS-Chem to hemispheric CMAQ for lateral boundary conditions

CMAQ Simulation Details

Figure 1 shows the CMAQ simulation configuration for the CMAQv5.1 base simulation and the CMAQv5.2 final simulation. Both simulations cover the contiguous U.S. (CONUS), using the same version of MCIP, OMI data, and in-line photolysis scheme. Indicated in bold are those aspects of the modeling system that differ in the CMAQv5.2 simulation. These include updates to the WRF meteorology, emissions, lightning NO emissions, AERO version, chemical mechanism version and boundary conditions (BCs).

Summary table of the configuration options for the CMAQv5.1 and CMAQv5.2 simulations.Figure 1. Summary of the configuration options for the CMAQv5.1 and CMAQv5.2 simulations.

Ozone

Figure 2 shows the change in ground-level O3 for July 2011 from the various updates in the CMAQv5.2 modeling system. The top left plot in the figure shows the CMAQv5.1 modeled O3 mixing ratio, while the top right plot in the figure shows the change in O3 between CMAQv5.1 and CMAQv5.2. The remaining plots in the figure show the change in O3 from the individual updates to the modeling system. These updates are:

  • Transition to WRFv3.8 (WRFv3.8)
  • Lightning assimilation in WRFv3.8 (Lightning Assimilation)
  • Science updates in CMAQv5.2 (Updates in CMAQv5.2)
  • Update from NEIv2 eh to NEIv2 ek platform (Emissions Update)
  • Transition to the CB6 chemical mechanism (CB6r3 Chemical Mechanism)
  • Update to the lightning NO calculation (Lightning NO Calculation Update)
  • Transition from GEOS-Chem to Hemispheric CMAQ for Boundary Conditions
  • Update to the O3 dry deposition (M3Dry update)

Maps showing impact on July O3 mixing ratio (ppbv) from the various modeling system updates applied in the CMAQv5.2 simulation.Figure 2. Impact on July O3 mixing ratio (ppbv) from the various modeling system updates applied in the CMAQv5.2 simulation.

Overall, there is a large decrease in O3 in CMAQv5.2 compared to CMAQv5.1, with monthly average decreases exceeding 5 ppb in a large portion of the eastern U.S. All the updates shown contribute to the decrease in O3 at least in some part, with the largest decreases in O3 coming from the science updates in CMAQv5.2 and the transition to the CB6r3 chemical mechanism. The transition to WRF3.8 and the lightning assimilation in WRF also lower O3 mixing ratios in many areas in July. The emissions update had the largest effect on O3 over the Gulf of Mexico, while updates to the lightning NO calculation had the largest impact in the western U.S. The transition from GEOS-Chem to Hemispheric CMAQ as lateral boundary conditions results in a large reduction in O3 in July, whereas the update to the O3 dry deposition results in a widespread increase in O3 in July.

Diurnal time series of O3 mixing ratio (ppbv) for the various modeling system updates. Concentration is shown on top and bias is shown on the bottom. Figure 3. Diurnal time series of O3 mixing ratio (ppbv) for the various modeling system updates. Concentration is shown on top and bias is shown on the bottom. Figure 3 shows the monthly average observed and model simulation diurnal O3 mixing ratio for the simulations show in Figure 2. The observed/simulated O3 mixing ratio is shown on the top Figure 4, while the model bias is shown on the bottom. The CMAQv5.1 simulation (red) has the largest bias, with O3 overestimated throughout the entire day. The transition to WRFv3.8 and the lightning assimilation (shown in blue as a single update) decrease O3 throughout the entire day. The emissions update (green) primarily decreases O3 during the overnight hours, while the transition to the CB6r3 chemical mechanism (yellow) has the greatest decrease on O3 during the middle of the day. The lightning NO update (orange) has a small impact onO3, while the transition from GEOS-Chem to Hemispheric CMAQ as boundary conditions results in a large decrease in O3.

Change in MDA8 O3 mixing ratio bias (ppbv) between the CMAQv5.1 and v5.2 simulations. Cool colors indicate improved bias in the v5.2 simulation, while warm colors indicate worse bias in the v5.2 simulation.Figure 4. Change in MDA8 O3 mixing ratio bias (ppbv) between the CMAQv5.1 and v5.2 simulations. Cool colors indicate improved bias in the v5.2 simulation while warm colors indicate worse bias in the v5.2 simulation.


Figure 4 shows the change in daily maximum 8-hr average O3 (MDA8) bias between CMAQv5.1 and CMAQv5.2. Warm colors indicate higher bias in the CMAQv5.2 simulation, while cool colors indicate lower bias in the CMAQv5.2 simulation. With the exception of the San Joaquin Valley (SJV) in California, MDA8 O3 bias decreases with CMAQv5.2 for most sites. The largest decreases occur along the Gulf Coast in Louisiana and Texas. Overall, about 80% of the sites show a decrease in MDA8 O3 bias between the CMAQv5.1 and CMAQv5.2 simulations.

PM2.5

Figure 5 shows the change in monthly average PM2.5 between CMAQv5.1 and CMAQv5.2 for January 2011, along with the change in PM2.5 due to the various updates to the CMAQv5.2 modeling system. The upper left plot in the figure shows the CMAQv5.1 simulated PM2.5 concentration, while the upper left plot shows the change in monthly average PM2.5 between CMAQv5.1 and v5.2. The remaining plots in the figure show the change in PM2.5 from the individual updates to the modeling system as listed above.

Maps showing impact on January simulated PM2.5 concentrations (µgm-3) from the various modeling system updates applied in the CMAQv5.2 simulation.Figure 5. Impact on January simulated PM2.5 concentrations (µgm-3) from the various modeling system updates applied in the CMAQv5.2 simulation.


Diurnal time series of January PM2.5 concentrations (µgm-3) and bias for the various modeling system updates. Concentration is shown on top while bias is shown on the bottom.Figure 6. Diurnal time series of January PM2.5 concentrations (µgm-3) and bias for the various modeling system updates. Concentration is shown on top while bias is shown on the bottom.

There is a mixed single between increases and decreases in PM2.5 in January, with the science updates in CMAQv5.2 primarily responsible for the changes in PM2.5 concentration. Lightning assimilation resulted in a small increase in PM2.5, as did the emissions update. The transition to hemispheric CMAQ for BCs had a small impact in CONUS, but did impact areas in Canada and the Gulf of Mexico near the simulation boundaries.

Change in January PM2.5 bias (µgm-3) between the CMAQv5.1 and v5.2 simulations. Cool colors indicate improved bias in the v5.2 simulation while warm colors indicate worse bias in the v5.2 simulation.Figure 7. Change in January PM2.5 bias (µgm-3) between the CMAQv5.1 and v5.2 simulations. Cool colors indicate improved bias in the v5.2 simulation while warm colors indicate worse bias in the v5.2 simulation.

Figure 6 shows the change in diurnal PM2.5 concentrations for January. The science updates in CMAQv5.2 clearly had the biggest impact on PM2.5, greatly reducing the nighttime overestimation of PM2.5 and decreasing the small daytime underestimation in PM2.5. Stacked bar plots of speciated PM2.5 for observed (left), CMAQv5.1 (middle) and CMAQv5.2 (right) for January. The height of the bars indicated the total observed or model simulated PM2.5.Figure 8. Stacked bar plots of speciated PM2.5 for observed (left), CMAQv5.1 (middle) and CMAQv5.2 (right) for January. The height of the bars indicated the total observed or model simulated PM2.5.The other updates to the modeling system had minimal overall impact on the PM2.5 concentrations in January. Figure 7 shows the difference in bias between CMAQv5.1 and v5.2, with warm colors indicating higher bias in CMAQv5.2 and cool colors indicating lower bias. Many of the sites (~62%) show a decrease in bias in the CMAQv5.2 modeling system versus CMAQv5.1, with the largest decreases occurring in the Northeast corridor, the Great Lakes region and along the Gulf of Mexico. Figure 8 shows stacked bar plots of observed and modeled PM2.5 speciation for January. Decreases in PM2.5 with the CMAQv5.2 modeling system were primarily due to decreases in organic carbon (OC) and non-carbon organic matter (NCOM).

Figure 9 shows the change in monthly average PM2.5 between CMAQv5.1 and v5.2 for July 2011, which the CMAQv5.1 simulation PM2.5 concentration in the upper left and the difference between v5.1 and v5.2 in the upper right (v5.2 – v5.1). PM2.5 concentrations are largely higher in July in the CMAQv5.2 modeling system, with the increase primarily due to science updates in CMAQv5.2, followed by the transition to Hemispheric CMAQ BCs and updates to the WRF meteorology. 

Maps showing impact on July simulated PM2.5 concentrations (µgm-3) from the various modeling system updates applied for CMAQv5.2.Figure 9. Impact on July simulated PM2.5 concentrations (µgm-3) from the various modeling system updates applied for CMAQv5.2.

Diurnal time series of July PM2.5 concentrations (µgm-3) and bias for the various modeling system updates. Concentration is shown on top while bias is shown on the bottom.Figure 10. Diurnal time series of July PM2.5 concentrations (µgm-3) and bias for the various modeling system updates. Concentration is shown on top while bias is shown on the bottom.Figure 10 shows the diurnal PM2.5 concentrations from the various simulations, indicating that almost all the changes to the modeling system improved PM2.5 performance, with the science updates clearly driving the biggest improvement, especially during the daytime hours. Figure 11 shows the change in bias between CMAQv5.1 and v5.2, with the vast majority of sites (over 80%) indicating a decrease in bias, particularly in the eastern U.S. Figure 12 shows stacked bar plots of observed and modeled speciated PM2.5, indicating that increases in organic carbon (OC) and non-carbon organic matter (NCOM) are driving the increase in PM2.5 in CMAQv5.2.

Change in July PM2.5 bias (µgm-3) between the CMAQv5.1 and v5.2 simulations. Cool colors indicate improved bias in the v5.2 simulation while warm colors indicate worse bias in the v5.2 simulation.Figure 11. Change in July PM2.5 bias (µgm-3) between the CMAQv5.1 and v5.2 simulations. Cool colors indicate improved bias in the v5.2 simulation while warm colors indicate worse bias in the v5.2 simulation.Stacked bar plots of speciated PM2.5 for observed (left), CMAQv5.1 (middle) and CMAQv5.2 (right) for July. The height of the bars indicated the total observed or model simulated PM2.5 (µgm-3).Figure 12. Stacked bar plots of speciated PM2.5 for observed (left), CMAQv5.1 (middle) and CMAQv5.2 (right) for July. The height of the bars indicated the total observed or model simulated PM2.5 (µgm-3).

Wind-blown Dust

The wind-blown dust (WBD) model in CMAQv5.2 has been updated to eliminate large over-estimations of WBD that can occur with the WBD model in CMAQv5.1. Figure 13 shows results from a CMAQv5.1 simulation using the updated WBD model available in CMAQv5.2 for March 2011. On the left are model simulated concentrations of soil, PM2.5 and PM10 from a CMAQv5.1 simulation with no WBD, while on the right is the contribution to each of those species from the new WBD model. The largest concentrations of soil, PM2.5 and PM10 occur in the western U.S., particularly the desert southwest where concentrations of WBD would be expected to be the greatest.

Maps showing the impact of the updated wind-blown dust (WBD) model implemented in CMAQv5.2, tested in a version of CMAQv5.1Figure 13. Impact of the updated wind-blown dust (WBD) model implemented in CMAQv5.2, tested in a version of CMAQv5.1. The top panels show the model simulated Soil without WBD (left) and then the contribution to soil from WBD on the right. The same thing is shown for PM2.5 in the middle panels and PM10 on the bottom panels. All plots are in units of µgm-3.

References

Appel, K., Napelenok, S., Hogrefe, C., Foley, K.M., Pouliot, G., Murphy, B., Luecken, D., Heath, N. (2016). Evaluation of the Community Multiscale Air Quality (CMAQ) Model Version 5.2. 2016 CMAS Conference, Chapel Hill, NC. Presentation available from the CMAS conference website. Exit