Potential Errors in CMAQ NO:NO2 Ratios and Upper Tropospheric NO2 Impacting the Interpretation of TROPOMI Retrievals

Abstract

Although Chemical Transport Models (CTMs) such as the Community Multiscale Air Quality Model (CMAQ) have been used in linking observations of trace gases to emissions and developing vertical column distributions, there remain consistent biases between CTM simulations and satellite retrievals. Simulated tropospheric NO₂ vertical column densities (VCDs) are generally higher over areas with large NO _x sources when compared with retrievals, while an opposite bias is found over low NO _x regions. Artificial (i.e., numerical) dilution in the model, where emissions are mathematically dispersed uniformly within the originating CTM grid, can impact modeled NO:NO₂ ratios, while lower CTM VCD levels often observed over rural areas can be attributed to missing emission sources of NO _x or flawed horizontal/vertical transport. Potential causes of both low and high biases are assessed in this study using CMAQ and Tropospheric Monitoring Instrument (TROPOMI) NO₂ retrievals. It was found that more detailed modeling of NO _x plumes to assess the NO:NO₂ ratio in two power plant plumes can mitigate the effect of artificial computational dilution, reducing the bias and overall differences in the observed vs modeled plumes (errors reduced by 30%). Adjustments of upper tropospheric NO₂ led to overall improvements, with a reduction in CMAQ bias (-43% to -29%) and improved spatial correlation (0.81 to 0.86). This study highlights the importance of having accurate modeled NO:NO₂ ratios when comparing models to retrievals and the impact of unintentional numerical dilution.

Keywords: NO2 vertical column densities; NOx ratio; TROPOMI; chemical transport models; power plants; satellite retrievals.

1

CMAQ modeling domain. Plant Bowen and Plant Scherer (both large electricity generation units) in northern Georgia (USA), with associated adjusted grids are shown, as are interstate highways. ATL is the Atlanta Hartsfield–Jackson International Airport.

2

Fraction of the total NO _x that is NO and NO₂ plotted at various distances from the plume source. a) NO and NO₂ ratios calculated using the Gaussian plume and PSS approximation model. b) NO and NO₂ ratios near Plant Scherer (Figure ) as calculated from the CMAQ simulations.

3

a) Changes in model grid volume as dimensions in x and y are adjusted. b) Shows for three different emission rates how NO₂ increases steadily as the grid sizes are adjusted to what is shown in (a).

4

a–c) TROPOMI NO₂ VCD plots. d–f) CMAQ-TROPOMI NO₂ VCD difference plots. g–i) Respective density scatter plots from scenarios in figures d to f. (a,d,g) Represent the base case with no adjustments for potential NO:NO₂ model differences in NO _x elevated plumes at two EGUs (Plant Bowen and Plant Scherer) and in the upper troposphere. (b,e,h) Show adjustments of NO _x bias in elevated plumes at the two EGUs. (c,f,i) Show adjustments of NO _x bias in elevated plumes at both EGUs and at higher altitudes above 8 km. Black lines are Interstate highways. The red and black dots in (g–i) represent the EGUs and the surrounding 8 grid cells as depicted in Figure . Data from the plots are averaged over the 17 selected days in August 2019.