Hello,
I have a problem with the NOx concentration in the CMAQ simulation.
When I simulated the CMAQ, the NOx concentration was significantly underestimated.
Initially, I thought the emission might be the problem, but other teams using the same emission didn’t show this much deviation. Compared to that, ozone concentration is unusually high.
I read a paper describing a CMAQ problem related to vertical dilution, which can exaggerate it to reduce NOx concentration. might it be the case?
CMAQ version is 5.5, compiled with intel Fortran oneAPI.
Thank you.
Hello,
NOx concentrations can be impacted by a variety of different causes (not just vertical mixing) - this may require a bit more digging and research on your part. I would suggest looking at your logs to make sure that emissions are properly mapped and that there aren’t any errors in your run. From there you may want to evaluate modeled meteorological variables that are linked to NOx formation and degradation as well as sink species for your particular domain/ep.
Just curious, what paper are you referencing? What domain/episode are you modeling? and what NOx observations are you using?
-Sara
Thank you for your response. I’m also doubting the emission mapping. I do not have any errors during the compilation and running of CMAQ. How can I find the problem in emission mapping? If you have any suggestions on checking issues in mapping, I would appreciate them.
For the vertical mixing related to NO and NO2, I’m referencing paper below:
Potential Errors in CMAQ NO:NO2 Ratios and Upper Tropospheric NO2 Impacting the Interpretation of TROPOMI Retrievals
Thank you.
If you read through your CTM_LOG files (just pick one), it will tell you if certain emissions streams in your runscript are mapped or not.
@slfarrell is spot on. The first step is to make sure the emission files and variables are being read. CMAQ uses DESID to translate emission variables to concentrations. In the DESID configuration, file-variables are mapped to CMAQ-variable and math is applied (e.g., added, scaled, masked). If you’re concerned that something is being left out, I recommend using the budget tool and/or the emission diagnostic tool to check that the emissions put into the model match the total from the input files.
For example, if you had a BUDGET file the total domain-wide emissions are the 19th field.
More details can be found in the documentation at GitHub - USEPA/CMAQ: Code for U.S. EPA’s Community Multiscale Air Quality Model (CMAQ) for estimating ozone, particulates, toxics, and deposition of acids and nutrients at neighborhood to global scales. · GitHub.
Lastly, NOx performance is not the same everywhere all the time. Model performance in the upper tropospheric paper is not relevant to an urban monitor. Performance in one city may not be relevant to another. So, as @slfarrell said, context about where and when you are modeling would be helpful and it would be good to know what monitor(s) and measurement method you’re comparing to.
You’re saying “NOx” and I assume you’ve included NO and NO2 from the model. Is the observation a NOx measurement or the sum of NO and NO2? It is often useful to compare to a known good NO2 measurement (being careful of techniques that have artifacts).
Thank you for your comments. I’m checking out other primary variables such as hydrocarbons; they might also have problems.
In the log file, NO and NO2 seems linked properly, without any errors or substitutions. I think mapping is not a problem. Because of that, I’m investigating something else.
