Some Questions on CMAQ Ozone Modeling (Comparisons of CMAQ v5.3 and CAMx v6.5 runs)

We run CMAQ v5.3 and CAMx v6.5 with 36-12-4km nested domain system. The inputs including emissions and meteorology for the both models are the same. The model runs from CMAQ and CAMx for coarse domains (36km and 12km) are very similar and comparable based on model performance evaluation using AMET v1.4. However, CMAQ shows very poor performance for daily maximum 8-hour average ozone compared with CAMx. Please see attached monthly statistical plot illustrating the monthly O3_8hrmax from two models for 12-km and 4-km domains. My question is why the modeled ozone with 4-km by CMAQ is so low from July to October compared to CAMx runs or coarse domain results even using exactly the same inputs? I get confused and have no idea what wrong with my CMAQ runs from July to October because the model performance should not be so difference between coarse and finer domains, 12-km and 4-km according to my experience. It is highly appreciated if you can provide some comments and suggestions to help me to figure out.
Thank you very much.

I am not sure if this explains all the differences you are seeing, but I noticed that the observations are not the same between the CAMx and CMAQ runs. You said the emissions and meteorology are the same, but if so, why are the obs different? Are the domains and/or time periods different?

Hi Chris Nolte,
That is a good question. We have the same observations for the both models performance evaluation for 4-km domain for sure. I need to clarify for 12-km domain that we evaluate CAMx performance for 12-km using the same monitors within 4-km domain, however we evaluate CMAQ performance using all observational monitors in 12-km domain. We should have done the both models performance for 12-km using the same observations. Our target domain actually is 4-km domain. But the point is why CMAQ performs so poor compared with CAMx from July to October.
Thank you very much

Hi Chris Nolte,
Please see attached for the same observations compared with model results of O3_hrmax.

Is this the release version of the model? Are you using any non-default build options (e.g., chemical mechanism)?
Are you using BEIS for biogenic emissions?
Have you looked at any other species (NOx or CO?)

Hi Chris,
Thanks for your review and response. We use the default mechanism (cb6r3_ae7_aq). I turned off the in-line biogenic emissions because we used MEGAN v3 to estimate biogenic emissions stand alone and then merged by SMOKE.
I looked at NOx and CO for 4-km domain in April (the best model performance for O3-8hrMax) and July (the worst). CMAQ results show overestimated NOx during night time in July (see attached). I am still struggling to find the reason.
Thank you very much.

Hi fliu,

 Just a few questions.  Are your CAMx 12-km and 4-km runs in two-way nesting setting?  Also, I am wondering if you have done CMAQ 4-km run using the same input files but with older version of CMAQ and if yes, how did it look?  Thanks.


Hi Winston,
Thanks for your response and questions.

  1. Yes, Our CAMx 12-km and 4-km runs are in two-way nesting setting.
  2. No, I did not run older version of CMAQ using the same inputs. We developed all inputs for the CMAQ v5.3.

Hi Fliu,

 I have one more question.  Are the lateral boundary condition files of your 4-km CMAQ run coming from the 12-km run?  Do they look o.k. for the months of July to October?  Thanks.


Hi Winston,
Thanks for response again.
Yes, the lateral boundary condition files of 4-km domain are generated from CCTM_CONC files of 12-km using run_bcon.csh . They look fine.


I have a problem with model outputs. There is a lot of difference between the model outputs and observation for O3 and CO. Could you help me, please?
Thank you

@esm, that question is too vague and open-ended for Forum users to be able to help. Also, it it not clear that it pertains to the thread you are posting in.

If yours is a new problem, please post a new thread, and provide more details as to what is your problem and what assistance you are seeking. What do “online” and “offline” mean? Are your model outputs for CO identically zero?


Thank you for your reply.

Offline = first run WRF and after run CMAQ
Online = use WRF-CMAQ two-way model

I use these models for Iran-Tehran. December 2017.

No, CO is very small.

Thank you.