CMAQ simulation data is unreasonable

Dear teachers, I used CMAQ5.2 to simulate the ozone concentration data in China from June 1 to 10, 2020. China’s MEIC list was used for emission sources.The simulated ozone data is not reasonable (the spatial distribution and data are not reasonable).
At the same time, I also made the zero emission inventory into the CMAQ model to simulate the ozone emission on June 1, 2020.
The ozone data obtained from the two simulations are not very different, which is something I am more confused about. The zero emission inventory should simulate ozone concentrations close to the background value, but why is this result?

Spatial distribution of ozone: (MEIC inventory on the left, zero emission inventory on the right)

I checked the following:
First, Does CMAQ read emission sources correctly?I looked at the LOG file of the two runs for the emis path and the model should read the corresponding emissions respectively.

Second, I checked the initial and boundary values:The ozone concentration in ICON file is 0.035 ppmv, and the ozone concentration in BCON file is 0.03-0.07 ppmv.

Third, is there a problem with the MEIC list? The following picture shows the spatial distribution of some compounds in EMIS.nc input into CMAQ. From the picture, the spatial distribution of these compounds is reasonable, and I don’t see any problem.

Fourth, compare with observed values: I randomly selected the ozone data of a dozen monitoring stations during the same period and roughly drew scatter plots. As shown in the figure, the ozone number I simulated was concentrated in the range of 65-70ug/m³ (the simulated ozone concentration of all grids was also in this range, which was obviously wrong). In addition, I simulated the ozone concentration data in January, also ranging from 60 to 70ug/m³, with no seasonal variation.

Is there any teacher who can help me look at my problem, which has been bothering me for a long time . Thank you very much for your answers!!!
run_cctm1.csh (24.6 KB)
zeroemis_log.txt (2.6 MB)

Hello Sunchanglei,

could you please provide some additional information?

  • which time period is shown in the ozone maps?
  • which time period is shown in the ozone scatter plot?
  • can you confirm that the first day of the simulation was June 1, 2020, and that the model was initialized with file ICON_v52_china6_profile on that day for both simulations?

At first glance, I don’t see any major mismatches between emission species and mechanism species being reported in the log file you posted, and the magnitude and spatial pattern of the emission maps you posted seem reasonable. That said, if you used a more recent version of CMAQ, you could enable DESID emission diagnostic output files that would then allow you to verify that the emissions being used in the model calculations match those in the input file after applying the DESID emission mapping rules specified in the control namelist file.

1 Like

Hello, teacher. Thank you for your reply.
Regarding the several questions you asked, I will re-describe them as follows:

  1. The ozone map on the left(anthropogenic emissions) shows the average daily ozone concentration from June 1st to 10th, 2020. and the right(zero-emission) shows the ozone concentration on June 1, 2020. That is, the two simulation times are 10 days and 1 day respectively
  2. The ozone scatter plot shows the average daily ozone concentration data for the 10 days from June 1st to 10th in the anthropogenic emission simulation.
  3. I confirm that the first day of the simulation was June 1, 2020. And the model was initialized with file ICON_v52_china6_profile for both simulations.
    Because in my icon script,i set:
    set IC = profile #> either profile or m3conc
    set DATE = date +%Y%j -d $rundate
    set rundate = 20200601 #> only needed for nested runs

Unfortunately, I didn’t use the latest version of CMAQ. Due to various factors, I eventually chose version 5.2. I really have no idea where else problems could arise :sob:

Thanks for sharing these additional details.

If you look at hourly ozone time series for June 1 - June 10 for the anthropogenic emissions case at some of the monitors used for your scatter plot, what do you see? The scatter plot and maps show that the spatial variability is much smaller than expected, but they do not provide any information on temporal variability. In addition to ozone, I suggest also looking at modeled CO, NO, NO2, and PM2.5 maps and time series for the anthropogenic emissions case.

1 Like

Hello, teacher. Sorry for replying to you so long. I have been dealing with this problem during this period of time. But it still couldn’t be solved. Now a new situation has emerged:

  1. I re-ran the entire CMAQ process. Now, the spatial distribution of the average ozone concentration over the ten days from 20200601 to 20200610 that I simulated is like this (left figure), which is completely opposite to the observed values. Meanwhile, compared with other studies (as shown in the right figure, the simulation time of this study was July),the areas where my simulation should have shown high ozone values showed low values instead.


  2. Similarly, I also simulated the situation in January 2020. The spatial distribution and concentration range of ozone are not much different.

3.The following images show the spatial distribution of NO2 from 20200601 to 20200610 (left image) and 202001 (right image),


4.As you mentioned, I only ran the MEIC anthropogenic source list data from 20200601 to 20200610 for ten days, but only extracted ozone and NO2, as shown in the following figure. Similarly, the spatial distribution of ozone remains unreasonable, showing a situation where it is higher in the west and lower in the east.