Extremely low NO2 concentration

Hi teachers,

I used the 2020 baseline scenario of the DPEC emission inventory (which follows the same format as the MEIC inventory for future scenarios), ERA5 meteorological data, WRF version 3.8.1, and CMAQ version 5.4 to simulate ozone (O₃) and nitrogen dioxide (NO₂) concentrations over the Yangtze River Delta region for January 2020.

However, the simulation results are quite unsatisfactory, especially for NO₂, which appears to be nearly zero across most of the domain. I have attached the run scripts and log files for your reference. For now, I have only included anthropogenic emissions, and all other sources (e.g., biogenic, light) have been turned off.

I suspect that the problem may be related to the emission files. For example, HONO—which is assigned a large portion of the NOₓ emissions—shows extremely low values, and in most time steps and regions, the values are zero.

Could this be the reason for the unrealistically low NO₂ concentrations?

run_cctm_Bench_2016_12SE1.csh (38.1 KB)
CTM_LOG_174.v54_intel_case1_20200102.txt (4.2 MB)
PA_REPORT.20200102.txt (115.2 KB)
PA_REPORT.20200103.txt (115.2 KB)
(This log file was artificially interrupted when CMAQ was just run)

emission file (max and min for NO NO2 HONO):



extreme low NO2 result (unit is ppb):

Comparison between obs and simu of O3 :

The upper range of the emissions shown in your ncview plots seems reasonable, but why are the cells all blue? Do you have cells with emissions toward the upper end of the range? Toggle the “linear” button to “low” – does the spatial distribution of NO, NO2, and HONO emissions look right?

The overall distribution appears to cover the target region, but as shown in the figure, there are very few grid cells with values close to the maximum. Most of the grid cells have very small emission values. Could you please confirm if this is expected?



I recently switched to the SAPRC chemical mechanism in my CMAQ simulations, and the results have significantly improved. The NO₂ concentration curves are now smoother, and the expected peaks during the morning and evening rush hours (associated with traffic emissions) are well captured, with no more anomalous continuous zero values.

However, I’ve noticed that NO₂ concentrations over the North China Plain and Northeast China are still substantially underestimated compared to observations. Additionally, I also observed abnormally high ozone concentrations in some areas, which seem inconsistent with the typical wintertime photochemical environment.

Could these discrepancies be caused by inaccurate WRF physics parameterization? For example, is it possible that heavy snowfall or cold surface conditions in northern China are not properly handled in the WRF settings, thus artificially suppressing NO₂ concentrations or overestimating ozone?

5-days NO2 average (unit: μg/m³)

5-days O3 average (unit: μg/m³)

Your NO2 emissions look ok to me.

The chemical mechanism you use in CMAQ must be consistent with the speciation used in your emissions files. Does your log file indicate many emissions species are not found, or that there are species in your emissions inputs that are not being used?

Why do you say your O3 is too high? 80 ug/m3 is about 40 ppb, which is close to background. Your urban ozone is lower than background due to NO titration.

Hi cgnolte,

I am using the <saprc07tc_ae6> chemical mechanism in my CMAQ simulation, and the emission files were downloaded from the official MEIC website. They are specifically prepared for the SAPRC07 mechanism and have been processed through temporal, spatial, and speciation allocation. I believe the mechanism setup between MEIC and my CMAQ configuration should be consistent.

I have verified in the CMAQ_Control_DESID_saprc07tc_ae6_aq.nml file that all required emission species are indeed present in the input. The missing species reported by CMAQ are shown in the screenshot below (Fig1&2), and I’ve also attached the full CTM_LOG output for your review. Additionally, I’m including my run script — I would appreciate it if you could help check whether there is anything inappropriate in the setup.In this simulation, I did not enable photolysis, lightning NOx, or soil NOx emissions. Could this be a reason for the underestimation of NO₂, especially in northern and northeastern China, given that these regions are also characterized by high anthropogenic NOₓ emissions? (Fig4).

Regarding ozone concentration: I noticed that the background values are quite high. However, this simulation is for January (winter), and O₃ levels are expected to be low under typical conditions. The results I obtained, however, resemble a summer scenario with unusually high O₃. I have also attached the hourly trends of NO₂, NO, and O₃ (Fig3) at a single grid cell for one day as a reference.

Looking forward to your insights and suggestions. Thank you in advance!

CCTM run script:
run_cctm_Bench_2016_12SE1.csh (38.2 KB)

CCTM log file:
CTM_LOG_000.v54_intel_case1_20200108.txt (538.2 KB)

Fig1&2


Fig3:

Fig4: