Low Ozone and PM2.5 concentrations with MOZART-MOSAIC 202 option

[struggling_phd]

Dear community,

I am Leo, and I am currently a PhD student at University of Houston and I am a part of the environmental fluid mechanics lab (..:: Environmental Fluid Mechanics Lab at UH ::..). Currently I am working on a project with WRF-CHEM application within Urban Area (Houston).

After the initial struggle, I was able to successfully run multiple WRF-CHEM simulations over the domain I am investigating. The meteorological outputs are very good, telling me that the model is configured well. (I've checked winds, temperature and am on my way to check relative humidity.) However, the problem is with ozone and PM2.5 results.
I am using WRF/CHEM 4.2.2.

Please take a look at the screenshot I've attached. I cannot reason out why are some months so under predicted by the model regarding the ozone, and then again why are some months so over-predicting for pm 2.5.

For the anthropogenic emissions, I am using the anthro_emiss utility with edgar_htap_2010_0.1x0.1 data,

For bioemiss it's megan

for BC I use mozbc with the attached .inp file.

I am also attaching my namelist to run the sim, as well as the input file for mozbc, and some time-series of ozone, pm25 and CO.

Any kind of tips / suggestions would be appreciated!
Thanks!

 

[jordanschnell]

Hi,

I have a few suggestions -

1) Where did you obtain the namelist for mozbc? It seems it is incomplete (i.e., nh4 is the only aerosol considered)
2) I would suggest turning off the chemical LBCs to see if that is your issue.
3) If not, then you'll likely need to turn processes off one at a time until you see your issue - this is especially true for aerosol processes (e.g, dust, sea salt) but could also be impacting gases (e.g., chem_conv_tr, aer_ra_feedback.
- For aerosols, is there a particular species that is elevated?
4) Is there a reason you've turned off cldchem_onoff and conv_tr_aqchem?
5) You could try halving your emissions to see if the ozone responds - that will likely be harder to diagnose what the problem is.

Jordan