Strange chemical species distribution in output MOZART MOSAIC

[catalinaporaicu]

Hello,
I am trying to run WRF-Chem with MOZART MOSAIC for Reunion Island, off the coast of Madagascar. I am using WRF-Chem V4.1.2 with KPP.
I am using CAMS and CAM-Chem for the initial and boundary conditions, which look okay. However, some species (namely CO) has a very strange distribution in the output, where it looks like a numerical error that occured between the start of the run.
I attached the output of CO at the surface for the initial condition, and then for 5 minutes later. This is occuring in the d02 domain, however this of course impacts the parent domain. The strange distribution after 5 minutes occurs over the entire vertical grid, not just at the surface. This shape shows up in other species too, but CO is the most obvious one with the most distinct shape.

It is as if there is an "overlay" of some sort over the correct values (which is the green in the second picture, corresponding to realistic CO output values). However, the dark blue is very, very low and it has a very strange, jagged distribution with distinct changes at the edges.

I checked my emission input files for the run and they do not contain any NaNs or negative values that might cause this strange numerical error. I already tried to do a test run where I turned off chemical emissions and kept it to chemistry based on the boundary and initial conditions, which still showed the same shape.

What might this be caused by? Please, any help would be greatly appreciated. I am happy to give more information that could help elucidate this problem.

 

[jordanschnell]

Hi, please try setting:

chem_in_opt = 1, 1

Jordan

 

[catalinaporaicu]

Hi, please try setting:

chem_in_opt = 1, 1

Jordan

Hello Jordan,

I keep getting an error that the wrf_chem_input files are not found, which makes sense. How do I overcome this? Thanks

 

[jordanschnell]

Hi, are you getting this error running real.exe?

If so, try running real.exe with chem_in_opt = 0, then run wrfchembc to put the necessary info into the wrfbdy and wrfinput files. Then run wrf.exe with chem_in_opt = 1.

Jordan

 

[catalinaporaicu]

Hi, are you getting this error running real.exe?

If so, try running real.exe with chem_in_opt = 0, then run wrfchembc to put the necessary info into the wrfbdy and wrfinput files. Then run wrf.exe with chem_in_opt = 1.

Jordan

Doing this overcame the error message, however I am still getting the strange distribution in the output for the chemical species when running wrf.exe

 

[jordanschnell]

Hi,

In your original message, you note that the problem occurs with "namely CO"

- Are gas and aerosol species both impacted?
- Are species that do NOT get LBCs from CAM-Chem also erroneous?

Also, please try and adjust these (i.e., change = 1 for d02)

gas_bc_opt = 1,1,0,
gas_ic_opt = 1,1,0,
aer_bc_opt = 1,1,0,
aer_ic_opt = 1,1,0,

 

[catalinaporaicu]

Hi,

In your original message, you note that the problem occurs with "namely CO"

- Are gas and aerosol species both impacted?
- Are species that do NOT get LBCs from CAM-Chem also erroneous?

Also, please try and adjust these (i.e., change = 1 for d02)

gas_bc_opt = 1,1,0,
gas_ic_opt = 1,1,0,
aer_bc_opt = 1,1,0,
aer_ic_opt = 1,1,0,

Hello Jordan,
In the end it was an error with the biogenic input files generated by the megan biogenic preprocessor, there were some negative and nan numbers in the d02 domain which were causing a numerical error.

Thank you for your help!

 

[jordanschnell]

Great to hear!

 

[mona]

Hello,
I am trying to run WRF-Chem with MOZART MOSAIC for Reunion Island, off the coast of Madagascar. I am using WRF-Chem V4.1.2 with KPP.
I am using CAMS and CAM-Chem for the initial and boundary conditions, which look okay. However, some species (namely CO) has a very strange distribution in the output, where it looks like a numerical error that occured between the start of the run.
I attached the output of CO at the surface for the initial condition, and then for 5 minutes later. This is occuring in the d02 domain, however this of course impacts the parent domain. The strange distribution after 5 minutes occurs over the entire vertical grid, not just at the surface. This shape shows up in other species too, but CO is the most obvious one with the most distinct shape.

It is as if there is an "overlay" of some sort over the correct values (which is the green in the second picture, corresponding to realistic CO output values). However, the dark blue is very, very low and it has a very strange, jagged distribution with distinct changes at the edges.

I checked my emission input files for the run and they do not contain any NaNs or negative values that might cause this strange numerical error. I already tried to do a test run where I turned off chemical emissions and kept it to chemistry based on the boundary and initial conditions, which still showed the same shape.

What might this be caused by? Please, any help would be greatly appreciated. I am happy to give more information that could help elucidate this problem.

Hi @catalinaporaicu
I am trying to run WRF-Chem, and I want to use cams-global-reanalysis-eac4 data from Copernicus Climate Data Store to create initial and boundary conditions using MOZBC utility. I tried to follow How to run the WRF-Chem model using CAMS data as initial and boundary conditions (BC)? - Copernicus Knowledge Base - ECMWF Confluence Wiki, but I encountered some problems in providing PS data. I can't see any PS data (which contains hyam and hybm variables) in cams-global-reanalysis-eac4. where Can I get this data?
How did you create yours?
Could you please guide me?
I would appreciate your help.
Thanks in advance.

 

[catalinaporaicu]

Hi @catalinaporaicu
I am trying to run WRF-Chem, and I want to use cams-global-reanalysis-eac4 data from Copernicus Climate Data Store to create initial and boundary conditions using MOZBC utility. I tried to follow How to run the WRF-Chem model using CAMS data as initial and boundary conditions (BC)? - Copernicus Knowledge Base - ECMWF Confluence Wiki, but I encountered some problems in providing PS data. I can't see any PS data (which contains hyam and hybm variables) in cams-global-reanalysis-eac4. where Can I get this data?
How did you create yours?
Could you please guide me?
I would appreciate your help.
Thanks in advance.

Yes, of course, Mona.
When using CAMS, you have to download a few separate things.

1. Firstly, this is the 60 levels with hyam and hybm information: L60 model level definitions - User Documentation - ECMWF Confluence Wiki
2. Then, when you download the reanalysis data from CAMS, in the variable selection in the "download data" section, you need to select surface pressure from the single level drop-down menu.
3. Then you would of course select the multi level chemical variables you plan to use as initial and boundary conditions.
4. Make sure you download the data as model level (and select all from 0 to 60), not pressure level, as this corresponds to the 60 levels in the file in step 1.

I think this will be sufficient in obtaining the data to use mozbc with CAMS!

 

[mona]

Yes, of course, Mona.
When using CAMS, you have to download a few separate things.

1. Firstly, this is the 60 levels with hyam and hybm information: L60 model level definitions - User Documentation - ECMWF Confluence Wiki
2. Then, when you download the reanalysis data from CAMS, in the variable selection in the "download data" section, you need to select surface pressure from the single level drop-down menu.
3. Then you would of course select the multi level chemical variables you plan to use as initial and boundary conditions.
4. Make sure you download the data as model level (and select all from 0 to 60), not pressure level, as this corresponds to the 60 levels in the file in step 1.

I think this will be sufficient in obtaining the data to use mozbc with CAMS!

Hi, great! Thank you.