error running aerosol-aware thompson microphysics scheme

[psagar]

I am trying to run the aerosol-aware version of thompson microphysics scheme with mp_physics = 28 option. I am following all the instructions and have correctly used all namelist settings and aerosol data required. The metgrid.exe ran well and I can see the aerosol variables from QNWFA_QNIFA_QNBCA_SIGMA_MONTHLY.dat being ingested in the met_em* files. However, when I run real.exe, the model fails -- the real.exe program just freezes without any errors after displaying the following rsl logs. I am using v4.4 for both WRF and WPS.
I have used enough nodes so it should not be due to memory problem. Also, I have previously used another aerosol-aware version (mp_physics = 40) successfully. The real.exe/wrf.exe works fine with the earlier mp_physics = 8 option also. I have three domains with grid size ~ 9, 3 and 1km. I tried to run only in one domain (d01) but the problem remains. Please let me know if you have any suggestions. I have attached my namelist.input file for your reference.


..........................
rsl.error.0000 message before the model fails:

Yes, this special data is acceptable to use: OUTPUT FROM METGRID V4.4
Input data is acceptable to use: met_em.d01.2023-02-10_00:00:00.nc
metgrid input_wrf.F first_date_input = 2023-02-10_00:00:00
metgrid input_wrf.F first_date_nml = 2023-02-10_00:00:00
Timing for input 51 s.
flag_soil_layers read from met_em file is 1
Max map factor in domain 1 = 1.10. Scale the dt in the model accordingly.
Using sfcprs2 to compute psfc

 

[haiqingsong]

I am trying to run the aerosol-aware version of thompson microphysics scheme with mp_physics = 28 option. I am following all the instructions and have correctly used all namelist settings and aerosol data required. The metgrid.exe ran well and I can see the aerosol variables from QNWFA_QNIFA_QNBCA_SIGMA_MONTHLY.dat being ingested in the met_em* files. However, when I run real.exe, the model fails -- the real.exe program just freezes without any errors after displaying the following rsl logs. I am using v4.4 for both WRF and WPS.
I have used enough nodes so it should not be due to memory problem. Also, I have previously used another aerosol-aware version (mp_physics = 40) successfully. The real.exe/wrf.exe works fine with the earlier mp_physics = 8 option also. I have three domains with grid size ~ 9, 3 and 1km. I tried to run only in one domain (d01) but the problem remains. Please let me know if you have any suggestions. I have attached my namelist.input file for your reference.


..........................
rsl.error.0000 message before the model fails:

Yes, this special data is acceptable to use: OUTPUT FROM METGRID V4.4
Input data is acceptable to use: met_em.d01.2023-02-10_00:00:00.nc
metgrid input_wrf.F first_date_input = 2023-02-10_00:00:00
metgrid input_wrf.F first_date_nml = 2023-02-10_00:00:00
Timing for input 51 s.
flag_soil_layers read from met_em file is 1
Max map factor in domain 1 = 1.10. Scale the dt in the model accordingly.
Using sfcprs2 to compute psfc

You will need to add the following line in the &metgrid section of the namelist.wps file:
constants_name = 'QNWFA_QNIFA_SIGMA_MONTHLY.dat'
To use the data in program real.exe and model wrf.exe, add these namelists:
&domains
wif_input_opt = 1,
num_wif_levels = 30,

&physics
mp_physics = 28, 28, 28,
use_aero_icbc = .true.

 

[psagar]

Hi Haiqingsong, thank you for your reply. I have added all those settings correctly both in namelist.wps and namelist.input files. As I mentioned, I can see the aerosol variables in my met_em* files so that means that the WPS processed the input data QNWFA* correctly. I have the file QNWFA_QNIFA_QNBCA_SIGMA_MONTHLY.dat (with _QNBCA_ in the file name), so I put the name accordingly.

 

[Ming Chen]

There are 2 issues here:

(1) Please specify the options below in your namelist.input,

&physics
mp_physics = 28
use_aero_icbc = .true.
aer_opt = 3

&domains
wif_input_opt = 1,
num_wif_levels = 30,

(2) Please use the the data "QNWFA_QNIFA_QNBCA_SIGMA_MONTHLY.dat" in namelist.wps, i.e.,

&metgrid
constants_name = 'QNWFA_QNIFA_QNBCA_SIGMA_MONTHLY.dat',

Let me know if you still have trouble to run the case. Thanks.

 

[psagar]

Hi Ming, thank you for your response. The only thing I am missing from your suggestion is that I didn't specify aer_opt = 3. This is some new information to me because it was not mentioned in the Thomspon aerosol-aware documentation here:
WRF Model Users Site. But now I found it was mentioned in WRF user guide. Thank you for this.

Our supercomputer is down currently for maintenance, I will try adding aer_opt = 3 option and see if that fixes my problem of running real.exe. I don't have problem running WPS -- I can see the aerosol variables such as I_WIF_JUN etc. in my met_em* files.

 

[psagar]

Hi Ming, I tried adding aer_opt = 3 option but it didn't change anything. The real.exe program still fails after showing the above message. I attached my namelist.input for your reference.

 

[psagar]

Hi Ming, I figured out my problem. I was using an older version of METGRID.TBL rather than the one that came with the WRFv4.4 so it caused some conflict. I see that there were several changes between the two versions of METGRID.TBL (e.g. the use of _CL flag). Anyway, the real.exe program runs well for me now. Sorry for the trouble. I found that these changes have been documented here already for anyone's reference:

Removing legacy code and cleaning up logic for real processing of aer… · wrf-model/WRF@61a1359

…osol for mp_physics=28 (#1567) TYPE: enhancement KEYWORDS: real, Thompson aerosol-aware, microphysics SOURCE: Timothy W. Juliano and Pedro A. Jimenez (NCAR/RAL) DESCRIPTION OF CHANGES: ...

github.com

 

[Ming Chen]

Hi Sagar,

Many thanks for the update. I am glad it finally works. The troubles caused by inconsistency between different versions are really annoying.