WRF-Chem hung: optical averaging mie fatal error DMIESS: 34

[limillet]

Dear WRF-Chem support team & users,

I ran WRF-Chem for a three-domain system with coarse (18 km), intermediate (6 km) & fine (2 km) domains. The wrf executable hung when the fine domain was introduced with the error message shown below:

-------------- FATAL CALLED ---------------
FATAL CALLED FROM FILE: <stdin> LINE: 76
optical averaging mie fatal error DMIESS: 34
------------------------------------------------

I found from module_optical_averaging.F that this error will be reported when the angle between the directions of the incident and the scattered radiation become greater than 90.0 degrees. However, according to the comments in module_optical_averaging.F, the supplementary value for the angle is supposed to be used when it greater than 90.0 degrees:

! THETD(J): ANGLE IN DEGREES BETWEEN THE DIRECTIONS OF THE INCIDENT
! AND THE SCATTERED RADIATION. THETD(J) IS< OR= 90.0
! IF THETD(J) SHOULD HAPPEN TO BE GREATER THAN 90.0, ENTER WITH
! SUPPLEMENTARY VALUE, SEE COMMENTS BELOW ON ELTRMX;

Any guidance about this issue is appreciated.
Thank you very much for your attention and time!

 

[douglowe]

Please can you post your namelist, and tell us what you are trying to study in the 2km domain?

Also, you should try running your model without aerosol radiative feedbacks switched on (set both aer_ra_feedback and aer_op_opt = 0), and look at your aerosol fields, to determine what black carbon mass loadings you have on your 2km domain - probably it's very high values of these which are causing the problem.

 

[limillet]

Please can you post your namelist, and tell us what you are trying to study in the 2km domain?

Also, you should try running your model without aerosol radiative feedbacks switched on (set both aer_ra_feedback and aer_op_opt = 0), and look at your aerosol fields, to determine what black carbon mass loadings you have on your 2km domain - probably it's very high values of these which are causing the problem.

Hi, Douglowe, Thank you very much for your attention to this matter.
The namelist.input has been attached below. I want to study the influence of sulfate aerosols on the deep convection in TCs.
Then, I will check aerosol fields when the simulation without aerosol radiative feedbacks is complete.

 

[limillet]

Please can you post your namelist, and tell us what you are trying to study in the 2km domain?

Also, you should try running your model without aerosol radiative feedbacks switched on (set both aer_ra_feedback and aer_op_opt = 0), and look at your aerosol fields, to determine what black carbon mass loadings you have on your 2km domain - probably it's very high values of these which are causing the problem.

Please refer to the black carbon mass concentration below:

 

[limillet]

On the other hand, there is a new finding on the original run when aerosol radiative feedbacks were still on.

Null value was found at the edge of domain 2 exclude the area of domain 3 (the dark blue area).
I just wondering if this is the cause of the model crash. May I have some suggestions to improve?

 

[douglowe]

Hi limillet,

Apologies for the delay in replying.

Your black carbon concentrations in domain 3 look okay. Your sulphate plot for domain 2 is from the start of your simulation, and you have feedback (from inner domains to outer domains) switched on - perhaps you haven't initialised the chemical fields in your outer domains? I've found that mozbc has to be run seperately for each model domain (so setting domain=3 will only process domain 3, it wont process domains 1 & 2 as well), so maybe you have the same problem? You should check the values in your wrfinput_d* files after running mozbc, to make sure that they are sensible (and consistent).

Regarding your namelist - I would recommend switching off feedbacks from the inner to outer domains for the moment. I think that your time_step might be a little long (rough guide is dx(in km)*6 - which would be 108 seconds for your outer domain) - this might not cause issues for your outer domains, but could cause issues for the inner domain. You could try reducing the time_step value, or increasing the parent_time_step_ratio value to compensate for this. Also, your radt value is strange (27?) - you should try to keep this (and other timestep settings) to either 0, or a clean integer multiplier of your time_step value (so maybe 30, similar to your photdt value, would be wise).

cheers,
Doug