Errors related to points exceeded cfl=2 and gasodesolver_lsodes failure in running WRF-Chem using CBMZ-MOSAIC chemical mechanisms

[Ankan]

Dear all,
I am performing WRF-Chem v3.9.1 simulations using the CBMZ-MOSAIC chemistry option (chem_opt=32). My parent domain simulation was completed successfully. But, after using ndown to generate boundary conditions for the nested domain files, I gave a run for the nested domain, and the simulation was stopped after throwing some errors in the rsl.error.0022 and rsl.out.0022 files. The errors are related to 'points exceeded cfl=2' and 'gasodesolver_lsodes failure' (the rsl.error.0022 file is attached for your convenience). I want to mention that the horizontal resolution of my domains is 15 km and 3 km (i.e., 1:5 parent_grid_ratio), so I set time_step 4*dx, i.e., time_step=60 for the parent domain and ndown run, and time_step=12 for the nested domain run. As the cfl errors are coming, I thought reducing the time step could solve the problem and gave it some test runs. The results of the test runs using varying time_steps are as follows:
Test1: time_step = 12 (4*dx)
"BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES" came in the wrfrun.log and simulation stopped after throwing some errors related to points exceeded cfl=2 and gasodesolver_lsodes failure
Test2-4: time-steps of 18 seconds (6*dx), 15 seconds (5*dx), 9 seconds (3*dx)
Simulation was stopped but no BAD TERMINATION error came. However, the errors related to points exceeded cfl=2 and gasodesolver_lsodes failure came. All the rsl.error.0022 files and namelist.input files are attached for your convenience. I found an old thread gasodesolver_lsodes failure no related to similar errors using CBMZ-MOSAIC. So, my question is that is it related CBMZ-MOSAIC? or is it related time_step or other namelist settings? Can anyone please help me in this regrad to solve this issue? That will be highly appreciated. Thank you for your time and consideration.
With regards,
Ankan

 

[zhaneden]

Hello Ankan, I had similar issue though i ran chem_opt=203, mine was also terminated by cfl errors. But i tried my setting with chem off and it works. Can i ask if you solved your problem and how you solved it? Thanks. Zhan

 

[Ankan]

Hi Zhan,
It was a long time ago. However, what I recall is that I changed the time step to 45 seconds instead of 60 seconds, and I got successful runs. In general practice, whenever someone is getting CFL errors, they should first reduce the time step and see which setup is giving the successful runs. If it is not, then there may be other issues in your namelist as CFL errors can come from various reasons. So, my suggestion is that if it is not being solved by changing the time step, then consult the experts of the WRF-Chem model. They will help you in this regards.
Best wishes,
Ankan

 

[zhaneden]

Hi Zhan,
It was a long time ago. However, what I recall is that I changed the time step to 45 seconds instead of 60 seconds, and I got successful runs. In general practice, whenever someone is getting CFL errors, they should first reduce the time step and see which setup is giving the successful runs. If it is not, then there may be other issues in your namelist as CFL errors can come from various reasons. So, my suggestion is that if it is not being solved by changing the time step, then consult the experts of the WRF-Chem model. They will help you in this regards.
Best wishes,
Ankan

Hello Ankan, thank you for your reply. The model appears to run when using a really really small timestep (9s). I only tested it for a short period. But, since I also have multiple high-resolution child domains (<=1km), completing even a 1-hour simulation at this timestep would be very computationally expensive. I am planning to use other chem_opt like 202, probably return to 203 later and run more tests because I really want to make it work. Thanks. Zhan