The model suddenly crashes

[fei Xiao]

Hi guys. I would like to ask for your assistance. I am running WRF 4.1.5, and WPS 4.1. I am using the FNL data. Real.exe runs fine but there seems to be problem with wrf.exe. And wrf.exe stopped after two minutes without any errors. I have tried different time_steps, but it still stopped. By the way, I also tried ERA-5 data.
The rsl_real.tar.gz is the rsl.error and rsl.out after running real.exe. The rsl_wrf.tar.gz is the rsl.error and rsl.out after running wrf.exe.
I tried looking for answers in the net, but there was no definite answers. Thanks for your help.

 

[fei Xiao]

I also set chem_opt=0 and run real.exe and wrf.exe. The rsl of wrf.exe is as follows:

 

[kwerner]

Hi,
The error when you set chem_opt = 0 is

-------------- FATAL CALLED ---------------
FATAL CALLED FROM FILE:  <stdin>  LINE:     337
ERROR: wet scavenging option requires chem_opt = 8 through 13 or 31 to 36 or 41 to 42 or 109 or 503 or 504 or 601 or 611 to function.
-------------------------------------------

which indicates that you can't set this to 0 with other options you're using.

If going back to the original namelist.input file you sent, can you try to run this with 100 processors, instead of 140. This likely won't make a difference, but it's worth a try to see if the model doesn't like the decomposition.

Otherwise, it would seem this may be a chemistry-specific option. Our team is not experienced with WRF-chem, and unfortunately WRF-chem has discontinued all support services for their version of WRF. However, there is a still a wrf-chem section on this forum and you can try posting there to see if anyone may be able to help you. If you'd like to ensure it's a chemistry problem, you can build WRF without any chemistry turned on, and run this test without any chem options in the namelist to see if it runs. If so, they you'll know it's related to wrf-chem.

 

[fei Xiao]

Thank you for your reply. I have resolved the issue you mentioned.
I was preparing to run WRF-Chem, but before that I had to ensure WRF could run successfully, so I chem_opt = 0.
I reduced the processor from 140 to 63 and the model ran successfully. But when I further used nesting, the model still crashed. I have tried different time steps and processors, but all of them didn't work. The problem may be related to the Qinghai Tibet Plateau region that I am simulating. The simulation of WRF on steep terrain is very unstable. I want to know if there is any way to make it more stable. I have already used low resolution geographic static data.
Thanks for your help.

 

[fei Xiao]

Hi,
The error when you set chem_opt = 0 is

-------------- FATAL CALLED ---------------
FATAL CALLED FROM FILE:  <stdin>  LINE:     337
ERROR: wet scavenging option requires chem_opt = 8 through 13 or 31 to 36 or 41 to 42 or 109 or 503 or 504 or 601 or 611 to function.
-------------------------------------------

which indicates that you can't set this to 0 with other options you're using.

If going back to the original namelist.input file you sent, can you try to run this with 100 processors, instead of 140. This likely won't make a difference, but it's worth a try to see if the model doesn't like the decomposition.

Otherwise, it would seem this may be a chemistry-specific option. Our team is not experienced with WRF-chem, and unfortunately WRF-chem has discontinued all support services for their version of WRF. However, there is a still a wrf-chem section on this forum and you can try posting there to see if anyone may be able to help you. If you'd like to ensure it's a chemistry problem, you can build WRF without any chemistry turned on, and run this test without any chem options in the namelist to see if it runs. If so, they you'll know it's related to wrf-chem.

I also found that the smaller the time step, the shorter the duration of model integration. For example, when the time step is 240, the model can be integrated to 4 hours. But when the time step is 120, the mode only integrated for 1-2 hours.