cfl error over nested domain run

[samudra]

I am trying a run a high resolution simulation, taking ERA5 atmopsheric dataset for 12 hours. I have attached the namelist.wps file for the domain details. I am using pbs to give the runs on 256 cores in parallel. I have also included the namelist.input file here, and two rsl.error. files,one highlighting the error and second showing nan values at a particular gridpoint. The error is there in multiple rsl file, although the error is almost smae in all of them, related to cfl conditions not meeting. The domain is also attached for reference.
Kindly look into it soon.
Thanks in advance.

 

[meteoadriatic]

Hi,
try epssm = 1.0 in &dynamics as a first measure but I doubt that it will be enough to solve CFL crashes in your case.

Also, the possible place that might bite you is the combination of feedback = 1 and different values of radt in domains. When feedback is enabled, they should be the same, but if you ask me, I would disable feedback as I only had issues with it (it might even be the cause of crashes).

As you run over Himalayas at 1.5 km grid distance, the most probable reason for crash is too steep terrain slope. I expect that you will have to smooth terrain out. For that, look into GEOGRID.TBL for options of smoothing HGT field (increase number of passes and rerun geogrid).

 

[kwerner]

I am trying a run a high resolution simulation, taking ERA5 atmopsheric dataset for 12 hours. I have attached the namelist.wps file for the domain details. I am using pbs to give the runs on 256 cores in parallel. I have also included the namelist.input file here, and two rsl.error. files,one highlighting the error and second showing nan values at a particular gridpoint. The error is there in multiple rsl file, although the error is almost smae in all of them, related to cfl conditions not meeting. The domain is also attached for reference.

Hi, take a look at Segmentation Faults and CFL Errors for some potential solutions to your CFL issue.