Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

Segmentation fault

This post was from a previous version of the WRF&MPAS-A Support Forum. New replies have been disabled and if you have follow up questions related to this post, then please start a new thread from the forum home page.

Zhi Mao

New member
Hello,
I am simulating a convective precipitation process over the Tibetan Plateau.I know steep terrain will cuase seg-fault(my rsl.error file show that),so I typed "ulimit -s unlimited",set time_step=4*DX,add smooth_cg_topo = .true. in the &domains section and set epssm = 0.2.But after I did that,the seg-fault still occurred.I can think of only one reason:I yerterday set dx=30000 for d01 and dx=10000 for d02 was OK,but today I just changed dx=12000 for d01 and dx=4000 for d02.Whether resolution will also cause seg-fault? The following are my namelist.wps and namelist.input.
 

Attachments

  • namelist.wps
    1.3 KB · Views: 55
  • namelist.input
    3.9 KB · Views: 62
I've solved the problems I've had,the seg-fault(or CFL errors) do not occur.BUT for the stability of the model,I set time_step=36 for d01(dx=12000,293x193 grids) and time_step=4 for d02(dx=4000,331x304 grids),it takes too much time.Although I typed "mpirun -np 64 ./wrf.exe",it could take more than ten hours to run one day.:(
 
Please set epssm = 0.9, 0.9, 0.9 in your nameless and increase time step.Please let me know whether this works.
 
Dear Ming,
I have done as you said,set epssm = 0.9, 0.9 and increase time step in namelist.input,but the seg-fault still occur.After that,I changed namelist.input many times until the model worked.Here are my final namelist.wps, namelist.input,and domain preview(the two black frames are d01 and d02).It maybe take a lot of time,although I typed "nohup mpirun -np 64 ./wrf.exe > nohup.log 2>&1 &".
 

Attachments

  • namelist.wps
    1.3 KB · Views: 49
  • namelist.input
    3.9 KB · Views: 74
  • domain_preview.png
    domain_preview.png
    65.4 KB · Views: 2,216
Hi Zhi,
We are aware that many issues may arise due to the complex terrain and large topography gradient over the Tibetan Plateau.
We usually recommend avoiding high-resolution run over that region, and turn on damping options.
epssm and w_damping can be helpful in certain cases.
You may also try diff_opt =2, 2, which is more physically reasonable.
 
Hi Ming,
Thanks for your valuable suggestions!
Because the area of d01 is so big, I changed the resolution dx=20000 for d01,but didn't change dx=4000 for d02.
This time I set epssm=0.9,0.9,w_damping=1 and diff_opt =2, 2.The time_step=3dx is fine for d01 and d02.
Now do I just need to wait for the model to complete the calculation? Although I may have to wait a long time.
 

Attachments

  • namelist.wps
    1.3 KB · Views: 52
  • namelist.input
    3.9 KB · Views: 64
I've solved the problem.Because of large topography gradient over the South Tibetan Plateau,the resolution less than 6km is not OK.Thanks!
 
Top