How to make inner 2 nests vortex-following?

[astansfield]

Hello,

I have been attempting to run a slightly modified TC test case simulation (see namelist attached) with 3 domains. Around 26 hours or so into the simulation, I keep getting a "MPT ERROR: Rank 282(g:282) received signal SIGSEGV(11)." error. I've attached a sample rsl error file. At first I thought this was a memory error, so I upped the number of nodes requested on Cheyenne and also requested the high memory nodes, but I keep running into this issue. When I looked at the PSFC variable at the last timestep before the model run fails, I noticed a large pressure anomaly near the bottom boundary of d02 (see bottom left panel in attached .png). I think this could be related to the TC getting too close to the d02 boundary?
I wanted both d02 and d03 to be vortex-following nests, but I'm not sure how to make 2 nests be vortex-following? Am I initializing the vortex too close to the d02 boundary perhaps?
I'm open to any other suggestions.

Thank you,
Alyssa

 

[Ming Chen]

Alyssa,
I looked at your namelist.input, which indicates that this is pretty big case with triply nested domains and large grid numbers. I am not sure which version of WRF you are running and how many nodes you used for this case. But I am highly suspicious that this failed case could be attributed to insufficient memory. I found the error message in your rsl file:

grid=<error reading variable: value of type `domain' requires 289448 bytes, which is more than max-value-size>,

Also, the PSFC pattern apparently is wrong along the boundary, which may indicate that some physics also went wrong.

We need more details (wrf version, compiling mode, run style etc.) before we can figure out the issue.

 

[astansfield]

Hi Ming,

Thank you for your reply. I am running WRF version 4.1.3. I configured WRF on Cheyenne using option 15 (dpmar, INTEL (ifort/iccc)) and option 3 for vortex-following nest. I compiled the idealized TC test case (em_tropical_cyclone).

 

[astansfield]

I am running the simulations using 25 high-memory nodes.

 

[Ming Chen]

Please note that vortex-following is designed for real-data case. It is not appropriate for the ideal TC case.

 

[astansfield]

Ming,
So the vortex-following nest function will not work for idealized cases? I have read a few recent papers that seem to suggest they use idealized WRF with vortex-following nests (Idealized Tropical Cyclone Responses to the Height and Depth of Environmental Vertical Wind Shear and Outer Rainbands–Driven Secondary Eyewall Formation of Tropical Cyclones).

 

[astansfield]

Posting again to see if anyone else has run idealized TC simulations using a vortex-following nest?

 

[Ming Chen]

I just take a quick look at the paper, --- i believe that the authors have modified the codes to realize their simulations. I will keep you updated if I find more information regarding this issue.