Segmentation Fault in WRF d03

[horinouchi]

Hi WRF Community,

I am attempting to reproduce heavy rainfall in d03 using WRF and a three-stage downscaling process.

I am encountering a Segmentation Fault error during a WRF run. The crash consistently occurs in domain 3 (d03), but I have not been able to resolve it despite trying several standard troubleshooting steps.I am currently repeating the experiments while varying the experimental settings.

Problem Summary:

• The model crashes with a Segmentation Fault during the integration of d03.
• Looking at the 1-hour rainfall distribution for d03 just before the crash across various sensitivity tests (testNo.20–36), there appears to be a common pattern where strong precipitation/convection is located near the domain boundary.

Troubleshooting Already Attempted (No Effect):

1. Reduced the number of MPI tasks / parallel cores.
2. Reduced time_step.
3. Set ulimit -s unlimited (stack size).
4. Enabled Adaptive Time-Stepping (use_adaptive_time_step = .true.).
5. Enabled Vertical Velocity Damping (w_damping = 1).
6. Enabled Smoothing options (smooth_option).

System & Model Setup:

• WRF Version: 4.5
• Compiler / MPI: [例: Intel OneAPI 2023 / OpenMPI 4.1]
• Domain Setup: 3 domains (d01, d02, d03), [2-way nesting]

Attachments / Log Snippets:

• Attached: namelist.input
• Attached: rsl.error.0000
• Attached: rsl.error.0007
• Attached: Plot of d03 1-hour rainfall just prior to the crash

Could this issue be related to boundary instabilities caused by strong convection passing through the d03 boundary, or a numerical instability in specific physics options? Any suggestions on how to further isolate or fix this problem would be greatly appreciated.

Thank you in advance for your support!

Best regards,

horinouchi