Hello everyone,
I'm encountering a strange issue with a 3-domain nested simulation where wrf.exe exits prematurely but without any error messages. My WRF model's version is 4.4.
My Goal:To run a simulation with spectral nudging applied only to the outermost domain (d01). The inner domains (d02, d03) should run freely without nudging.
The Problem:The real.exe step completes successfully and creates wrfinput_d01/d02/d03, wrfbdy_d01, and wrffdda_d01. However, when I run wrf.exe, it initializes domain 1, correctly reads the spectral nudging settings, and then stops right as it begins to allocate space for domain 2. The rsl.out.0000 file ends cleanly, and the job completes with a success code (ExitCode 0:0). There are no segmentation faults or CFL errors.
Troubleshooting Steps:I have confirmed the following through testing:
1. I can run the 3-domain nested simulation without the configurations in &fdda module.
2. A single-domain (d01 only) simulation runs perfectly with the exact same spectral nudging settings enabled.
This isolates the problem to the interaction between a nudged parent domain (d01) and a non-nudged child domain (d02) during initialization.
I have already tried several potential solutions without success, including:
Has anyone encountered this specific behavior before? Is there a known namelist setting or workaround required for a nested run where only the parent domain is nudged?
I have attached my namelist.input, namelist.wps, and the rsl.out.0000 file from the failed 3-domain run.
I'm encountering a strange issue with a 3-domain nested simulation where wrf.exe exits prematurely but without any error messages. My WRF model's version is 4.4.
My Goal:To run a simulation with spectral nudging applied only to the outermost domain (d01). The inner domains (d02, d03) should run freely without nudging.
The Problem:The real.exe step completes successfully and creates wrfinput_d01/d02/d03, wrfbdy_d01, and wrffdda_d01. However, when I run wrf.exe, it initializes domain 1, correctly reads the spectral nudging settings, and then stops right as it begins to allocate space for domain 2. The rsl.out.0000 file ends cleanly, and the job completes with a success code (ExitCode 0:0). There are no segmentation faults or CFL errors.
Troubleshooting Steps:I have confirmed the following through testing:
1. I can run the 3-domain nested simulation without the configurations in &fdda module.
2. A single-domain (d01 only) simulation runs perfectly with the exact same spectral nudging settings enabled.
This isolates the problem to the interaction between a nudged parent domain (d01) and a non-nudged child domain (d02) during initialization.
I have already tried several potential solutions without success, including:
- Ensuring every parameter in &fdda has placeholders for all 3 domains.
- Setting grid_fdda = 2, 2, 2 but with nudging coefficients for d02 and d03 set to 0.0.
Has anyone encountered this specific behavior before? Is there a known namelist setting or workaround required for a nested run where only the parent domain is nudged?
I have attached my namelist.input, namelist.wps, and the rsl.out.0000 file from the failed 3-domain run.