v_cfl problem when coupling with OASIS

[laetis]

Hi,
I am trying to run WRF+NEMO coupled with oasis on a global run.
I keep having this error right at the beginning:
Timing for main: time 2024-12-15_06:00:20 on domain 1: 7.16088 elapsed seconds
9340 points exceeded v_cfl = 2 in domain d01 at time 2024-12-15_06:00:20 hours
Max W: 343 1 33 W: ******* w-cfl: ******* dETA: 0.01
Followed by a segmentation fault.
I set a very little time_step (20).
What is strange is that if I run the exact same case without coupling (with wrf non coupled executable) without changing anything it runs fine at this point.
My namcouple part for ocean-> atmosphere is:
O_SSTSST WRF_d01_EXT_d01_SST 1 10800 1 fdocn1.nc EXPOUT
180 148 361 181 ocnt atmt LAG=10800
P 2 P 0
SCRIPR
BILINEAR LR SCALAR LATLON 1
# Ocean X-velocity (o->a 2)
O_OCurx1 WRF_d01_EXT_d01_UOCE 1 10800 1 fdocn1.nc EXPOUT
180 148 361 181 ocnt atmt LAG=10800
P 2 P 0
SCRIPR
BILINEAR LR SCALAR LATLON 1
# Ocean surface Y-velocity (o->a 3)
O_OCury1 WRF_d01_EXT_d01_VOCE 1 10800 1 fdocn1.nc EXPOUT
180 148 361 181 ocnt atmt LAG=10800
P 2 P 0
SCRIPR
BILINEAR LR SCALAR LATLON 1

I set all fields in the restart files to 0.
Thanks for insights.
Best regards

 

[Ming Chen]

I don't know much of the ocean model NEMO and OASIS. So the ideas below are only valid for WRF-ARW

The v_cfl = 2 error indicates that the model becomes unstable during its integration. The instability can be attributed to various factors. What we usually do to suppress the instability includes:

(1) turn on w_damping
(2) increase epssm
(3) reduce time_step.

If they didn't work, then something may be wrong in the model physics. Hope this is helpful for you.

 

[laetis]

Thanks for your answer.
I tried all these, the fact that the simulation runs fine without coupling make me think I probably did something wrong in the coupling.
Best

 

[Ming Chen]

Coupling ocean model with WRF is challenging. It has been proposed long time ago that WRF should be coupled with a regional ocean model, but this work work has never been done. Now NCAR freezes WRF development, and we have to leave this task to the community.