-
Notifications
You must be signed in to change notification settings - Fork 249
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Regional restart run is creating different results. #408
Comments
@JacobCarley-NOAA May I ask who may work on this issue? |
Was this resolved with @DomHeinzeller's commit (#417) or is there still an outstanding issue? (I think this may only be in the release branch at this time.) |
This was resolved for the release/public-v2 branch, and the only changes I had to make were (a) code that was added to develop between July 2020 (when the release branch was cut) and December 2020 (from Moorthi and Shan), and (b) NoahMP-specific code changes (these are not yet in develop). |
@climbfuji I see the restart reproducibility tests in release/public-v2 are for suite files FV3_GFS_v15p2 and FV3_RRFS_v1alpha. While the regional LAM tests are using FV3_GFS_v15_thompson_mynn, and the restart test does not reproduce the control with this suite file. So the release/public-v2 does not support FV3_GFS_v15_thompson_mynn? |
It doesn't support it officially, but it will give b4b reproducible results if you add the suite manually and compile it. |
I tested the latest develop branch, the restart test using FV3_GFS_v15_thompson_mynn do not reproduce from control. Do you mean the release/public-v2 has restart reproducibility for FV3_GFS_v15_thompson_mynn? |
Yes, 99% confident that it has. But according to Lucas Harris, we shouldn't bring over all these changes 1:1, but fix the regional boundary update script and then remove any unnecessary mpp_update_domain calls in the dycore. I am happy to assist or lead this effort, but not this week. |
The issue is fixed. |
Description
Regional restart regression test is comparing results to itself, and not to control run.
For global application all is OK,
To Reproduce:
I used regression test to create CONTROL and RESTART work directories and repeated runs using quilting=trrue
phyf*** and dynf*** files differed between the two runs.
Additional context
Output
Screenshots
output logs
The text was updated successfully, but these errors were encountered: