[RegCNET] [EXTERNAL] An error in running simulations
Joshua Fafanyo
fafanyoj at gmail.com
Sat Mar 12 04:50:41 CET 2022
Hello everyone, decreasing the value for the dt worked fine.
I am however surprised that dt did not affect any of the other test files I
am running (I am running several test files and changing only the
convective schemes while holding all other parameters constant). This
particular test file was running on the Kuo convective scheme.
Did the convective scheme contribute to this?
Thank you all.
Regards
On Fri, Mar 11, 2022 at 3:22 PM Ashfaq, Moetasim <mashfaq at ornl.gov> wrote:
> Hi Travis and Johsua,
>
> The CFL violation error is seemingly not over the high topography areas
> (see ii,jj values). The CORDEX Africa domain also cuts through the Iranian
> Plateau in the east, so topography near the boundary may not be an issue
> here. In our experience, if reducing dt was not resolving the problem, the
> error most likely was due to problems in the input data. A careful
> inspection of input data files at all levels around the time step where
> model crashes may also help diagnose the problem.
>
> Thanks,
>
> -Moet
>
> On Mar 10, 2022, at 8:07 AM, O'Brien, Travis Allen <obrienta at iu.edu>
> wrote:
>
> Dear Joshua,
>
> I can’t say for certain what the problem is, but there are two common
> things to check when getting this issue with a new domain: (1) reduce dt,
> and (2) modify the domain to avoid cutting across high topography at the
> boundaries.
>
> My guess is that (1) isn’t going to help, since you have a dt of 120s in
> your file and a grid spacing of 60km: that corresponds to an advective CFL
> parameter of about 0.1 for ~60 m/s upper-level winds, which should be
> stable. It’s worth a try though: reduce dt by ½ and see if it helps.
>
> My hunch is that (2) is the problem. I’ve attached an annotated version of
> the topography that is generated when I run terrain on your namelist file.
> I’ve circled regions where the boundary cuts across relatively high
> topography. Such regions can be problematic for stability. I suggest
> trying to change the geometry of your domain to avoid cutting across high
> topography to the extent possible. It’s not always possible to totally
> avoid cutting across topography; in that case, try to cut across as low an
> elevation as possible.
>
> With Kind Regards,
> -Travis-
>
> --
> Travis A. O'Brien
> Pronouns: he/him
>
> Assistant Professor
> Earth and Atmospheric Sciences
> Indiana University, Bloomington
>
> Visiting Faculty
> Climate and Ecosystem Sciences Division
> Lawrence Berkeley Lab
>
> Topical Editor
> Geoscientific Model Development
>
> hxxps://earth.indiana.edu/directory/faculty/obrien-travis.html
> +1 (812) 856-6298
>
> *From:* RegCNET <regcnet-bounces at lists.ictp.it> *On Behalf Of *Joshua
> Fafanyo
> *Sent:* Monday, March 7, 2022 2:32 AM
> *To:* regcnet at lists.ictp.it
> *Subject:* [RegCNET] An error in running simulations
>
> I am trying to run a simulation over west Africa but I keep getting this
> error message despite efforts to resolve it:
>
>
> "READY BC from 1992060118 to 1992060200
> ATM variables written at 1992-06-01 18:00:00 UTC
> SRF variables written at 1992-06-01 18:00:00 UTC
> SHF variables written at 1992-06-01 18:00:00 UTC
> RAD variables written at 1992-06-01 18:00:00 UTC
> MAXVAL ATEN T : 2.0630375226624444
> II : 40 , JJ : 76 , KK : 13
> MAXVAL ATEN U : 1.0182738287789692
> II : 40 , JJ : 77 , KK : 10
> MAXVAL ATEN V : 1.0219311819551942
> II : 40 , JJ : 76 , KK : 2
> WHUUUUBBBASAAAGASDDWD!!!!!!!!!!!!!!!!
> No more atmosphere here....
> CFL violation detected, so model STOP
> #####################################
> # DECREASE DT !!!! #
> #####################################
> -------------- FATAL CALLED ---------------
> Fatal in file: mod_tendency.F90 at line: 695
> CFL VIOLATION
> -------------------------------------------
> mod_tendency.F90 : 695: 1
> Abort called by computing node 0 at 2022-03-07 06:42:27.650
> +0000
> Execution terminated because of runtime error
> --------------------------------------------------------------------------
> MPI_ABORT was invoked on rank 0 in communicator MPI COMMUNICATOR 3 DUP
> FROM 0
> with errorcode 1.
>
> NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes.
> You may or may not see output from other processes, depending on
> exactly when Open MPI kills them."
>
> Please find attached the test file.
> Thank you for your help.
> <test_002_topo_annotated.png>
> _______________________________________________
> RegCNET mailing list
> RegCNET at lists.ictp.it
> hxxps://lists.ictp.it/mailman/listinfo/regcnet
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.ictp.it/pipermail/regcnet/attachments/20220312/4a1b72bb/attachment.htm>
More information about the RegCNET
mailing list