Update JRA55 forcing implementation #562
Conversation
Several bug fixes addressed - Reading two timesteps instead of 1 now (was reading the same timestep into both slots) - Advancing files by year (was just cycling the initial year in a multi-year run) - Time interpolation updated (was not working in multi-year runs) - Interpolate 4 input states in time, but hold 3 input fluxes constant over 3 hour forcing period - Additional output diagnosing forcing filename as it evolves
apcraig
requested review from
eclare108213,
dabail10,
daveh150,
rallard77 and
TillRasmussen
|
A couple questions/comments.
|
|
I don't think we need QC for this one. The important thing is to get this fixed quickly. I think there is some forcing documentation, somewhere, since we reference the data sets, but it might not explain how the forcing module/routines work. That information should be added, preferably in this PR unless it'll take a long time -- I'd be fine with addressing that as a separate issue, e.g. as part of a 'clean up' phase. |
|
The code changes Tony committed look good. I like putting the reads in a for loop. Were we going to standardize the JRA55 forcing names to include the grid. So it would be JRA55_03hr_forcing_gx3_2005.nc If we do that, I think we can remove the hardcoded '/8XDAILY/ ' part. That should be defined in atm_data_dir namelist varaible. |
|
I agree with Dave's suggestion on standardizng file names and removing
the hardcoded '/8XDAILY" and putting that in the atm_data_dir namelist
variable.
Rick
On 2/10/2021 8:32 AM, daveh150 wrote:
The code changes Tony committed look good. I like putting the reads in
a for loop.
Were we going to standardize the JRA55 forcing names to include the
grid. So it would be
JRA55_03hr_forcing_gx3_2005.nc
JRA55_03hr_forcing_gx1_2005.nc
JRA55_03hr_forcing_tx1_2005.nc
If we do that, I think we can remove the hardcoded '/8XDAILY/ ' part.
That should be defined in atm_data_dir namelist varaible.
—
You are receiving this because your review was requested.
Reply to this email directly, view it on GitHub
<#562 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AG63LBZ5HF7SSRX5RXQSBOTS6KKJJANCNFSM4XFD2JZQ>.
--
logo
*Richard A. Allard *
*Head, Nearshore and Coupled Models Systems Section*
*Ocean Sciences Division NRL Code 7322*
*Stennis Space Center, MS 39529*
*T: 228.688.4894*
richard.allard@nrlssc.navy.mil <mailto:richard.allard@nrlssc.navy.mil>
****
******
|
| The ``default`` ocean setting is the standard setting used in standalone CICE runs. | ||
| In this mode, the sea surface salinity is set to 34 ppt and the sea surface | ||
| temperature is set to the freezing temperature at all grid points and | ||
| held constant. Other ocean coupling fields are set to zero. No files are read. |
There was a problem hiding this comment.
held constant unless the mixed layer parameterization is turned on, in which case the SST evolves
There was a problem hiding this comment.
and when the mixed layer parameterization is on, it can use data from the ocean forcing file, which we only have for gx1. It might not be used in our testing and might not be offered as a forcing data set...
There was a problem hiding this comment.
I updated the text a bit, but did not add the stuff about an ocean forcing file. I don't think we test anything other than the default, and I don't think we've released any ocean forcing dataset.
PR checklist
Update JRA55 forcing implementation
apcraig
Full test suite run on cheyenne with 3 compilers, all results change (as expected) https://github.com/CICE-Consortium/Test-Results/wiki/cice_by_hash_forks#96a5103a495df1caa5703c478f143a4bc81cde79
Also ran 10+ years for 1995-2005 on cheyenne with debugging output on (for part of the time). As far as I can tell, things are working correctly.
Several bug fixes addressed
See #249, #343, #482, #558