the coupled model requires FV3 and CMEPS updates for frac grid normalization #366
Comments
|
This change will involve normalizing the fluxes from ATM to the OCN by the open water fraction that ATM used to calculate those fluxes. In CMEPS, the field imported from the ATM will have the name The normalization change will be implemented in esmFieldsExchange_nems when the fluxes are mapped from ATM to OCN by changing the normalization from Because the packed field bundles can have only one normalization type, with this change I needed to change the When the mapping type is changed from consf to bilinear for sea level pressure, the fields of sea level pressure exported to the OCN are in fact different (by small amounts). Two effects are seen. First, in the 1deg ocean configuration, the answers were bit identical to mapping sea level pressure conservatively. This can only be true if the 1deg configuration is not actually using this field. See MOM6 issue #58. Secondly, the creation of the bilinear routehandle resulted in a small number of points now being mapped from OCN to land points (on the ATM grid) when those fields were mapped with conservative destination. This indicates some cross-contamination of the routehandles, either at the ESMF or CMEPS level. Following a suggestion by @rsdunlapiv, I added the retrieval of the We're continuing to debug this issue now, with additional help from the ESMF team. |
|
@denise Worthen - NOAA Affiliate <[email protected]> While reading
this, may I ask if we can have wind stress interpolated through vector
bilinear rather than scalar bilinear? It will make significant differences
near the pole region.
…On Tue, Mar 2, 2021 at 8:45 AM Denise Worthen ***@***.***> wrote:
This change will involve normalizing the fluxes from ATM to the OCN by the
open water fraction that ATM used to calculate those fluxes. In CMEPS, the
field imported from the ATM will have the name Sa_ofrac and will be
carried in the fractional FieldBundle with the name aofrac.
The normalization change will be implemented in esmFieldsExchange_nems
when the fluxes are mapped from ATM to OCN by changing the normalization
from none to aofrac.
Because the packed field bundles can have only one normalization type,
with this change I needed to change the maptype for the sea level
pressure from consf to bilinear. This is actually the correct mapping type
for sea level pressure since it is a global unmasked field. (I believe that
consf mapping type was carried over from the NEMS mediator, when all fields
mapped from ATM to OCN used nearest neighbor conservative.)
When the mapping type is changed from consf to bilinear for sea level
pressure, the fields of sea level pressure exported to the OCN are in fact
different (by small amounts). Two effects are seen. First, in the 1deg
ocean configuration, the answers were bit identical to mapping sea level
pressure conservatively. This can only be true if the 1deg configuration is
not actually using this field. See MOM6 issue #58
<NOAA-EMC/MOM6#58>.
Secondly, the creation of the bilinear routehandle resulted in a small
number of points now being mapped from OCN to land points (on the ATM grid)
when those fields were mapped with conservative destination. This indicates
some cross-contamination of the routehandles, either at the ESMF or CMEPS
level. Following a suggestion by @rsdunlapiv
<https://github.com/rsdunlapiv>, I added the retrieval of the dstStatusField
from the RegridStore in CMEPS. This field contains a flag for reporting
the type of masking which is done at each point in the destination grid
when the routehandle is being created. The values contained in this field
are described here
<http:https://earthsystemmodeling.org/docs/nightly/develop/ESMF_refdoc/node9.html#SECTION090150000000000000000>.
In the case where a bilinear routehandle is being created, the points which
now appear as OCN->ATM land are all associated with a dstStatusField of
5. When no bilinear routehandle is created, these points all contain a
dstStatusField of 0. Since these points are in fact land, the correct
value in this case is in fact 0.
We're continuing to debug this issue now, with additional help from the
ESMF team.
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#366 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AI7D6TOONE7G455U4YTJKZDTBTTV7ANCNFSM4WBBG4LA>
.
|
|
@junwang-noaa I would need to understand more before giving a certain answer. I only have a very vague memory of this from very early days when we were coupling GSM and HYCOM. |
|
Please see the link below on the impact interpolation on wind related
fields and surface pressures.
https://docs.google.com/presentation/d/1I47QQsxC8dH1FIsGJmwiKPg4We9j8Ytw/edit#slide=id.p20
…On Tue, Mar 2, 2021 at 1:01 PM Denise Worthen ***@***.***> wrote:
@junwang-noaa <https://github.com/junwang-noaa> I would need to
understand more before giving a certain answer. I only have a very vague
memory of this from very early days when we were coupling GSM and HYCOM.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#366 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AI7D6TL5HA7HJAMKWLL3STTTBURWHANCNFSM4WBBG4LA>
.
|
|
CMEPS has this mapping type available: mappatch_uv3d = 9 ! rotate u,v to 3d cartesian space, map from src->dest, then rotate back |
|
Great, it's good to use that mapping method for wind related fields (any
vector component fields).
…On Wed, Mar 3, 2021 at 1:29 PM Denise Worthen ***@***.***> wrote:
CMEPS has this mapping type available:
mappatch_uv3d = 9 ! rotate u,v to 3d cartesian space, map from src->dest,
then rotate back
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#366 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AI7D6TKXFP56CYIBAC36HZLTBZ5YRANCNFSM4WBBG4LA>
.
|
…on for geogrid files (fix_am directory) (#366) * Point exregional_make_ics.sh script to the new location for geogrid files (fix_am directory). Set the config.community.sh external model staging user-defined variable to "false." * Change path for geogrid files on Hera. * Remove if statements for geogrid files in exregional_make_ics.sh, and update paths to fix_am/fix_sfc_climo in setup.sh to be consistent across platforms and remove paths to personal directories. * Remove GRID_GEN_METHOD and reset EXPT_SUBDIR value. Keep USE_USER_STAGED_EXTRN_FILES set to "TRUE" but comment it out and add additional information for the user to uncomment this line if they wish to use this feature. * Reset numerous variables to correct values
Description
Both FV3 and CMEPS will have an update to normalize the fluxes being sent from FV3 through CMEPS for the ocean and ice. When these are ready, the coupled model will need to be updated and tested.
Related Issues
FV3 #230
CMEPS #29
The text was updated successfully, but these errors were encountered: