g.e23_a17c.GJRAv4.TL319_t232_zstar_N65.2024.041

[iangrooms]

Purpose:
Last summer I implemented a kinetic-energy conserving correction to the velocity remap. You can use this in combination with any remap scheme; the correction preserves the order of the scheme while adjusting the velocity profile so that kinetic energy is conserved to machine precision. The goal of this run is to test the effect of this scheme in the 2/3 degree configuration. There is also a new diagnostic that measures the rate of depth-integrated KE loss associated with the remap (before the correction).

---

Description:
I made this as close to run 004 as I could, but ended up using sandbox alpha17a instead of alpha16g.

---

Case directory:

• Locally (if still available):
  /glade/work/igrooms/ALE/g.e23_a17a.GJRAv4.TL319_t232_zstar_N65.2024.041

---

Sandbox:

• Locally (if still available): /glade/work/igrooms/ALE/cesm2_3_alpha17a

• On github: https://github.com/ESCOMP/cesm/tree/cesm2_3_alpha17a

• MOM6 code changes: https://github.com/iangrooms/MOM6/tree/vel_remap_ke_cons

---

Diagnostics:

• Diags (if available):
  TODO

---

Output:

• Raw output on derecho (if still available): /glade/derecho/scratch/igrooms/archive/$CASE/

---

Contacts:
@iangrooms

---

Extra details:
--compset 2000_DATM%JRA-1p4-2018_SLND_CICE_MOM6_DROF%JRA-1p4-2018_SGLC_SWAV_SESP

MOM_override:

#override INPUTDIR = "INPUT"
STOCH_EOS = True
STANLEY_COEFF = 0.5
USE_STANLEY_PGF = True
USE_STANLEY_ML = True
USE_STANLEY_ISO = True
USE_STANLEY_GM = True
NDIFF_TAPERING = True
KHTR_USE_EBT_STRUCT = True
DO_GEOTHERMAL = True
GEOTHERMAL_SCALE = 1.0
GEOTHERMAL_FILE = "geothermal_davies2013_tx2_3_20231120_cdf5.nc"
GEOTHERMAL_VARNAME = "geothermal_hf"
REMAP_VEL_CONSERVE_KE = True

[iangrooms]

Mean rate of KE loss due to velocity remap (years 31-60):

The rate of KE loss is not very big, and is very localized along the equator, near complex topographic features, and in the strait of Gibraltar and in the straits between the Aegean and Black Seas.

Salinity bias at 700m in 004 (004 minus obs; years 31-60):

There is a high-salinity bias in the Atlantic and a corresponding low salinity bias in the western Mediterranean. The Black Sea is also too salty at depth.

Difference in mean salinity at 700m, 041 minus 004 (years 31-60):

The sign convention is such that if the colors in the bottom figure are opposite to the colors in the middle figure, then the bias is reduced. The salinity biases in the Atlantic, western Mediterranean, and Black Sea are all reduced with the KE-conserving velocity remap. This is presumably because the corrected remap improves the structure of the flow through the straits connecting these seas, which is consistent with the top figure above, showing where the correction is largest.

[iangrooms]

I ran a hybrid-coordinate case and confirmed that it has a similar effect on the salinity bias. The code was less stable in the hybrid case, so I changed the code to limit the size of the correction, which enabled it to run stably for a full JRA cycle. Evidently the KE loss during remap can intermittently be quite large for the hybrid coordinate, in which case my correction is also large but produces an unrealistic velocity profile leading to truncations; limiting the correction avoids this. The code is now in a PR to dev/ncar.