GOSI10: Investigating ocean salinity response to sea ice developments

[emmafiedler]

For a run from 1958-2022, sea ice dynamics developments for GOSI10.beta.5 were found to increase the salt content in the subpolar gyre and Labrador Sea from around the mid-1980s, and the heat content of the subpolar gyre from around 2000, compared to GOSI10.beta.4. This issue documents investigations into which sea ice changes are causing this.

[emmafiedler]

AMOC behaviour in beta.5 compared to beta.4. Runaway changes are seen in the SPG (see results below) which are not acceptable.

MOC as a function of depth compared to OSNAP array

Experiments

Five runs are considered here to isolate the impact on the ocean of the beta.5 sea ice changes:

• GOSI10.beta.4 release [beta.4] (u-dq383)

• GOSI10.beta.5 release [beta.5] (u-dw515)

• GOSI10.beta.5 without Rothrock (1975) strength scheme [noR75] (u-dx103)
  (i.e. reverting to Hibler (1979): ln_str_h79=.false. and ln_str_r75=.true.; and reverting to linear redistribution function: ln_distf_exp=.false. and ln_distf_lin=.true.)

• GOSI10.beta.5 without aEVP rheology [noaEVP] (u-dx104)
  (i.e. reverting to EVP: ln_aevp=.false.)

• GOSI10.beta.5 without landfast ice scheme [noLF] (u-dx105)
  (ln_landfast_l16=.false.)

These are all JRA forced runs, at eORCA025, run for 65 years from 1958.

They all use the MLF time-stepping scheme.

Results

Beta.5 shows a substantial increase over time in SPG salt content, particularly compared to beta.4 which shows a reduction (compare green and blue lines). Over the first 30 years, the noLF run decreases SPG salt content, and noR75 and noaEVP both increase it. For noLF, the salt content increases markedly in the mid-1990s. For noR75, the salt content remains steady over the 65-year period, with oscillations. For noaEVP, salt content increases at a similar rate over the period to beta.5. Salt content itself is higher than beta.5 until it drops below in the final decade.

Until the mid-1990s, noR75 SPG heat content is closest to beta.4. In the mid-1990s there is a jump in SPG heat content for all runs. After this time, noLF is closest to beta.4 (and observations).

Further work

It has since been found that there is a difference in the response of the ocean model in the north Atlantic depending on whether each of the sea ice updates is switched off from beta.5 (as done here), or switched on from beta.4 (as done on #41), where switching them on does not cause these impacts on the SPG and the AMOC. There is also a differing response between using MLF or RK3 time-stepping.

This implies that the ocean impact cannot be attributed to a single beta.5 sea ice update, but rather is a cumulative response to all the sea ice changes together, and that this full set of changes pushes the AMOC towards a tipping point. Further work on this issue is documented on #41.