Ocean Initialization

[fobryan3]

I have recently been testing how different choices of ocean initialization impact the solutions. In particular, I have tried initializing forced ocean-ice cases with World Ocean Atlas 2023 JAN climatologies from 1955-1964, 1965-1974, and 1955-2023 as compared to the current default using WOA 2018 all time average. The main motivation was to see how avoiding the influence of the late 20th-early 21st century global warming signal might change (improve?) the spin-up. As you might expect, the SST converges very quickly across cases because the heat flux has a strong feedback with SST. The sea surface salinity stays a little more differnt despite restoring to the same WOA18 salinity.

However, the volume average ocean temperature (equivalently the total ocean heat content) stays distinct through a full cycle of JRA forcing (1958-2018). The cases initialized with the earlier decade observations stay cooler throughout the run. Notice that the 1965-1974 initialized case starts out the coolest but conversges with the 1955-1964 run mid-run. The case initialized with the all decades WOA23 data starts out warmer than the all decades WOA18 case presumably because it is increasingly dominated by the recent Argo data and stays warmer through the whole run.

The next steps are to compare the end-of-run errors relative to the 1991-2020 WOA23 30 year "climate normal" and to compare the spinup of coupled runs with initialized with early states of these forced runs.