The Relationship between the Southern Ocean and the Eastern tropical Pacific in unforced and forced climate model simulations

The sea surface temperature (SST) over the eastern tropical Pacific significantly influences global-mean climate feedback and may be driven in part by the SST over the Southern Ocean. Previous studies demonstrated a teleconnection from the Southern Ocean to the eastern tropical Pacific by perturbing the Southern Ocean climate. We investigate if this teleconnection holds in a fully coupled, freely running climate system using CMIP6 models. We assess the relationship between the Southern Ocean (SO) and the eastern tropical Pacific (SEP) by calculating correlations between SO and SEP SST timeseries within each model and regressions between mean SO and SEP SSTs across models. We show robust, positive SO-SEP relationships in an unforced climate using pre-industrial SSTs, in a forced climate using SST anomalies between pre-industrial and quadrupled CO2 simulations, and in the SST pattern of the forced response relative to the global-mean SST anomaly. The strength of SO-SEP correlations is positively related to the stratocumulus cloud feedback off the west coast of South America, and negatively related to ocean heat uptake in the same region. As both shortwave cloud feedback and ocean heat uptake are underestimated in climate models, understanding their effects on SO-SEP teleconnections and their interactions is crucial for determining the strength of SO-SEP teleconnection in the real world and its trustworthiness in climate model projection.

Methods
The original data for SST and for calculating the radiative restoration strength, shortwave cloud feedback, and ocean heat uptake are downloaded from https://aims2.llnl.gov/search.

The original monthly data has been converted to annual data and re-grided to 2.5o x 2.5o spatial resolution.

SST anomalies are calculated for each grid by subtracting the linear regression fit of timeseries of piControl simulations from abrupt-4xCO2 simulations.

The main SST pattern used in the study is calculated by dividing SST anomalies at each grid point by the global-mean SST anomalies at each timestep. Other SST patterns are calculated by subtracting local SST anomalies at each grid point from the global-mean SST anomalies at each timestep, by regressing local SST anomalies at each grid point onto the global-mean SST anomalies across the entire timeseries (150 years), and by applying principal component analysis (PCA) to the three-dimensional SST anomalies.

The climatological total wind advection is calculated using climatological zonal and meridional winds by √(U2 + V2) in pre-industrial control simulations across the entire timeseries (150 years). 

The anomalous total wind advection is calculated by subtracting the climatological total winds from the forced total winds that calculated in abrupt4xCO2 simulations across the entire timeseries.

The wind-evaporation-SST (WES) feedback is calculated using the surface upward latent heat flux (LH), the wind advection, SSTs, and sea-level air pressure. Regressing surface upward latent heat flux, SSTs, sea-level air pressure, zonal (u) and meridional winds onto normalized SST averaged over a small region in the southeast Pacific (19°S–15°S, 103°W–107°W) in pre-industrial simulations provides a spatial pattern of WES feedback mechanism in an unforced climate. 

The WES feedback on SST pattern is calculated by regressing anomalous surface upward latent heat flux, SSTs, sea-level air pressure, zonal and meridional winds onto normalized SST patterns averaged over a small region in the southeast Pacific (19°S–15°S, 103°W–107°W).

The WES parameter (WESp) in pre-industrial control simulation is calculated based on WESp = LH · u / wind_bar2. wind_bar accounts for submonthly wind variance and is calculated by √(U2 + V2 + w_hat2), where w_hat is set to 4m/s.

The changes in WESp is calculated by subtracting WESp in pre-industrial control simulation from WESp in abrupt4xCO2 simulations.

Radiative restoration strength is calculated by regressing the local shortwave cloud radiative effect onto local SST in pre-industrial control simulations across the entire timeseries (150 years). The shortwave cloud radiative effect is the difference between the outgoing shortwave flux at the top-of-atmosphere and the outgoing shortwave flux at top-of-atmosphere when there are no clouds (clear sky).

Shortwave cloud feedback is calculated by regressing anomalies of the local shortwave cloud radiative effect onto local SST anomalies across the entire timeseries (150 years). The anomalies of the shortwave cloud radiative effect are calculated by subtracting the linear regression fit of timeseries of piControl simulations from abrupt-4xCO2 simulations.

Ocean heat uptake is represented by the net surface downward heat flux, which is the difference between surface downward longwave flux, surface upward longwave flux, surface downward shortwave flux, surface upward shortwave flux, surface upward sensible heat flux, and surface upward latent heat flux.

Changes in ocean heat uptake are calculated by subtracting the linear regression fit of timeseries of piControl simulations from abrupt-4xCO2 simulations.