Fig. 5: Meridional ocean heat transport MHT [1 PW = 10¹⁵W] in the Southern Hemisphere and its link to zonal-mean potential temperature [°C] and zonal-mean sea-ice changes [%].

a MHT in the low-resolution (LR) and high-resolution (HR) experiment for 1990–2019 (light green/grey) and 2070–2099 (dark green/black). b Zonal-mean sea-ice change and c zonal-mean ocean potential temperature change in LR and HR between 1990–2019 and 2070–2099. Vertical solid lines in panel (b) give the location of the September sea-ice edge in 1990–2019, vertical dashed lines the location in 2070–2099. d Zoom (40–80° S) into the anomalous MHT (2070–2099 minus 1990–2019). Climate models participating in the Coupled Model Intercomparison Project (CMIP5) are grouped into two sets with ‘higher’/‘lower’ resolution than 90 km (for the computation, see “Methods”). e Difference in the response (HR-LR) of the panels in (b). Vertical solid (dashed) lines denote the location of the September sea-ice edge in 1990–2019 (2070–2099) in LR, HR, and in CMIP5 (“Methods”). f Difference in the response (HR-LR) of the panels in (c). The signals are significantly larger than typical unforced running 30yr-variability (standard deviation) of zonal-mean potential temperature in the 1950 control simulations. Hatching indicates non-significant regions where the signals and difference in the response are smaller than two standard deviations (see “Methods”). The “delta approach” (see "Methods") has been applied in (b)–(d).