Extended Data Fig. 7: Comparison of SST-based AMOC indices.
a Comparison of four SST-based indices in observations (HadISST). SSTDIPOLE (black)20: North Atlantic (70°W-30°E; 45°N-80°N) minus South Atlantic (70°W-30°E; 40°S-0) SSTs. SSTDIPOLE-SWSA due to the Rossby wave propagation (pink): subpolar North Atlantic (60°W-10°W; 50°N-70°N) minus southwestern South Atlantic SSTs (70°W-30°W; 45°S-10°S). SSTDIPOLE-SESA due to Kelvin wave propagation (green): subpolar North Atlantic (60°W-10°W; 50°N-70°N) minus eastern South Atlantic (5°E-20°E; 30°S-10°S). SSTSPG-PA (blue): averaged SST anomalies from Nov-May in the subpolar gyre (as per ref. 23). In all indices twice the global mean temperature SST anomaly has been subtracted to remove the global warming signal7. The signal-to-noise ratio (R) is indicated for each index in observations, computed as the ratio between the mean change (values from 2000-2023 minus 1900-1950) and the standard deviation from 1900-1950. b-c Comparison of simulated and observed b SSTDIPOLE and c SSTDIPOLE-SESA. Red lines indicate the observed indices. Grey shading indicates the 10-90 percentile range of a 40-member ensemble of the ACCESS-ESM1.5 hist and ssp585 simulations. Thin grey, blue and purple lines show a single member of each ensemble as a reference for the model’s internal variability. Light blue shading indicates 8-member ensemble range of hist-fw simulations and purple shading indicates 8-member ensemble range of ssp585-fw simulations. The same comparison for SSTSPG-PA and SSTDIPOLE-SWSA is found in Fig. 4 of the main paper, where SSTDIPOLE-SWSA is referred to as the ‘modified SSTDIPOLE index’ (SSTDIPOLE*).
