Skip to main content
. 2022 Nov 21;119(47):e2209431119. doi: 10.1073/pnas.2209431119

Fig. 4.

Fig. 4.

(A) TMT trend (1979 to 2014) for the CESM2-SBB ensemble average minus the CESM2 ensemble average. (B) Probability distribution function of tropical (30°S to 30°N) TMT trends for CMIP6 models (black line with units of [K⋅decade−1]−1 corresponding to the y axis). Individual model ensemble averages are denoted by the short vertical lines along the x axis; the “X” denotes the multimodel average). The distributions of CESM2 and CESM2-SBB trends are shown with purple “violin” probability distributions (their vertical position on the y axis is arbitrary). The difference between the CESM2 and CESM2-SBB ensemble average trends represents the effect of BB aerosol forcing artifacts, and the magnitude of this difference is denoted with an arrow labeled “SBB sensitivity.” As in Fig. 3, we subtract the estimated impact of internal variability (−0.07 ± 0.07 K⋅decade−1) from the observed range of trends (purple shading), which has the effect of shifting the observations to the right (red shading). The magnitude of this shift is denoted with an arrow labeled “Internal variability effect.”