Fig. 7: Gaussian model for hot day probabilities.
From: Quantifying the role of variability in future intensification of heat extremes
a Theoretical probability changes obtained by integrating Eq. (2) (main text) over the +2K warming range and using multimodel grid-point results for the changes in the annual average TXm and the early-industrial standard deviation TXsdEI. b, c Theoretical changes calculated from Eq. (2) as above, but each of the two parameters, in turn, is held fixed throughout and equal to its area-weighted global average, to disentangle the relative contributions. Respectively, b TXm = < TXm > g and c TXsdEI = < TXsdEI > g. The fraction R2 of total spatial variation of projected changes in hot day probabilities (FCEP-99th, Fig. 6c) explained by the full Gaussian model F(TXm, TXsdEI) (a) amounts to ~0.9, and it reduces to ~ 0.5 if TXm = < TXm > g (b). Instead, if TXsdEI = < TXsdEI > g (c) R2 becomes negative (~ −0.9), meaning that if the total spatial variation of TXsdEI is not accounted for, the Gaussian model has worse predictions than the baseline global average of FCEP-99th.
