Figure 10. Median signal-to-noise ratio (SNR) for TXx and TNn
under SSP5-8.5 and RCP8.5 in the year 2050. (a) SNR for TXx under
SSP5-8.5 in the year 2050; (b) SNR for TNn under SSP5-8.5 in the year
2050; (c, d) same as (a) and (b), but for RCP8.5.
As for the temporal evolution of SNR (Figs. 8 and 9), in general, the
multi-model medians of SNR in TXx and TNn are slightly larger in
SSP5-8.5 than RCP8.5 (e.g., 3.22 under SSP5-8.5 and 2.77 under RCP8.5 in
2050 over AUS); while over some southern regions for TNn (e.g., SA, SSA,
SWA), the two CMIP ensemble show higher similarity. In addition, the
medians of signal and noise for the two indices are also comparable in
the two scenarios (Figs. S18-S20). It is noted that the differences in
signal between CMIP6 and CMIP5 in the end of the century resemble that
shown in Figs. 3a and 3d, which may further imply that the regional
climate sensitivity in CMIP6 and CMIP5 is comparable indicated in
previous studies (Palmer et al., 2021; Seneviratne & Hauser, 2020). In
terms of the inter-model spread, although the spreads of the signals for
TXx and TNn in SSP5-8.5 are commonly larger than RCP8.5, in which there
are more models showing stronger signal in SSP5-8.5 (Figs. S18 and S19),
the ranges in noise (Fig. S20) also contribute to the uncertainty of
SNR. Consequently, the relative magnitudes of SNR in SSP5-8.5 and RCP8.5
may change (e.g., Fig. 8a), compared to the signal (e.g., Fig. S18a).
For example, the spread of the signal in SSP5-8.5 is slightly larger in
the end of the century than RCP8.5; however, influenced by the noise,
the resulting range of SNR in SSP5-8.5 becomes narrower. Over the
regions, the ranges of SNR for TXx are usually narrower over southern
regions (e.g., SSA and SEA); in contrast, for TNn, northern regions such
as TA exhibit less uncertainty for SNR and TOE. In other scenarios
(Figs. S24-S27), the medians in SNR for TXx and TNn are lower, compared
to SSP5-8.5&RCP8.5; and the medians in SSPs are generally still higher
than that in RCPs. Also, the spreads of SNR and signal in the lower
forcing pathways is generally narrower, consistent with the time series
of projected changes.
To estimate the TOE for TXx and TNn, we use SNR > 1 and SNR
> 2 as the thresholds (Hawkins and Sutton 2012; Frame et
al. 2017; Hawkins et al. 2020) and present the spatial patterns for
multi-model median TOE under SSP5-8.5 and RCP8.5 (Fig. 11). As TOE
occurring at the end of the century may be a temporary change, which is
considered as “pseudo-emergence”, we exclude the TOE occurring after
the year 2050 (Abatzoglou et al., 2019; Diffenbaugh & Scherer, 2011;
Hawkins et al., 2014; King, Donat, et al., 2015).