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).