Editing IPCC:AR6/SR15/Chapter-3 (section)

==== 3.3.2.2 Projected changes in regional temperature means and extremes at 1.5°C versus 2°C of global warming ====

<div id="section-3-3-2-2-block-1"></div>

There are several lines of evidence available for providing a regional assessment of projected changes in temperature means and extremes at 1.5°C versus 2°C of global warming (see Section 3.2). These include: analyses of changes in extremes as a function of global warming based on existing climate simulations using the empirical scaling relationship (ESR) and variations thereof (e.g., Schleussner et al., 2017; Dosio and Fischer, 2018; Seneviratne et al., 2018c <sup>[[#fn:r65|65]]</sup> ; see Section 3.2 for details about the methodology); dedicated simulations of 1.5°C versus 2°C of global warming, for instance based on the Half a degree additional warming, prognosis and projected impacts (HAPPI) experiment (Mitchell et al., 2017) <sup>[[#fn:r66|66]]</sup> or other model simulations (e.g., Dosio et al., 2018; Kjellström et al., 2018) <sup>[[#fn:r67|67]]</sup> ; and analyses based on statistical pattern scaling approaches (e.g., Kharin et al., 2018) <sup>[[#fn:r68|68]]</sup> . These different lines of evidence lead to qualitatively consistent results regarding changes in temperature means and extremes at 1.5°C of global warming compared to the pre-industrial climate and 2°C of global warming.

There are statistically significant differences in temperature means and extremes at 1.5°C versus 2°C of global warming, both in the global average (Schleussner et al., 2016b; Dosio et al., 2018; Kharin et al., 2018) <sup>[[#fn:r69|69]]</sup> , as well as in most land regions ( ''high confidence'' ) (Wartenburger et al., 2017; Seneviratne et al., 2018c; Wehner et al., 2018b) <sup>[[#fn:r70|70]]</sup> . Projected temperatures over oceans display significant increases in means and extremes between 1.5°C and 2°C of global warming (Figures 3.3 and 3.4). A general background on the available evidence on regional changes in temperature means and extremes at 1.5°C versus 2°C of global warming is provided in the Supplementary Material 3.SM.2. As an example, Figure 3.5 shows regionally-based analyses for the IPCC SREX regions (see Figure 3.2) of changes in the temperature of hot extremes as a function of global warming (corresponding analyses for changes in the temperature of cold extremes are provided in the Supplementary Material 3.SM.2). As demonstrated in these analyses, the mean response of the intensity of temperature extremes in climate models to changes in the global mean temperature is approximately linear and independent of the considered emissions scenario (Seneviratne et al., 2016; Wartenburger et al., 2017) <sup>[[#fn:r71|71]]</sup> . Nonetheless, in the case of changes in the number of days exceeding a given threshold, changes are approximately exponential, with higher increases for rare events (Fischer and Knutti, 2015; Kharin et al., 2018) <sup>[[#fn:r72|72]]</sup> ; see also Figure 3.6. This behaviour is consistent with a linear increase in absolute temperature for extreme threshold exceedances (Whan et al., 2015) <sup>[[#fn:r73|73]]</sup> .

As mentioned in Section 3.3.1, there is an important land–sea warming contrast, with stronger warming on land (see also Christensen et al., 2013; Collins et al., 2013; Seneviratne et al., 2016) <sup>[[#fn:r74|74]]</sup> , which implies that regional warming on land is generally more than 1.5°C even when mean global warming is at 1.5°C. As highlighted in Seneviratne et al. (2016) <sup>[[#fn:r75|75]]</sup> , this feature is generally stronger for temperature extremes (Figures 3.4 and 3.5; Supplementary Material 3.SM.2 ). For differences in regional temperature extremes at a mean global warming of 1.5°C versus 2°C, that is, a difference of 0.5ºC in global warming, this implies differences of as much as 1°C–1.5°C in some locations, which are two to three times larger than the differences in global mean temperature. For hot extremes, the strongest warming is found in central and eastern North America, central and southern Europe, the Mediterranean, western and central Asia, and southern Africa (Figures 3.4 and 3.5) ( ''medium confidence'' ). These regions are all characterized by a strong soil-moisture–temperature coupling and projected increased dryness (Vogel et al., 2017) <sup>[[#fn:r76|76]]</sup> , which leads to a reduction in evaporative cooling in the projections. Some of these regions also show a wide range of responses to temperature extremes, in particular central Europe and central North America, owing to discrepancies in the representation of the underlying processes in current climate models (Vogel et al., 2017) <sup>[[#fn:r77|77]]</sup> . For mean temperature and cold extremes, the strongest warming is found in the northern high-latitude regions ( ''high confidence'' ). This is due to substantial ice-snow-albedo-temperature feedbacks (Figure 3.3 and Figure 3.4, middle) related to the known ‘polar amplification’ mechanism (e.g., IPCC, 2013; Masson-Delmotte et al., 2013) <sup>[[#fn:r78|78]]</sup> .

Figure 3.7 displays maps of changes in the number of hot days (NHD) at 1.5°C and 2°C of GMST increase. Maps of changes in the number of frost days (FD) can be found in Supplementary Material 3.SM.2. These analyses reveal clear patterns of changes between the two warming levels, which are consistent with analysed changes in heatwave occurrence (e.g., Dosio et al., 2018) <sup>[[#fn:r79|79]]</sup> . For the NHD, the largest differences are found in the tropics ( ''high confidence'' ), owing to the low interannual temperature variability there (Mahlstein et al., 2011) <sup>[[#fn:r80|80]]</sup> , although absolute changes in hot temperature extremes tended to be largest at mid-latitudes ( ''high confidence'' ) (Figures 3.4 and 3.5). Extreme heatwaves are thus projected to emerge earliest in the tropics and to become widespread in these regions already at 1.5°C of global warming ( ''high confidence'' ). These results are consistent with other recent assessments. Coumou and Robinson (2013) <sup>[[#fn:r81|81]]</sup> found that 20% of the global land area, centred in low-latitude regions, is projected to experience highly unusual monthly temperatures during Northern Hemisphere summers at 1.5°C of global warming, with this number nearly doubling at 2°C of global warming.

Figure 3.8 features an objective identification of ‘hotspots’ / key risks in temperature indices subdivided by region, based on the ESR approach applied to Coupled Model Intercomparison Project Phase 5 (CMIP5) simulations (Wartenburger et al., 2017) <sup>[[#fn:r82|82]]</sup> . Note that results based on the HAPPI multimodel experiment (Mitchell et al., 2017) <sup>[[#fn:r83|83]]</sup> are similar (Seneviratne et al., 2018c) <sup>[[#fn:r84|84]]</sup> . The considered regions follow the classification used in Figure 3.2 and also include the global land areas. Based on these analyses, the following can be stated: significant changes in responses are found in all regions for most temperature indices, with the exception of i) the diurnal temperature range (DTR) in most regions, ii) ice days (ID), frost days (FD) and growing season length (GSL) (mostly in regions where differences are zero, because, e.g., there are no ice or frost days), iii) the minimum yearly value of the maximum daily temperature (TXn) in very few regions. In terms of the sign of the changes, warm extremes display an increase in intensity, frequency and duration (e.g., an increase in the temperature of the hottest day of the year (TXx) in all regions, an increase in the proportion of days with a maximum temperature above the 90th percentile of Tmax (TX90p) in all regions, and an increase in the length of the WSDI in all regions), while cold extremes display a decrease in intensity, frequency and duration (e.g., an increase in the temperature of the coldest night of the year (TNn) in all regions, a decrease in the proportion of days with a minimum temperature below the 10th percentile of Tmin (TN10p), and a decrease in the cold spell duration index (CSDI) in all regions). Hence, while warm extremes are intensified, cold extremes become less intense in affected regions.

Overall, large increases in hot extremes occur in many densely inhabited regions (Figure 3.5), for both warming scenarios compared to pre-industrial and present-day climate, as well as for 2°C versus 1.5°C GMST warming. For instance, Dosio et al. (2018) <sup>[[#fn:r85|85]]</sup> concluded, based on a modelling study, that 13.8% of the world population would be exposed to ‘severe heatwaves’ at least once every 5 years under 1.5°C of global warming, with a threefold increase (36.9%) under 2°C of GMST warming, corresponding to a difference of about 1.7 billion people between the two global warming levels. They also concluded that limiting global warming to 1.5°C would result in about 420 million fewer people being frequently exposed to extreme heatwaves, and about 65 million fewer people being exposed to ‘exceptional heatwaves’ compared to conditions at 2ºC GMST warming. However, changes in vulnerability were not considered in their study. For this reason, we assess that there is ''medium confidence'' in their conclusions.

In summary, there is ''high confidence'' that there are robust and statistically significant differences in the projected temperature means and extremes at 1.5°C versus 2°C of global warming, both in the global average and in nearly all land regions <sup>[[#fn:6|6]]</sup> ( ''likely'' ). Further, the observational record reveals that substantial changes due to a 0.5°C GMST warming are apparent for indices related to hot and cold extremes, as well as for the WSDI ( ''likely'' ). A global warming of 2°C versus 1.5°C would lead to more frequent and more intense hot extremes in all land regions <sup>[[#fn:7|7]]</sup> , as well as longer warm spells, affecting many densely inhabited regions ( ''very likely'' ). The strongest increases in the frequency of hot extremes are projected for the rarest events (very likely). On the other hand, cold extremes would become less intense and less frequent, and cold spells would be shorter ( ''very likely'' ). Temperature extremes on land would generally increase more than the global average temperature ( ''very likely'' ). Temperature increases of extreme hot days in mid-latitudes are projected to be up to two times the increase in GMST, that is, 3ºC at 1.5ºC GMST warming ( ''high confidence'' ). The highest levels of warming for extreme hot days are expected to occur in central and eastern North America, central and southern Europe, the Mediterranean, western and central Asia, and southern Africa ( ''medium confidence'' ). These regions have a strong soil-moisture-temperature coupling in common as well as increased dryness and, consequently, a reduction in evaporative cooling. However, there is a substantial range in the representation of these processes in models, in particular in central Europe and central North America ( ''medium confidence'' ). The coldest nights in high latitudes warm by as much as 1.5°C for a 0.5°C increase in GMST, corresponding to a threefold stronger warming ( ''high confidence'' ). NHD shows the largest differences between 1.5°C and 2°C in the tropics, because of the low interannual temperature variability there ( ''high confidence'' ); extreme heatwaves are thus projected to emerge earliest in these regions, and they are expected to become widespread already at 1.5°C of global warming ( ''high confidence'' ). Limiting global warming to 1.5°C instead of 2°C could result in around 420 million fewer people being frequently exposed to extreme heatwaves, and about 65 million fewer people being exposed to exceptional heatwaves, assuming constant vulnerability ( ''medium confidence'' ).

<div id="section-3-3-2-2-block-2"></div>

<span id="figure-3.5"></span>
<!-- START IMG -->
<!-- IMG TITLE -->
'''Figure 3.5'''

<span id="projected-changes-in-annual-maximum-daytime-temperature-txx-as-a-function-of-global-warming-for-ipcc-special-report-on-managing-the-risk-of-extreme-events-and-disasters-to-advance-climate-change-adaptation-srex-regions-see-figure-3.2-based-on-an-empirical-scaling-relationship-applied-to-coupled-model-intercomparison-project-phase-5-cmip5-data-adapted-from-seneviratne-et-al.-2016-86-and-wartenburger-et-al.-2017-87-together-with-projected-changes-from-the-half-a-degree-additional-warming-prognosis-and-projected-impacts-happi-multimodel-experiment-mitchell-et-al.-2017-88-based-on-analyses-in-seneviratne-et-al.-2018c-89-bar-plots-on-regional-analyses-and-central-plot-respectively."></span>
<!-- IMG CAPTION -->
'''Projected changes in annual maximum daytime temperature (TXx) as a function of global warming for IPCC Special Report on Managing the Risk of Extreme Events and Disasters to Advance Climate Change Adaptation (SREX) regions (see Figure 3.2), based on an empirical scaling relationship applied to Coupled Model Intercomparison Project Phase 5 (CMIP5) data (adapted from Seneviratne et al., 2016 <sup>[[#fn:r86|86]]</sup> and Wartenburger et al., 2017) <sup>[[#fn:r87|87]]</sup> together with projected changes from the Half a degree additional warming, prognosis and projected impacts (HAPPI) multimodel experiment (Mitchell et al., 2017 <sup>[[#fn:r88|88]]</sup> ; based on analyses in Seneviratne et al., 2018c) <sup>[[#fn:r89|89]]</sup> (bar plots on regional analyses and central plot, respectively).'''
<!-- IMG FILE -->
[[File:9252f14e93b2d56d039676325a83d65b Figure_3_small-1024x717.jpg]]

For analyses for other regions from Figure 3.2 (with asterisks), see Supplementary Material 3.SM.2. (The stippling indicates significance of the differences in changes between 1.5°C and 2°C of global warming based on all model simulations, using a two-sided paired Wilcoxon test (P = 0.01, after controlling the false discovery rate according to Benjamini and Hochberg, 1995) <sup>[[#fn:r90|90]]</sup> . See Supplementary Material 3.SM.2 for details.

<!-- END IMG -->
<div id="section-3-3-2-2-block-3"></div>

<span id="figure-3.6"></span>
<!-- START IMG -->
<!-- IMG TITLE -->
'''Figure 3.6'''

<span id="probability-ratio-pr-of-exceeding-extreme-temperature-thresholds."></span>
<!-- IMG CAPTION -->
'''Probability ratio (PR) of exceeding extreme temperature thresholds.'''
<!-- IMG FILE -->
[[File:6baf6de9f8862a37c500c13b51df12df Figure_3.6-small-1024x576.jpg]]

(a) PR of exceeding the 99th (blue) and 99.9th (red) percentile of pre-industrial daily temperatures at a given warming level, averaged across land (from Fischer and Knutti, 2015) <sup>[[#fn:r91|91]]</sup> . (b) PR for the hottest daytime temperature of the year (TXx). (c) PR for the coldest night of the year (TNn) for different event probabilities (with RV indicating return values) in the current climate (1°C of global warming). Shading shows the interquartile (25–75%) range (from Kharin et al., 2018) <sup>[[#fn:r92|92]]</sup> .

<!-- END IMG -->
<div id="section-3-3-2-2-block-4"></div>

<span id="figure-3.7"></span>
<!-- START IMG -->
<!-- IMG TITLE -->
'''Figure 3.7'''

<span id="projected-changes-in-the-number-of-hot-days-nhd-10-warmest-days-at-1.5c-left-and-at-2c-middle-of-global-warming-compared-to-the-pre-industrial-period-18611880-and-the-difference-between-1.5c-and-2c-of-warming-right."></span>
<!-- IMG CAPTION -->
'''Projected changes in the number of hot days (NHD; 10% warmest days) at 1.5°C (left) and at 2°C (middle) of global warming compared to the pre-industrial period (1861–1880), and the difference between 1.5°C and 2°C of warming (right).'''
<!-- IMG FILE -->
[[File:4bf94f95131bf1bea5e0596fe9177107 Figure_3-small-1024x257.jpg]]

Cross-hatching highlights areas where at least two-thirds of the models agree on the sign of change as a measure of robustness (18 or more out of 26). The underlying methodology and the data basis are the same as for Figure 3.2 (see Supplementary Material 3.SM.2 for more details). Differences compared to 1°C global warming are provided in the Supplementary Material 3.SM.2.

Original Creation for this Report using CMIP5 multi-model ensemble output data.

<!-- END IMG -->
<div id="section-3-3-2-2-block-5"></div>

<span id="figure-3.8"></span>
<!-- START IMG -->
<!-- IMG TITLE -->
'''Figure 3.8'''

<span id="significance-of-differences-in-regional-mean-temperature-and-range-of-temperature-indices-between-the-1.5c-and-2c-global-mean-temperature-targets-rows."></span>
<!-- IMG CAPTION -->
'''Significance of differences in regional mean temperature and range of temperature indices between the 1.5°C and 2°C global mean temperature targets (rows).'''
<!-- IMG FILE -->
[[File:0c0469ed81921bcdf461eab1024baf70 Figure_3.8-new.jpg]]

Definitions of indices: T: mean temperature; CSDI: cold spell duration index; DTR: diurnal temperature range; FD: frost days; GSL: growing season length; ID: ice days; SU: summer days; TN10p: proportion of days with a minimum temperature (TN) lower than the 10th percentile of TN; TN90p: proportion of days with TN higher than the 90th percentile of TN; TNn: minimum yearly value of TN; TNx: maximum yearly value of TN; TR: tropical nights; TX10p: proportion of days with a maximum temperature (TX) lower than the 10th percentile of TX; TX90p: proportion of days with TX higher than the 90th percentile of TX; TXn: minimum yearly value of TX; TXx: maximum yearly value of TX; WSDI: warm spell duration index. Columns indicate analysed regions and global land (see Figure 3.2 for definitions). Significant differences are shown in red shading, with increases indicated with + and decreases indicated with –, while non-significant differences are shown in grey shading. White shading indicates when an index is the same at the two global warming levels (i.e., zero changes). Note that decreases in CSDI, FD, ID, TN10p and TX10p are linked to increased temperatures on cold days or nights. Significance was tested using a two-sided paired Wilcoxon test (P=0.01, after controlling the false discovery rate according to Benjamini and Hochberg, 1995) <sup>[[#fn:r93|93]]</sup> (adapted from Wartenburger et al., 2017) <sup>[[#fn:r94|94]]</sup> .

Original Creation for this Report using CMIP5 multi-model ensemble output data.

<!-- END IMG -->
<span id="regional-precipitation-including-heavy-precipitation-and-monsoons"></span>