Limiting global warming to 1.5 °C will lower increases in inequalities of four hazard indicators of climate change

Clarifying characteristics of hazards and risks of climate change at 2 °C and 1.5 °C global warming is important for understanding the implications of the Paris Agreement. We perform and analyze large ensembles of 2 °C and 1.5 °C warming simulations. In the 2 °C runs, we find substantial increases in extreme hot days, heavy rainfalls, high streamflow and labor capacity reduction related to heat stress. For example, about half of the world’s population is projected to experience a present day 1-in-10 year hot day event every other year at 2 °C warming. The regions with relatively large increases of these four hazard indicators coincide with countries characterized by small CO2 emissions, low-income and high vulnerability. Limiting global warming to 1.5 °C, compared to 2 °C, is projected to lower increases in the four hazard indicators especially in those regions.


Introduction
The Paris Agreement sets a goal of 'holding the increase in the global average temperature to well below 2°C above pre-industrial levels and pursuing efforts to limit the temperature increase to 1.5°C above pre-industrial levels, recognizing that this would significantly reduce the risks and impacts of climate change' (United Nations Framework Convention on Climate Change (UNFCCC 2015). The most vulnerable and least developed countries have been calling for the 1.5°C limit for many years, to reduce risk of Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence.
Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. dangerous anthropogenic interference with the climate system (Tschakert 2015, Schleussner et al 2016a. Since the adoption of the agreement, a range of studies have examined how changes in extreme weather events and impacts would be lowered by limiting global warming to 1.5°C, compared to 2°C (Tschakert 2015, Rogelj and Knutti 2016, Schleussner et al 2016b, Mitchell et al 2016, Lewis et al 2017, Hoegh-Guldberg et al 2018. In the Paris Agreement, the issues of equity and climate justice are inherent in discussions of climate change adaptation, loss and damage, as well as mitigation (Okereke andCoventry 2016, Morgan andNorthrop 2017). The 1.5°C special report of the Intergovernmental Panel on Climate Change (IPCC) (Hoegh-Guldberg et al 2018) concluded that, with respect to the 'distribution of impacts', a transition from moderate to high risk is located between 1.5°C and 2°C of global warming. 'Risk of climate-related impacts results from the interaction of climate-related hazards with the vulnerability and exposure of human and natural systems' (Field et al 2014). Vulnerability and adaptive capacities are unevenly distributed. In general, the least developed countries in tropical and subtropical areas are among the most vulnerable (African Development Bank et al 2003, Field et al 2014, Department of Economic and Social Affairs of the United Nations Secretariat 2016). The ability to cope with the impacts of climate variability and extreme weather events depends strongly on the level of economic development and governance (Field et al 2014, Tschakert 2015. The differential vulnerability and adaptation capability is projected to lead to relatively larger impacts in the least developed countries that have emitted less CO 2 and have fewer financial resources (i.e. relatively little mitigation capacities) (Field et al 2014). There are also spatial heterogeneities in the future changes of extreme weather events (hazards). Some studies have suggested that the frequencies of hot days will increase more rapidly in the poorest countries in the tropics than in countries in mid-tohigh-latitude regions (Mahlstein et al 2011, Harrington et al 2016, Hoegh-Guldberg et al 2018. It has been suggested that 1.5°C-2°C differences of increases in heatwave exposure (Russo et al 2019), high streamflow (Döll et al 2018) and multi-sector (water, energy, food and environment) risks (Byers et al 2018) are larger for low-income countries/populations than high-income countries/populations.
The aim of this study is to investigate whether keeping global warming to 1.5°C, compared to 2°C, lowers changes in extreme weather events (extreme daily temperature and precipitation events), high streamflow and heat-related labor capacity reduction (LCR) from the present climate. We also examine if keeping global warming to 1.5°C limits the unequal distribution of these hazard indicators in terms of not only incomes of countries but also their responsibilities and vulnerabilities.

Global climate model simulations
The large variability of the four hazard indicators cause challenges in distinguishing 2°C from 1.5°C when using small ensembles of climate model simulations. Furthermore, the Coupled Model Intercomparison Project Phase 5 (CMIP5) (Taylor et al 2012) that contributed to the previous IPCC report was not designed to compare changes in climate at the 1.5°C and 2°C stabilization levels (Tschakert 2015, Rogelj andKnutti 2016). Although it is possible to extract anomalies of climate variables at 1.5°C and 2°C from the transient scenario experiments of CMIP5, those can be different from anomalies in stabilized 1.5°C and 2°C simulations (e.g. figure 1 of Mitchell et al 2016).
To overcome these challenges and inform policy dialogues, a multi atmosphere-land global climate model (AGCM) intercomparison project, the Half a degree Additional warming, Prognosis and Projected Impacts (HAPPI), was proposed (Mitchell et al 2016. In the HAPPI project, we have performed 10 year time-slice ensembles for the present (2006-2015, Hist), 1.5°C and 2°C warming climates relative to preindustrial levels using 6 AGCMs (supplementary table 1 is available online at stacks.iop.org/ ERL/14/124022/mmedia, supplementary methods). This design enables us to examine anomalies of climate variables of the 1.5°C and 2°C stabilized simulations. Furthermore lower computing costs of AGCM than fully coupled models make it easier to perform large ensembles. The ensemble sizes for each experiment range from 83 to 125. These sizeable ensembles enable us to robustly examine the effects of a 0.5°C difference in global warming on changes in the four hazard indicators.

Changes in the frequencies of extreme weather events
We investigate the annual warmest daily maximum temperatures (TXx) and the annual maximum consecutive 5 d precipitation (Rx5day) obtained from the HAPPI simulations (supplementary methods). We investigate the exposure ratios of land areas (which are important for identifying impacts on natural systems) and populations to extreme weather events (figure 2). Figure 2(a) indicates the fraction of the global land area where PR(TXx) exceeds a given threshold (shown on the horizontal axis). We omit the Antarctic region from this analysis because of its insignificant population. In the multi-AGCM average of the 2.0°C runs, PR(TXx) exceeds a factor of 5 (i.e.  frequencies of extreme events increase by factors of >5) over 39% of the global land area (the min-max range of the AGCMs is 31%-47%). In contrast, under the 1.5°C runs, the fraction of the land area where PR(TXx) exceeds a factor of 5 declines to 7% (3%-13%). The results for PR(Rx5day) are not as drastic as those for PR(TXx), but the changes in the frequency of the extreme rains are not negligible ( figure 2(b)). The fraction of the land area where PR(Rx5day) exceeds a factor of 1.5 is 37% (28%-46%) and 15% (9%-23%) for the 2.0°C and 1.5°C runs, respectively.

Labor capacity reduction
We examine heat-related LCR. To keep core body temperature within a safe range in hot environments, per-hour amount of physical activity must be limited (NIOSH 2016, ISO 2017, which means LCR. The future warming will increase LCR. If workers decrease their actual working time according to LCR, it leads to economic losses (Takakura et al 2017(Takakura et al , 2018. If they do not, the risk of heat-related hazards, some of which are fatal, will be elevated. LCR can be one of the dominant sources of the expected total economic loss caused by climate change amongst many other climate-induced effects (Takakura et al 2019). Thus, LCR would be a good indicator to gauge potential effects of high temperature on humans whereas this indicator alone does not consider the difference in vulnerability among regions. We estimate the annual LCR for outdoor workers with moderate physical activity (300 W) in the present, +1.5°C and +2.0°C climates (supplementary methods). Figures 3(a), (b) indicate changes in 1-in-10 year LCR (ΔLCR) for the 1.5°C and 2.0°C runs from Hist, respectively. ΔLCR is greater than 8% in some tropical countries (greater positive values indicate more reduction of labor capacity) in the 2.0°C runs, and the difference between the 1.5°C and 2.0°C runs (figures 3(c) and (a)) is largest in those tropical countries.

High streamflow
We also investigate annual highest 7 d streamflow (Q 7hf ), calculated by Döll et al (2018) using two hydrological models and the HAPPI runs of four AGCMs (supplementary methods). We investigate 1-in-10 year Q 7hf , which may lead to inundations. Human assets can be damaged by inundation, while floodplain habitat requires inundation. Figures 3(d), (e) show the relative changes in 1-in-10 year Q 7hf (ΔQ 7hf ) for the 1.5°C and 2.0°C runs compared to the Hist runs, respectively (supplementary methods). In the 2.0°C runs, increases in Q 7hf of more than 8% compared to the Hist runs occur in the many tropical countries while decreases of Q 7hf are found in some   at the same conclusion. We further indicate that, compared to the 2.0°C goal, meeting the 1.5°C goal both decreases PR(TXx) (the blue lines are lower than the red lines) and limits the increases of the inequalities: the amplitudes of the decreases in PR(TXx) are larger in lower-income regions with smaller emissions.

Uneven distributions of the four hazard indicators
Changes in LCR are greater in the lower-income regions with smaller emissions in 2.0°C (figures 5(a), (b)). The differences of ΔLCR between 1.5°C and 2.0°C are larger in those regions.
There are uneven distributions in PR(Rx5day) in the 2.0°C runs, but not so evident than those in PR(TXx) and ΔLCR (figures 4(c) and (d)). Although the solid red regression lines obtained using the 2.0°C simulations of all the AGCMs have negative slopes (i.e. apparent unequal distributions), some of the AGCMs have positive slopes. Nevertheless, one important finding holds: the reductions in PR(Rx5day) from a 2.0°C warming to a 1.5°C warming are greater in the regions with smaller emissions and lower-income.
Similar to PR(Rx5day), ΔQ 7hf does not decrease with increasing emissions/wealth in case of a 1.5°C world, while it clearly decreases in case of a 2°C world, where it would increase uneven distributions (note that the differences of the regression slopes between 1.5°C and 2.0°C are statistically significant) (figures 5(c), (d)).
We also use the University of Notre Dame Global Adaptation Initiative (ND-GAIN) Country Index (Chen et al 2015), which summarizes the vulnerability of countries to climate change and other global challenges (in the food, water, health, ecosystem services, human habitat and infrastructure sectors), combined with the readiness of countries to improve their resilience in 2015 (figures 6, 7). Lower values indicate greater vulnerability and lower readiness. There are obvious inequalities in the 2.0°C runs: countries with lower ND-GAIN values (i.e. more vulnerable countries) are projected to suffer greater increases in extreme hot days, heavy rains and LCR. The severity of these inequalities is lower for these three indices when global warming is limited to 1.5°C, rather than 2°C.
In the 2°C runs, 1-in-10 year Q 7hf largely increases in countries with low ND-GAIN values, and decreases in countries with high ND-GAIN values ( figure 7(b)). The slope of the 1.5°C runs is approximately zero. When we consider the human assets that can be damaged by inundation, the associated inequalities in the 1.5°C world are lower than those in the 2.0°C world, while that may not be the case for floodplain habitat requiring inundation.

Summary and discussion
By performing 2°C and 1.5°C warming runs of AGCMs, computing LCR and analyzing the outputs of the global hydrological model simulations, we show that keeping global warming to 1.5°C, rather than  2.0°C, lowers PR(TXx), PR(Rx5day), ΔLCR and ΔQ 7hf . Here we also examine the distributions of these four hazard indicators in relation to per capita CO 2 emissions, per capita GDP and ND-GAIN. There are uneven distributions: the regions with large increases of these four hazard indicators are characterized by small responsibility (small emissions/capita), low capability (low-income/capita) and high vulnerability (low-ND-GAIN). Keeping global warming to 1.5°C, rather than 2.0°C, limits these uneven distributions. King and Harrington (2018) indicated that the ratios of annual mean temperature differences (2.0°C minus 1.5°C) and the internal variability are larger in lower-income countries. Russo et al (2019) suggested that heatwave exposure and an illustrative heatwave risk index (the product of the probability of heatwave occurrence, exposure and a proxy for vulnerability) at the 1.5°C warming level for the population living in low development countries is expected to be greater than those at the 2°C warming level for the population living in very high development countries. Döll et al (2018) showed that the effect on ΔQ 7hf of meeting the 1.5°C goal rather than the 2.0°C would be felt more strongly in the low-income country groups than other country groups. Our results are consistent with these previous studies. Furthermore we suggest that meeting the 1.5°C goal limits the uneven distributions of the four hazard indicators in relation to not only GDP, but also CO 2 emissions and ND-GAIN values.
Our results are also consistent with Byers et al (2018) that analyzed distributions of a broader set of hazards and vulnerability indicators. They showed that global population exposure to multi-sector (water, energy, food and environment) risks approximately doubles between 1.5°C and 2°C warming. Large parts of global exposure to the multi-sector risks fall to Asian and African regions where the most of exposed and vulnerable population (income <$10/ day) exist. Our four hazard indicators (not examined by Byers et al 2018) also have the largest changes in countries with low-income as well as low-emission and high-vulnerability, providing more evidence that a 2°C warming is projected to increase the uneven distributions of hazard indicators compared to a 1. Tim Trautmann https://orcid.org/0000-0001-8652-6836 Petra Döll https://orcid.org/0000-0003-2238-4546 Sebastian Ostberg https:/ /orcid.org/0000-0002-2368-7015 Hannes Müller Schmied https:/ /orcid.org/0000-0001-5330-9923 Fahad Saeed https:/ /orcid.org/0000-0003-1899-9118