Future changes in thermal comfort conditions over China based on multi-RegCM4 simulations

GAO Xu-Ji,WU Ji,SHI Ying,WU Jia,HAN Zhn-Yu,ZHANG Dong-Fng,TONG Yao,LI Rou-K,XU Yingan GIORGI Filippo

aClimate Change Research Center,Institute of Atmospheric Physics,Chinese Academy of Sciences,Beijing,China;bUniversity of Chinese Academy of Sciences,Beijing,China;cNational Climate Center,China Meteorological Administration,Beijing,China;dShanxi Climate Center,Taiyuan,China;eGaizhou Meteorological Bureau,Yingkou,China;fThe Abdus Salam International Center for Theoretical Physics,Trieste,Italy

ABSTRACT A set of high resolution(25 km)21st century climate change projections using the regional climate model RegCM4 driven by four global model simulations were conducted over East Asia under the mid-range RCP4.5 scenario.In the present paper,the authors investigate the change in thermal comfort conditions over china based on an ensemble of the projections,using the index of effective temperature(ET),which considers the aggregate effects of temperature,relative humidity,and wind on human thermal perception.The analysis also accounts for exposure as measured by distributed population amount scenarios.The authors find that the general increase in ET leads to a large increase in population exposure to very hot days(a China-aggregated sixfold increase in ‘person-days’by the end of the 21st century.There is a decrease in cool,cold,and very cold person-days.Meanwhile,a decrease in comfortable day conditions by 22%person-days is found despite an increase in climate-based comfortable days.Analysis of the different contributions to the changes(climate,population,and interactions between the two)show that climate effects play a more important role in the hot end of the thermal comfort categories,while the population effects tend to be dominant in the cold categories.Thus,overall,even a mid-level warming scenario is found to increase the thermal stress over China,although there is a strong geographical dependence.The inclusion of population exposure strongly modulates the climateonly signal,which highlights the need for including socioeconomic factors in the assessment of risks associated with climate change.

KEYWORDS Thermal comfort conditions;RegCM;climate change;population

1.Introduction

One of the main concerns related to global warming is the change in thermal conditions for the living environment of human beings.Both hot and cold extremes have been widely studied at the global and regional scale using different in dices(Collins et al.2013),such as the number of days in a year when maximum(minimum)temperature is above(below)a certain percentile(e.g.90%or 10%)of the present day distribution(Abatan et al.2016;Zhang et al.2011),spells of days with temperature above a threshold(heat waves),or long period return values(Xu et al.2018).Besides temperature,humidity is also an important factor influencing human comfort/discomfort,morbidity,and mortality,and thus it is often included in analyses of future heat stress(Diffenbaugh et al.2007;Fischer,Oleson,and Lawrence 2012;Zhao et al.2015;Kjellstrom et al.2016;Pal and Eltahir 2016;Im,Pal,and Eltahir 2017).Furthermore,population information can be included to account for human exposure to heat stress risk and loss of labor productivity(Jones et al.2015;Zhao et al.2016;Liu et al.2017).

However,while much was learned from previous studies on extreme thermal stress categories,little attention has been paid to less extreme conditions,e.g. mild and comfortable climates (Molenaar,Heusinkveld,and Steeneveld 2016;Perch-Nielsen,Amelung,and Knutti 2010;van der Wiel,Kapnick,and Vecchi 2017).In addition,wind effects on human thermal perception(Epstein and Moran 2006;Blazejczyk et al.2012)have been seldom considered.Effective temperature(ET),which considers the aggregate effects of temperature,relative humidity,and wind speed,can thus be a simple but effective indicator of thermal perception (McGregor1995;Liand Chan 2000;Blazejczyk et al.2012).Based on ET values,thermal perception is classified into different categories,ranging from ‘very hot’to ‘very cold’(Table 1).

With a population of 1.3 billion heterogeneously distributed across a vast territory,different climates,and large seasonal cycles,China can be especially vulnerable to changes in thermal conditions.By using the ET index,observed changes in thermal conditions over China were investigated for the most recent decades of the observed record(Wu et al.2017).It was found that China is characterized by a predominance of uncomfortable thermal conditions,mostly because of high numbers of cold days.Increased ET throughout the most recent decades has led to fewer cold days and more comfortable days in colder areas and seasons,but to an increase in heat stress during the summer in warmer regions.The study clearly illustrated the large spatial and seasonal variability of changes in thermal perception,a factor especially important because of the pronounced heterogeneity of population density throughout China.It is thus critical to assess how these variability factors can affect future changes in human thermal conditions throughout the country under increased concentrations of greenhouse gases.

Within this context,previous studies have shown the advantages of high resolution Regional Climate Models(RCMs)in generating climate projections over China and capturing the complex physiography and unique weather and climate systems of the region(Gao et al.2012).Recently,a new and unprecedented set of 21st century climate projections at 25 km grid spacing over East Asia were completed by using the RegCM4 RCM(Giorgi et al.2012)driven by four different General Circulation Models(GCMs)under RCP4.5,which lies towards the middle of the emissions scenario range.This ensemble thus offers an excellent opportunity to investigate the projected changes in thermal conditions over China by using the ET index.

Table 1.Thermal comfort categories based on effective temperature.

As already mentioned,in order to assess the effect of climate on people’s thermal comfort,it is also necessary to include exposure information.This is accomplished here by multiplying present day and future population amounts by the climate-based number of days in a thermal comfort category(resulting in units of ‘person-days’).

2.Model and methods

2.1.Model and experimental design

The climate projections analyzed here were completed with the Abdus Salam International Centre for Theoretical Physics RegCM4 (Giorgi et al. 2012) run over East Asia with model physical and parameter configurations identified in previous applications(Gao,Shi,and Giorgi2016;Gao et al.2017).The model was driven by four Coupled Model Intercomparison Project Phase 5(CMIP5)(Taylor,Stouer,and Meehl 2012)GCM simulations under the greenhouse gas concentration pathway RCP4.5(Moss et al.2010).The GCMs are CSIRO-Mk3.6.0,EC-EARTH,HadGEM2-ES,and MPI-ESM-MR(see Table S1),and were selected due to their higher resolution,good performances over the region(Jiang,Tian,and Lang 2016),as well as the data availability.The model domain follows the specifications of the International Coordinated Regional Climate Downscaling Experiment(CORDEX)(Giorgi,Jones,and Asrar 2009;Jones,Giorgi,and Asrar 2011;)at 25 km grid spacing,and the simulations cover the period 1979–2098,where 1981–2010 is considered as the reference period(present day)and 2069–98 as the end of the 21st century.

2.2.Bias correction and ET

The variables used to calculate the ET from the RegCM4 simulations,i.e.daily mean surface air temperature,relative humidity,and wind speed,were bias corrected using a quantile-mapping method(Gudmundsson et al.2012;Tong et al.2017).Specifically,for a given variable,the cumulative density function(CDF)of the daily values for the historical simulation in the reference period is first matched with the CDF of the observation dataset of CN05.1(Wu and Gao 2013),generating a transfer function depending on the quantile.Then,this transfer function is used to correct the variable from the future simulation on a quantile by quantile basis.

The ET was originally developed based on laboratory experiments(see Wu et al.2017 for more detail),and is calculated using the formula in the present study as(Landsberg 1972;Hentschel 1987):

where T is daily mean temperature(°C),RH is relative humidity(%),and v is wind speed(m s-1).The thermal comfort categories based on the ET values are provided in Table 1.

2.3.Population,person-days,and contributions from different factors

The population data are obtained from a scenario developed by the International Institute on Applied System Analysis(Riahi and Nakicenovic 2007)at a resolution of 0.5°× 0.5°(longitude × latitude),with the distribution of population for the present day shown in Figure S1(a).The future population associated with the B1 scenario was selected to match RCP4.5.The model and population data were bilinearly interpolated onto common 0.25°× 0.25°(longitude×latitude)grids.Note that these population data were also used to investigate future changes in heating and cooling degree days over China in a previous study(Shi et al.2016).The unit of ‘person-day’is used to measure the influence(exposure)of thermal conditions on population,which is calculated by multiplying the population in each grid box by the annual number of days in each ET category for the corresponding grid box.

The overall changes in person-days for a given category are divided into three contributions:climate,population,and interactive effects(Jones et al.2015;Liu et al.2017).The climate effect is measured by keeping the population fixed at the reference period level,while the population effect is measured by keeping climate fixed at the reference period level.The interactive effect is the difference between the total change and the climate and population effects.

3.Results

3.1.Present day

The mean annual present day ET shows values below the 17°C comfortable threshold dominating over the region,with cold or very cold conditions over most of western and northern China(Figure S1(b)).Figure 1(a)–(g)present the spatial distribution of the person-days in different thermal categories for the reference period(1981–2010)from the ensemble of simulations.The person-day amount for the very hot category is mostly limited to the region extending from the Yangtze River valley to North China,along with the southern coasts and the Sichuan Basin.The person-days for the hot category show a similar pattern,but with a larger spatial extent and higher amounts(10–50 × 106persondays).Lower person-day values are found for the warm category compared to the hot,while comfortable conditions occur mostly throughout eastern and northeastern China,with maximum values exceeding 25×106person-days mainly over North China and the Sichuan Basin.The person-days for the cool and cold categories are widespread and range from 25×106to a maximum of 50×106person-days over North China and the Sichuan Basin.Finally,the very cold category has a sharp south–north gradient,from<5 × 106along the southern coasts to>50×106person-days over North China.

The total number of person-days aggregated over the Chinese territory for the different thermal categories are summarized in Figure 1(h).The largest values are found for the very cold category(190×109,accounting for 30%of the total),followed by the cool(150×109)and cold(126×109)ones.The value for the comfortable category is also relatively large(78×109person-days or 12%of the total),while values for the hot and warm categories are 43×109and 37×109,respectively,and only 3×109person-days(less than 1%of the total)are found for the very hot category.It is thus clear that exposure to uncomfortably cold conditions dominates over most of China in present day climate,particularly in the central/eastern and northeastern regions.

Compared to the distribution of the number of days in different thermal categories without inclusion of population information(Figure S2),Figure 1 clearly shows that the uneven distribution of population over China strongly modulates the distribution of thermal comfort conditions.For example,while very cold conditions are dominant in the climate-only case,their prevalence decreases in the population-weighted case because of sparse population in the mountainous regions. This emphasizes the importance of including the population factor in climate impact studies,as reported by Shi et al.(2016).

Figure 1.Ensemble average person-days of different thermal comfort categories over China in present day conditions(1981–2010).(a–g)Spatial distribution of(a)very hot,(b)hot,(c)warm,(d)comfortable,(e)cool,(f)cold,and(g)very cold conditions(106 person-days;gray areas:values less than 0.1).(h)Aggregated value over China(109person-days).

3.2.Future changes

Concerning future conditions,population in China is projected to peak around the 2030s,and then decline from the present day value of 1.08×109to 0.81×109by the end of the century,representing a decline of 25%.However,this trend is not geographically homogeneous,with a redistribution of population consisting of a less pronounced decrease in more economically developed regions,and even an increase over North China and the Beijing area(Figure S3(a)and(c)).

A ubiquitous increase of ET is found in the ensemble simulations(Figure S3(b)),mostly caused by an increase in temperature and a decrease in wind speed,while relative humidity shows a decrease over eastern China(except for the coastal regions)and an increase over central and northern China( figures not shown for brevity).The regional mean increase of ET is slightly larger in December–January–February compared to June–July–August,and is consistent among the simulations,with the inter-model spread showing a tendency to increase in the latter half of the century(Figure S3(d)).

Figure 2.Projected changes in ensemble average person-days in different thermal comfort categories over China by the end of the 21st century,with respect to present day conditions(2069–98 minus 1981–2010).(a–g)Spatial distribution of(a)very hot,(b)hot,(c)warm,(d)comfortable,(e)cool,(f)cold,and(g)very cold conditions(106person-days;gray areas:values less than 0.1).(h)Aggregated value over China(109person-days).

Figure 3.Ensemble average amount of population subjected to different numbers of days per year in a given thermal comfort category for present day conditions(1981–2010,blue)and the end of the century(2069–98,red)(106persons).(a–g)Spatial distribution of(a)very hot,(b)hot,(c)warm,(d)comfortable,(e)cool,(f)cold,and(g)very cold conditions.The length of the parallel bars in(a)indicate the total population in the present day(1.08 billion)and at the end of the century(0.81 billion).The‘w’and ‘m’on the x-axis represent week and month,respectively.

Figure 2 shows the projected ensemble average change in person-days for each comfort category by the end of the 21st century (2069–98 minus 1981–2010),while Figure 3 presents the amount of people in the present and future time slices subjected to different numbers of days in a given thermal category.The pattern of increase in very hot person-days is similar to the distribution of this category in the present day(Figures 1(a)and 2(a)).Aggregated over the country,the number of person-days subjected to very hot conditions at the end of the century is 19.8×109,i.e.a six-fold increase compared to the present day value of～3.3×109(Figures 1(h)and 2(h)),despite the general decline in population.In particular,the population exposure to more than one and two months of very hot days increases from 3×106and 0 to 143×106and 23×106,respectively(Figure 3(a)).

The largest increase in hot person-days,10–25 × 106,is found in North China(Figure 2b),although the increase in climate-based hot days is not so large( figure not shown for brevity).A decrease in hot person-days is found in the Sichuan Basin and middle reaches of the Yangtze River following the population decline and the hot days moving to the very hot category.The aggregated change in hot person-days over China is still an increase(Figure 2(h)),with the population with less than one month of hot days dropping from 660×106to 270×106(～40%),and longer than four months of hot days increasing from 5×106to 16×106(approximately three-fold)(Figure 2(b)).

Figure 4.Temporal evolution of ensemble average person-days in different thermal comfort categories and contributions from climate,population,and interactive effects(109person-days).(a)Different thermal comfort categories;(b–h)contributions and the total change in(b)very hot,(c)hot,(d)warm,(e)comfortable,(f)cool,(g)cold,and(h)very cold conditions.

A decrease in warm and comfortable person-days is found in broad areas of southern China,except the southwest,where the increase in comfortable climate based days overwhelms the decline in population(Figure 2(c)and(d)).For the China-wide average,warm and comfortable person-days decrease by 19%and 22%,respectively(Figure 2(h)),despite an increase in climate-based comfortable days(～12%)(Figure S4(d)and(h)).This shows that fewer people will benefit from the improved thermal conditions.In fact,the population with over two months of comfortable days by the end of the century is projected to be 55%of the present day value(Figure 3(d)).

A general decrease in cool,cold and very cold person-days is found(Figure 2(e)–(g)),with the exception of an increase over the Tibetan Plateau and parts of North China for cool and cold,mostly due to a shift to warmer scales.Note that North China is a region with increases in both cold and hot person-days.Also,as shown in Figure 3(e)–(g),relatively large declines in population are found for long-lasting cold conditions.For example,the population with over four months of very cold days drops from 450×106to 210×106(～50%).Concerning the aggregated China-wide changes(Figure 2(h)),the largest signal is a decrease in very cold conditions,while comfortable conditions decrease by an amount similar in magnitude to the increase in very-hot conditions.In terms of persondays,very hot conditions show a large percentage change–about a factor of five.

3.3.Contributions from different factors

Figure 4 presents the temporal evolution of the Chinawide person-days in each category,along with the contributions from climate,population,and interactive effects to the overall changes.The person-days in the different categories increase until about the 2030s and then decline following the population trend,except for the very hot category,which shows a continuous increase until about the 2060s followed by a stabilization in the latter part of the century.We note that climate effects play a more important role in the hot end of the categories,while population effects tend to be dominant in the cold categories.To be more specific,the contributions of climate,population,and interactive effects at the end of the 21st century for very hot/hot are around130%/283%,-4%/-114%,and-26%/-69%,and for cold/very cold are 19%/43%,87%/69%,and-6%/-12%,respectively.The interactive component is in general small,as also found in a previous analysis of CMIP5 projections(Liu et al.2017).

4.Conclusions and discussion

Summarizing our results,although most of China is located in cold mid-and high latitudes,with many high mountains and plateaus,where climate warming will improve the thermal comfort conditions,most people live in warm and wet regions,which will undergo a substantial worsening of thermal stress.As a result,overall,although more climate-based comfortable conditions are projected throughout China,the persondays in the comfortable category are in fact projected to decrease.This indicates that the benefit from warming in cold climate regions throughout China will not be felt by large segments of the population.This result illustrates the relevance of the geographic and seasonal distribution of the ET changes,along with the importance of including exposure(measured here by the population amount)in analyses of impacts.

Several uncertainties are present in our study.First,the ensemble is based on only one RCM driven by multiple GCMs,and only one emissions and population scenario(the medium level RCM4.5 and B1,respectively).In fact,the high-end scenario(RCP8.5)would probably produce a larger increase in very hot stress conditions and a further decrease in comfortable conditions. The urban heat island effect is also not considered,which is particularly important over China,where many people live in large megacities.In addition,the index of ET is a relatively simple one(e.g.it does not consider radiative effects),and the thresholds for the different comfort categories should probably be adjusted over different regions.These uncertainties can be addressed in future work,especially within the context of the CORDEX framework(Giorgi,Jones,and Asrar 2009;Jones,Giorgi,and Asrar 2011),which will produce a new generation of high resolution projections over the region.However,despite these uncertainties,our work clearly indicates that the stress associated with climate warming can pose an important threat to future wellbeing in China.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This research was jointly supported by the National Key Research and Development Program of China (Grant No.2016YFA0600704) and the National Natural Science Foundation of China(Grant No.41375104).