Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

RESEARCH ARTICLE Extreme Wet‐Bulb Temperatures in China: The

10.1029/2019JD031477
Significant Role of Moisture
Key Points:
• Extreme wet‐bulb temperatures in
Pinya Wang1 , L. Ruby Leung2 , Jian Lu2 , Fengfei Song2 , and Jianping Tang1
China are investigated using 1
observations, considering both the
School of Atmospheric Sciences, Nanjing University, Nanjing, China, 2Atmospheric Sciences and Global Change
role of temperatures and moisture Division, Pacific Northwest National Laboratory, Richland, WA, USA
• Moisture contributes more to
extreme wet‐bulb temperatures than
temperature in most regions of Abstract This paper investigates the extreme wet‐bulb temperatures (TWs) in China, with a focus on
China
understanding the relative contributions of temperature and moisture to the extremes. Analysis of station
• The environment during moisture‐/
temperature‐dominated extreme observations shows that daily extreme temperatures (T), specific humidity (q), and TWs generally co‐occur in
wet‐bulb temperatures tends to be Southeastern China, while extreme TW and T rarely overlap in the arid and semiarid North and Northwest
convection favored/convection
China. Overall, q contributes more than T to extreme TWs, especially in North and Northwest China. Based on
inhibited.
the relative contributions of q and T, regional extreme TW events are classified as q dominated and T
dominated, respectively, to study their large‐scale environment. Cluster analysis of global reanalysis data
shows that extreme TWs are generally accompanied by increased surface air temperature and humidity,
Correspondence to:
L. R. Leung,
concomitant with anomalous high pressure and notable water vapor flux convergence. However, important
ruby.leung@pnnl.gov differences are also seen in the large‐scale environment during q‐dominated and T‐dominated extreme TWs.
During q‐dominated extreme TWs, the large‐scale environment favors convection, as indicated by ascending
Citation:
motions, decreased downward solar radiation as well as increased precipitable water and near‐surface
Wang, P., Leung, L. R., Lu, J., Song, F., relative humidity. In contrast, during T‐dominated extreme TWs, convection is inhibited by the large‐scale
& Tang, J. (2019). Extreme Wet‐Bulb environment that features descending motion and increased downward solar radiation, decreased precipitable
Temperatures in China: The Significant
Role of Moisture. Journal of Geophysical
water, and decreased relative humidity. Consistent with the contrasting environments, we demonstrate that
Research: Atmospheres, 124, T‐dominated extreme TWs tend to last longer than q‐dominated extremes. Given the significant role of moisture
11,944–11,960. https://doi.org/10.1029/ in extreme TWs, more research is needed to understand its impacts on heat stress now and in the future.
2019JD031477

Received 6 AUG 2019


Accepted 21 OCT 2019 1. Introduction
Accepted article online 25 OCT 2019
Published online 26 NOV 2019
With devastating impacts, extreme hot weather such as the 2003 European heat wave, the 2006 California
heat wave, and the 2010 Russian heat wave has raised significant societal concerns in the recent decades.
It is well accepted that high‐pressure systems are fundamental for the occurrence of anomalous high tem-
peratures (Della‐Marta et al., 2007; Freychet et al., 2017; Rohini et al., 2016; Xue et al., 2017). Other factors
such as drought also play an important role in prolonging extreme hot days through a positive feedback
between soil moisture and temperature (e.g., Fischer et al., 2007). Heat waves are also associated with remote
sea surface temperature anomalies that influence local circulation patterns and thus temperatures through
teleconnection mechanisms (i.e., Cassou et al., 2005; Huang & Li, 1989; Wang et al., 2017). Extreme heat has
been analyzed mostly based on a single variable, the extreme dry‐bulb temperatures.
Concurrent with anomalous temperatures, high humidity, stagnant winds, and air pollution can exacerbate
the adverse impacts of extreme heat on human discomfort and lead to enhanced mortality rates (Budd, 2008;
Conti et al., 2005; Fischer & Knutti, 2013). Under very humid and warm conditions, the human body has
difficulty in sweating. Accumulation of environmental and metabolic heat in the body may cause physical
disorders and even deaths (Budd, 2008). Hence, understanding humidity in addition to extreme tempera-
tures is of great significance for understanding the specific characteristics of humid heat stress.
Combining temperature and humidity, wet‐bulb temperature (TW) is an effective integrated measure for
characterizing extreme heat events. TW is the temperature a parcel of air would have if it is cooled adiaba-
tically to saturation at constant pressure by evaporating water into it through the latent heat supplied by the
parcel (AMS Glossary of Meteorology). Therefore, it is usually lower than the dry‐bulb temperature and
increases with the relative humidity. Raymond et al. (2017) investigated the spatiotemporal pattern and
synoptic conditions of extreme TW in the contiguous United States. Using 18 general circulation models,
©2019. American Geophysical Union. Coffel et al. (2017) projected the future changes of extreme TWs on a global scale under two representative
All Rights Reserved. concentration pathways. Focusing on the extreme TWs in North China Plain, Kang and Eltahir (2018) used

WANG ET AL. 11,944


21698996, 2019, 22, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2019JD031477 by Nat Prov Indonesia, Wiley Online Library on [14/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1029/2019JD031477

an ensemble of high‐resolution regional climate model simulations and projected deadly heat stress toward
the end of this century. Other integrated measure of humidity and temperature such as the wet‐bulb globe
temperature (WBGT, Yaglou & Minaed, 1957; Fischer & Knutti, 2013) and apparent temperature (AT; Russo
et al., 2017) has also been used in related studies. WBGT is more often used in health‐related studies because
there are established safety thresholds for WBGT (e.g., ISO, 1989; NIOSH, 1986). It has been suggested that
heat‐related morbidity and mortality and acute injury are higher with increasing WBGT (Alam et al., 2016;
Garzon‐Villalba et al., 2016). But WBGT is difficult to measure. In contrast, TW can be measured using stan-
dard method and is based on thermodynamic principle (Sherwood & Huber, 2010). Hence, TW has been
used more often in climate studies (e.g., Pal & Eltahir, 2016; Raymond et al., 2017).
In this study, we analyze the climatological characteristics of extreme TWs over China using observations
from meteorological stations and reanalysis data. Following Raymond et al. (2017), we disentangle the con-
tribution of q and T to extreme TWs and classify extreme TWs into q versus T dominance and investigate
their thermodynamical and dynamical environments. The remainder of this study is organized as follows:
section 3 describes the data and method. In section 3, we discuss the climatological characteristics of extreme
TWs in China and the dominant regional patterns of extreme TWs based on cluster analysis. We also inves-
tigate the synoptic characteristics of extreme TWs in each region. Section 4 summarizes and discusses the
key findings.

2. Data and Methodology


2.1. Data and the Calculation of TW
Following Coffel et al. (2017), daily maximum TW is calculated from the daily maximum air temperature
(T), daily mean relative humidity, and daily mean surface pressure (p) using the algorithm described in
Davies‐Jones (2008). Daily maximum specific humidity (q) is also derived from daily mean relative humidity
and p and daily maximum T. Buzan et al. (2015) implemented the algorithm in the Community Land Model,
which is then ported to MATLAB by Dr. Robert Kopp (available at http://www.bobkopp.net/software/code/
WetBulb.m). The daily data set of temperature, relative humidity, and surface pressure is provided by China
Meteorological Administration (http://data.cma.cn/en/) compiled from 1,710 observation stations across
China for the 1960–2015 period.
Daily data from the National Center for Environment Prediction/National Center for Atmospheric Research
reanalysis (Kalnay et al., 1996) are used to investigate the synoptic conditions during extreme TW days.
Specifically, the geopotential height at 500 hPa (H500) and meridional and zonal winds are used to describe
the circulation pattern. Specific humidity and winds at different pressure levels are used to estimate the
moisture transport, and surface pressure is used to account for the impacts of topography. The anomalies
of 2‐m air temperatures (T2m), 2‐m specific humidity, and near‐surface relative humidity (850 hPa) are
also analyzed.

2.2. Identification of Extreme TW and Regional Hot Days


We focus our analysis on the extreme TW in the warm season from 1 May to 30 September each year.
Extreme TW is calculated as the daily TW exceeding the local 90th percentile of the moving 21‐day centered
window for the specific calendar day during 1960–2015. Similar to Grotjahn et al. (2016), we use a relatively
low threshold of 90th percentile (compared to 95th or 99th percentile) to increase the sample size for analysis
to improve the statistical significance.
Extreme hot weathers are inherently regional as they affect an area rather than an individual station
(Gershunov et al., 2009; Ren et al., 2012; Wang et al., 2017). For instance, the 2003 and 2006 European heat
waves swept across a large region (e.g., Fischer et al., 2007; Chiriaco et al., 2014). Based on the method pro-
posed by Wang, Tang, Sun, et al. (2017), a regional hot day (RHD) is identified using three steps: (1) Two
neighboring sites of extreme TWs within a distance of 250 km are grouped into the same hot region, (2)
any sites of extreme TW outside of this hot region are grouped into other hot regions or not belong to any
hot regions, and (3) a RHD is identified if the hot region includes more than 20 sites of extreme TWs.
More details can be found in Wang, Tang, Sun, et al. (2017).

WANG ET AL. 11,945


21698996, 2019, 22, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2019JD031477 by Nat Prov Indonesia, Wiley Online Library on [14/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1029/2019JD031477

2.3. Cluster Analysis


Cluster analysis is a classical method used in atmospheric sciences to
characterize midlatitude weather (Cheng & Wallace, 1993; Mo &
Michael, 1988), including extreme events (Peng et al., 2011; Stefanon
et al., 2012; Wang, Tang, Sun, et al., 2017). In this study, we use the
hierarchical cluster analysis algorithm to obtain the dominant patterns
of RHDs in China and investigate their synoptic characteristics.
Clustering is conducted based on daily maps of the TW exceedance over
the threshold for all the RHDs during 1960–2015. TW exceedance is the
deviation of TW above the threshold of extreme TWs. An optimal cluster
must have a large intercluster dissimilarity and a small intracluster dis-
similarity. And the clusters must be reproducible (Cheng & Wallace,
1993; Wang, Tang, Sun, et al., 2017). The specific clustering procedures
are described in Wang, Tang, Sun, et al. (2017). The synoptic conditions
for each cluster are calculated through compositing the meterological
Figure 1. Spatial distribution of the averaged daily maximum TW (°C) dur- fields of the associated extreme RHDs.
ing May to September over the period 1960–2015. TW = wet‐bulb
temperature.
3. Results
3.1. Climatological Characteristics of Extreme TWs in China
Figure 1 shows the spatial pattern of averaged TWs in the boreal summer
monsoon season of May–September during 1960–2015. TW has lower values over Northern China but
higher values over Southern China. Note that the averaged TW and T (not shown) have similar spatial pat-
terns except in Northwestern China, where T is relatively high but q is the lowest, reflecting the arid and
semiarid climate (Peng & Zhou, 2017; Zhang et al., 2016), with a potential for large differences between
extreme dry temperatures and extreme TWs.
The average annual occurrence (days), annual maximum duration (days), and annual maximum magni-
tude (°C) of extreme TWs are shown in Figure 2. For each site, the annual maximum duration is the
annual maximum number of consecutive days of extreme TWs, and the annual maximum magnitude
is the annual maximum exceedance of extreme TWs. Extreme TWs show high occurrences in southern-
most China (Figure 2a), while extreme TWs generally last longer over Southeastern and Northeastern
China (Figure 2b). Note that if a minimum duration such as 3 days is required in the definition of
extreme TWs, the highest occurrences of extreme TWs would be located in Southeastern China, coincid-
ing with the region of longer lasting extreme TWs. Differences between the duration of extreme TWs and
the duration of extreme dry‐bulb temperatures are notable in the arid and semiarid Northern China
where the durations of extreme dry high temperatures are comparable to those in Southern China
and longer than those in Northeastern China (not shown), while extreme TWs are of shorter durations
than the extreme TWs in other regions. For the annual maximum magnitude of extreme TWs, there is a
positive gradient stretching from Southern China to Northern China (Figure 2c). The large maximum
magnitude in Northern China may be related to the low climatological average of TW in that
region (Figure 1).
3.2. Excursions of T and q to Extreme TWs
Anomalies of TW may be related to anomalies of either or both of its components (q and T). Here, we disen-
tangle the contributions of T and q to extreme TWs in China following a similar approach from Raymond et al.
(2017) using a standard‐anomaly ratio (SAR). However, our definition of SAR emphasizes the exceedance
above the extreme rather than the mean. Four cases can be defined based on the value of SAR equals to:

R1=R2 R1>0 & R2>0 (1)

0 R1≤0 & R2>0 (2)

maximum R1>0 & R2≤0 (3)

1 R1≤0 & R2≤0 (4)

WANG ET AL. 11,946


21698996, 2019, 22, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2019JD031477 by Nat Prov Indonesia, Wiley Online Library on [14/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1029/2019JD031477

where R1 and R2 are the normalized exceedance defined as R1 = [(T −


Tc)/Ts] and R2 = [(q − qc)/qs]. Tc and qc are the daily thresholds for
extreme T and q, similarly defined following the definition of threshold
for the extreme TW. Ts and qs are the standard deviations of the time
series of the 21‐day centered moving window for T and q during
1960–2015. In equation (3), maximum refers to the maximum value of
SAR in equation (1) across the whole China. We define the SAR ratio
as motivated by the fact that during extreme TW days, neither the local
T nor the local q is required to be extreme. A larger (smaller) SAR
implies a larger T (q) excursion, so T (q) has a relatively greater contri-
bution to the extreme TW. Thus, an extreme TW day can be classified as
T dominated (SAR > 1) or q dominated (SAR < 1) based on its SAR
value. Also, a single SAR value can be used to characterize all possible
cases: Both T and q are extreme (Case 1); q is extreme but T is not (Case
2); T is extreme but q is not (Case 3), and neither T nor q is extreme
(Case 4). Similar to the definition of extreme TWs, extreme q (T) indi-
cates q (T) > qc (Tc), and the thresholds are defined based on the 90th
percentile values.
Figure 3a shows the spatial pattern of the median values of SAR during
1960–2015 and the percentage of extreme TWs in the four cases
(Figures 3b–3e). At most stations, the median values of SAR are low
(median of SAR < 1), indicating that the extreme TWs are q dominated
in most of China. Extreme T and extreme q usually co‐occur during
extreme TW days (Case 1), especially in Southern and Northeastern
China (Figure 3b), but in the arid and semiarid Northwest and North
China, Case 2 with extreme q but nonextreme T occurs more often
(Figure 3c). Thus, previous studies focusing only on the dry‐bulb tempera-
tures in China will miss most of the TW extremes in Northern and
Northwestern China. Across China, extreme TWs in Case 3 are much less
frequent compared to Cases 1 and 2. Extreme TWs in Case 4 are negligi-
ble, suggesting that TW extremes are primarily favored by atmospheric
conditions that are either both very humid and very hot or very humid
but not very hot. In other words, atmospheric humidity is important for
TW extremes.

To further elucidate the relationship among T, q, and TW in China,


the correlation between summer daily T and q, as well as the differ-
ence between the standardized partial regression coefficients of sum-
mer daily TW on q and T are illustrated in Figure 4. Based on
Figure 2. Average annual number of days (a, days), annual maximum dura- multiple regression analysis, the partial correlation coefficients repre-
tion (b, days), and annual maximum magnitude (c, °C) of extreme wet‐bulb sent the rate of change in the dependent variable for a unit change
temperatures (TWs) during May to September over the period 1960–2015.
in the independent variable, under the constraint that all other inde-
For each site, the annual maximum duration is the annual maximum
number of consecutive days of extreme TWs, and the annual maximum pendent variables are held constant. The standardized regression coef-
magnitude is the annual maximum exceedance of extreme TWs during May ficients can be directly compared with each other to determine the
to September over 1960–2015. TW exceedance is the deviation of extreme most effective variables (Frossard et al., 2013; Goulden, 1939; Knibbe
TW above the threshold. et al., 2014; Wespes et al., 2017). Daily T and q have a strong linear
relationship over most of China during summertime, especially in
Southern and Northeastern China, consistent with the higher co‐occurrences of extreme T and q in
those regions (Figure 3b). The strong T and q relationship reflects the important role of moisture
supply. That is, with unlimited supply of moisture, the moisture of an air mass would scale with the
temperature following the Clausius‐Clapeyron relation. In contrast, T and q have a weak relationship
over arid and semiarid regions in North and Northwest China, such as over most of the Xinjiang
Province, as q is limited by moisture from local evapotranspiration over land. From Figure 4b, q
explains a larger fraction of the variance of annual mean TW than T over most of China, especially over

WANG ET AL. 11,947


21698996, 2019, 22, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2019JD031477 by Nat Prov Indonesia, Wiley Online Library on [14/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1029/2019JD031477

Figure 3. Spatial distributions of the median values of SAR during 1960–2015 (a) and the percentage (%) of extreme TWs belonging to the four catogories defined
based on the SAR value (b–e). SAR = standard‐anomaly ratio; TW = wet‐bulb temperature.

the arid and semiarid North and Northwest China. Therefore, the more frequent q‐dominated TW
extremes over the arid and semiarid regions (Figure 3c) are driven by the strong dependence of TW
on q in those areas.

WANG ET AL. 11,948


21698996, 2019, 22, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2019JD031477 by Nat Prov Indonesia, Wiley Online Library on [14/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1029/2019JD031477

Figure 4. Spatial distributions of linear correlation between daily maximum T and q (a) over May to September during 1960–2015 and the deviations between the
standardized multiregression coefficient of daily maximum TW onto q and T (b). The seasonal cycle of daily maximum T, q, and TW has been removed. Scatterplot
of the relationship between extreme TWs and the local T and q in the (c) SE (20–27°N, 105–125°E) and (d) NW (35–50°N, 75–90°E) regions. The colored
contour lines denote the constant TWs at the standard atmospheric pressure (1,013 hPa) as a function of T and q. Green dots denote the averaged TWs along the
constant TWs. The slope of the lines connecting two adjacent green dots indicates the sensitivity of extreme TWs to T and q. The steeper the slope, the more q
dominated the extreme TWs are. TW = wet‐bulb temperature.

Raymond et al. (2017) have argued that the baseline climate is a strong predictor of the q/T dominance of the
local extremes. That is, in a stationary climate, q dominance (T dominance) is more likely to occur over
regions with climatologically high T(q), especially when the climatological q (T) is concurrently low. To
explore this concept in China, Figures 4c and 4d show scatterplots of extreme TWs as a function of the local
T and q in Southeastern (SE) China (20–27°N, 105–125°E) and Northwestern (NW) China (35–50°N, 75–90°E),
respectively. Extreme TWs in the SE region have higher specific humidity than those in the NW region.
Nearly all the local q of extreme TWs in the SE region is higher than 20 g/kg, and it can even reach
40 g/kg (Figure 4c), consistent with the climatologically humid conditions of the region. In contrast, extreme
TWs in the NW have a much wider range of local q and local T. For extreme TWs in the NW region, the excee-
dance of TWs tends to be dominated by the increase of T when both T and q are low (q < 15 g/kg and T < 30 °C).
In contrast, the exceedance of TWs is more sensitive to q when T is higher than 30 °C with q mostly lower than
25 g/kg, so in the relatively hot and dry regime, q plays an increasing role for extreme TWs (Figure 4d). And
compared with extreme TWs in the SE region, q dominance occurs more often for extreme TWs in the NW
region partly because the climatological q is low so even low q values can exceed the q threshold and lead to
extreme TWs.
3.3. Synoptic Environments of Extreme TWs
So far, we have focused on the relative role of T and q in extreme TWs in China. As the daily T and q are
strongly influenced by the large‐scale circulation, we address the modulation of the q/T dominance by the
large‐scale circulation on synoptic time scale. To facilitate analysis of the synoptic environments of

WANG ET AL. 11,949


21698996, 2019, 22, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2019JD031477 by Nat Prov Indonesia, Wiley Online Library on [14/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1029/2019JD031477

Figure 5. (a–i) The spatial patterns of nine clusters of regional hot days. Shading is the average daily TW exceedance (°C). TW = wet‐bulb temperature.

extreme TWs, we first classify extreme TWs based on RHDs using cluster analysis to focus on extreme TWs at
regional scale. Then through composite analysis, we identify the synoptic patterns of the dominant RHDs
and investigate how the synoptic patterns modulate the q/T dominance in the RHD clusters. These
analyses provide a view on the dynamical control of extreme TWs to complement the thermodynamics
view discussed in section 3.2.
3.3.1. Classification of RHDs
In our definition, an RHD consists of several hot sites of extreme TWs; each has a SAR value. An RHD is
classified as q dominated if the number of hot stations with SAR < 1 is higher than the number of hot sta-
tions with SAR ≥ 1. Conversely, an RHD is classified as T dominated if the number of hot stations with
SAR values >1 is higher than the number of hot stations with SAR ≤ 1. With this classification, 66% of
RHDs are categorized as q dominated and 31% of RHDs are categorized as T dominated. Together, they
account for 97% of the total RHDs.
3.3.2. Synoptic Environments During RHDs for Different Regions
Using cluster analysis, we obtain nine reproducible clusters (Cheng & Wallace, 1993; Wang, Tang, Sun,
et al., 2017). Their spatial patterns are shown in Figure 5. Shading indicates the average TW exceedance
of all RHDs associated with each cluster. A green box is used to define the territory of extreme TWs in each
cluster. The nine clusters roughly cover the whole China. Consistent with the spatial distribution of annual
maximum magnitude of extreme TWs (Figure 2c), the TW exceedance shows higher values in the northern
part of China. For instance, Clusters 4, 5, and 9 (Figures 5d, 5e, and 5i) located north of the Yellow River
basin show higher TW exceedance than Clusters 2, 6, and 7 (Figures 5b, 5f, and 5g) in the southern part.

WANG ET AL. 11,950


21698996, 2019, 22, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2019JD031477 by Nat Prov Indonesia, Wiley Online Library on [14/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1029/2019JD031477

Table 1 With the nine clusters of RHDs identified, we investigate the synoptic con-
Percentages of q‐Dominated and T‐Dominated RHDs in Each Cluster ditions related to extreme TWs through compositing of the atmospheric
Clusters conditions during RHDs in each cluster. The fraction of q‐dominated
Classifications C1 C2 C3 C4 C5 C6 C7 C8 C9 and T‐dominated RHDs in each cluster is shown in Table 1. The classifica-
tion results are consistent with the spatial pattern of SAR values in
q dominated 71% 53% 53% 82% 77% 67% 66% 61% 65%
T dominated 28 % 45% 45% 16% 21% 33% 30% 38% 34%
Figure 3. T‐dominated RHDs are more located south of the Yellow River
(Clusters 2 and 3), where T‐dominated and q‐dominated RHDs have simi-
Note. The top two q‐dominated and T‐dominated percentages are shown
lar frequency of occurrence (45% versus 53%, Table 1). The RHDs in arid
in bold font. RHD = regional hot day.
and semiarid Northern China are more q dominated, especially in
Clusters 4 and 5, which account for 82% and 77% of the total RHDs within
those clusters, respectively (Table 1). We focus on the synoptic characteristics of Clusters 4 and 5 in contrast
with those of Clusters 2 and 3 to shed some lights on the different characteristics of the synoptic environ-
ments between q‐dominated and T‐dominated extreme TWs.
The composites of H500 and T2m during RHDs in each cluster are shown in Figure 6. For each cluster, posi-
tive T2m anomalies are evident and generally colocated with high H500 aloft. Both anomalies of H500 and
T2m are stronger in the northern part than the southern part of China, consistent with the stronger TW
exceedance in the northern part of China (Figure 5). However, the T2m and H500 anomalies are not always
cocentered, but instead, notable geographic separations are evident in some clusters. For example, the cen-
ters of increased T2m in Clusters 4 and 5 are located on the western flank of the anomalous H500 while the
center of increased H500 is located directly above the center of increased T2m in Cluster 8. Increased solar
radiation under clear sky associated with anomalous high‐pressure systems should be favorable for
increased surface air temperature (Rohini et al., 2016). These processes should result in cocentered anoma-
lies of H500 and T2m such as those depicted in Cluster 8 (Figure 6h).

Figure 6. (a–i) The spatial pattern of averaged anomalies (deviation to the monthly mean) of H500 (contour, m) and T2m (shading, °C) during the regional hot
days. Stippling indicates statistically significant T2m anomalies at p < 0.05.

WANG ET AL. 11,951


21698996, 2019, 22, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2019JD031477 by Nat Prov Indonesia, Wiley Online Library on [14/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1029/2019JD031477

2
Figure 7. (a–i) The spatial pattern of averaged anomalies of DSR (contour, W/m ) and W500 (shading, Pa/s) during the regional hot days (deviation from the
monthly mean). Stippling indicates statistical significant W500 anomalies at p < 0.05. DSR = downward solar radiation.

Figure 7 shows anomalies of vertical velocity at 500 hPa (W500) and downward solar radiation (DSR) dur-
ing extreme TWs in each cluster. Large differences in the averaged anomalies of W500 and DSR are
obvious among the clusters. While some clusters are dominated by descending motion and increased
DSR (Clusters 2 and 3 apparently, Figures 7b and 7c), which help to sustain the anomalous high tempera-
tures during RHDs, other clusters are dominated by ascending motion that favors cloud formation, consis-
tent with the reduced DSR colocated with the W500 anomalies (Clusters 4 and 5, Figures 7d and 7e). To
understand these differences, the averaged precipitable water anomalies during RHDs is further shown in
Figure 8. We can see reduced precipitable water in Clusters 2 and 3 as well as Cluster 6 (Figures 8b, 8c, and
8f), but the precipitable water is apparently increased in other clusters, especially Clusters 4 and 5
(Figures 8d and 8e). Considering the regional difference of q/T dominance in Clusters 2/3 and Clusters
4/5, we hypothesize that the synoptic environment for the q‐dominated RHDs favors convection while
the synoptic environment for the T‐dominated RHDs inhibits convection. More quantitatively, we
hypothesize that the strength of convection scales with the degree of q dominance of the RHDs.
Figure 9 tests this hypothesis using a scatterplot that compares the percentage of q‐dominated hot sites
(SAR < 1) of each RHD in all the clusters and the associated areal mean W500 anomalies within the ter-
ritories of RHDs. There is a significant decreasing slope in the areal mean W500 anomalies with the
increasing percentage of q‐dominated sites, lending support to our hypothesis. The relationship between
humid heat and convection has rarely been discussed, although Coffel et al. (2017) noted that a
convection‐inhibiting environment may be needed to achieve high TWs. Such condition is consistent with
the higher TWs over Southern China that are less q dominated (i.e., weaker convection). Our analysis sug-
gests that extreme TWs can also happen under convection‐favored environment, but such events do not
last as long as extreme TW events under strong subsidence and convective inhibition. Further research
on the relationship between humid heat and convection in other regions can provide important insights
on the generality of humid heat under conditions favorable for convection. In particular, in some regions

WANG ET AL. 11,952


21698996, 2019, 22, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2019JD031477 by Nat Prov Indonesia, Wiley Online Library on [14/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1029/2019JD031477

2
Figure 8. (a–i) The spatial pattern of averaged anomalies of RH at 850 hPa (contour, %) and PW (shading, kg/m ) during the regional hot days. Stippling indicates
statistically significant PW anomalies at p < 0.05. RH = relative humidity; PW = precipitable water.

(e.g., Clusters 1 and 9), the territory of extreme TWs is situated on the border between anomalous upward
and downward motions (Figures7a and 7i). Such displacement between the center of extreme TWs and the
center of convection/subsidence needs future investigations.

Figure 9. Scatterplot and the regression line (pink line) between the percentage of q‐dominated hot sites (SAR < 1) in the
RHDs of all clusters and the associated areal mean W500 anomalies. The zero line is shown as the black dashed line. RHDs
= regional hot days.

WANG ET AL. 11,953


21698996, 2019, 22, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2019JD031477 by Nat Prov Indonesia, Wiley Online Library on [14/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1029/2019JD031477

−1 −1
Figure 10. (a–i) The spatial pattern of averaged anomalies of vertically integrated moisture flux (vector, kg·m ·s ) and surface‐specific humidity (shading, g/kg).
−2 −1
The magenta dashed contours indicate moisture convergence (g·m ·s ). Stippling indicates statistical significant specific humidity at p < 0.05.

Composite anomalies of the vertically integrated moisture flux and moisture transport convergence and
surface‐specific humidity (shading) are depicted in Figure 10. Specific humidity increases are seen
within and in the vicinity of all clusters, and the anomalies are generally colocated with the moisture
flux convergence. The moisture flux convergence is stronger in Clusters 4 and 5 (Figures 10d and
10e) where the q‐T relationship is weaker than the other clusters (Figure 4a), indicating that the humid-
ity increases during extreme TWs may be primarily attributed to the water vapor transport. In other
words, Clusters 4 and 5 with the T2m anomaly centers located on the western flank of the anomalous
high‐pressure systems receive anomalous moisture supply by southerly moisture transport induced by
the high pressure to promote RHDs. Previous research suggests that heat stress may result from the
interaction of hot desert air masses with moisture advection from warm bodies of water (Pal &
Eltahir, 2016). Byrne and O'Gorman (2018) quantified the importance of atmospheric dynamics and
moisture advection in controlling continental humidity, which may also have implications for changes
in humid heat. Because the humidity increase is more related to moisture transport rather than local
evaporation, the q anomalies are less related to and can be larger than the increase in saturation specific
humidity due to the T anomalies following the Clausius‐Clapeyron relationship, causing an increase
in the near‐surface relative humidity (Figures 8d and 8e). Generally, clusters with larger q anomalies
(e.g., Clusters 1, 4, 5) are located on the western flank of the high‐pressure anomalies and receive large
anomalous moisture transport. In contrast, clusters with smaller q anomalies (e.g., Clusters 2, 3, and 9)
have their centers more colocated with the high‐pressure anomalies and receive only anomalous moist-
ure transport near the fringes of the cluster boundaries. For Clusters 2 and 3, the negative anomalies of
relative humidity at 850 hPa (Figures 8b and 8c) associated with small positive anomalies of surface‐
specific humidity and large T2m anomalies are consistent with the nearly equal T and q dominance
to RHDs in these areas.

WANG ET AL. 11,954


21698996, 2019, 22, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2019JD031477 by Nat Prov Indonesia, Wiley Online Library on [14/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1029/2019JD031477

Table 2 In summary, extreme TWs in China are generally accompanied by both


Percentages of Extreme Cases of q‐Dominated and T‐Dominated RHDs in anomalous high temperature and humidity, associated with or dominated
Each Cluster
by high‐pressure systems. However, clusters with more q‐dominated
Clusters RHDs are characterized by a more convection‐favored environment and
Classifications C1 C2 C3 C4 C5 C6 C7 C8 C9 clusters with less q dominated RHDs are characterized by more
q dominated 19% 14% 16% 65% 31% 18% 17% 23% 30%
convection‐inhibited environment. The latter resembles the environmen-
T dominated 2% 12% 25% 6% 2% 4% 3% 9% 10% tal conditions during extreme warm dry‐bulb temperatures. And the
higher the q dominance, the stronger the convection.
Note. The top q‐dominated and T‐dominated percentages corresponding
to Cluster 4 and Cluster 3, respectively, are shown in bold font.
3.4. Synoptic Environments for q‐ Versus T‐Dominated Extreme
TWs: A Canonical View
To further lend support to the above findings, we compare the synoptic environments between the extreme
cases of q‐dominated RHDs and the extreme cases of T‐dominated RHDs. The extreme cases of q‐dominated
RHDs mean that most sites have extreme q but nonextreme T (i.e., a plurality of sites in the region fall into
Category 2). In contrast, most sites in the extreme cases of T‐dominated RHDs have extreme T but nonex-
treme q (i.e., a plurality of sites in the region fall into Category 3).
Table 2 shows the percentage of the extreme cases of q‐/T‐dominated RHDs in each cluster. Extreme cases of
T/q dominated RHDs generally account for less than 50% of the total RHDs in each cluster, except for
Cluster 4 of which almost 65% is extreme cases of q‐dominated RHDs. The results are consistent with the
above finding that most TW extremes are accompanied by both anomalous high T and q. Moreover, com-
pared to the other clusters, Cluster 4 has relatively more extreme cases of q‐dominated RHDs, whereas
Cluster 3 has relatively more extreme cases of T‐dominated RHDs.
The composite anomalies of T2m, surface humidity, integrated water vapor transport, water vapor transport
convergence, and low‐level relative humidity for the extreme cases of q‐dominated RHDs in Cluster 4 and
the extreme cases of T‐dominated RHDs in Cluster 3 are compared in Figure 11, respectively. For clusters
with q‐dominated TWs, the regions of extreme TWs are dominated by increased T and adjacent to an anom-
alous high‐pressure system (Figure 11a). Ascending motions with decreased DSR (Figure 11c) and increased
precipitable water (Figure 11e) are seen, and surface‐specific humidity shows apparent increase over areas
where TW increases, accompanied by notable water vapor transport (Figure 11g). Specific humidity
increases are stronger than the T increases, so the near‐surface relative humidity is increased (Figure 11e).
The strong low‐level wind convergence and high relative humidity favor convection, consistent with the
reduced solar radiation. For the extreme cases of T‐dominated TWs, the regions of extreme RHDs are also
accompanied by increased T and high‐pressure systems (Figure 11b). However, descending motions with
increased DSR (Figure 11d) and precipitable water (Figure 11f) are notable, which benefit the maintenance
of high temperatures. Moreover, surface‐specific humidity shows weak increases (Figure 11h) and the rela-
tive humidity is decreased because of the large increases in T (Figure 11f).
The above results confirm that the atmospheric environment during q‐dominated RHDs tends to favor con-
vection, while the environment during T‐dominated extreme RHDs tends to inhibit convection. The latter
resembles the synoptic conditions during extreme warm dry‐bulb temperatures. The T‐dominated extreme
TWs may last longer than the q‐dominated ones because increased solar radiation during T‐dominated
extreme can amplify the surface warming, while convection during q‐dominated extreme can cool the sur-
face through downdraft that brings the cooler air aloft to the surface and causes evaporative cooling of the
raindrops, as well as cloudiness that reduces solar radiation. Although the high humidity of q‐dominated
TW extremes may trap heat more effectively to maintain the temperature anomalies, as a fast process, con-
vection may have larger effects in limiting the duration of q‐dominated TW extremes. This is confirmed by
Figure 12 showing the daily evolution of the T2m and TW anomalies normalized by their values on the onset
dates of the extreme cases of q‐dominated RHDs in Cluster 4 and the T‐dominated RHDs in Cluster 3. The
TW/T anomalies are averaged over the regions of the RHDs. The TW anomalies generally last longer than
the T2m anomalies. This suggests that the covariance of T and q anomalies may persist over time to maintain
the TW anomalies more than the T anomalies alone. Interestingly, for both TW and T2m, the anomalies last
longer for T‐dominated TW extremes than q‐dominated TW extremes. The difference in the anomaly time
scales of T‐ versus q‐dominated TW extremes is also supported by the climatology of the maximum duration
of TW extremes (Figure 2b), which shows longer duration in Southern China (relatively less q dominated)

WANG ET AL. 11,955


21698996, 2019, 22, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2019JD031477 by Nat Prov Indonesia, Wiley Online Library on [14/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1029/2019JD031477

Figure 11. The spatial pattern of averaged anomalies of atmospheric variables during the extreme cases of q‐dominated RHDs in Cluster 4 (a, c, e, and g) and the
extreme cases of T‐dominated RHDs in Cluster 3 (b, d, f, and h). (a, b) The spatial pattern of averaged anomalies of H500 (m, contour) and T2m (°C, shading)
2 2
during the RHDs. (c, d) Same as (a)–(c) but for W500 (Pa/s, shading) and DSR (W/m , contour). (e, f) Same as (a)–(c) but for PW (kg/m , shading) and relative humidity
−1 −1
(contour). (g)–(h) The spatial pattern of averaged anomalies of vertically integrated moisture flux (kg·m ·s , vector) and surface‐specific humidity (shading).
−2 −1
The green dashed contours indicate moisture convergence (g·m ·s ). DSR = downward solar radiation; RH = relative humidity; PW = precipitable water.

WANG ET AL. 11,956


21698996, 2019, 22, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2019JD031477 by Nat Prov Indonesia, Wiley Online Library on [14/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1029/2019JD031477

Figure 12. Daily evolution of the ratio of averaged TW (a) and T2m (b) anomalies following the specific dates of the
extreme cases of q‐dominated regonal hot days (RHDs) in Cluster 4 and the T‐dominated RHDs in Cluster 3. The TW
and T2m anomalies are averaged over the territories of RHDs. TW = wet‐bulb temperature.

but shorter duration in Northwest and Northern China (relatively more q dominated). The underlying
processes associated with T‐ and q‐dominated extremes also help explain the remarkable difference between
the duration of extreme high dry‐bulb temperatures and extreme TWs in North and Northwest China. As
mentioned in section 3.1, extreme dry‐bulb temperatures have long durations in North and Northwest
China where T‐q relationship is weak, which are comparable to the long durations of high temperatures
in Southern China and longer than those in Northeastern China. These results suggest that q‐dominated
TW extremes may happen less often in subtropical regions than desert regions. Hence, humid heat events
in the subtropics may last longer, allowing the intensity to strengthen over time and influence more of the
world's population than heat events in the desert regions.

WANG ET AL. 11,957


21698996, 2019, 22, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2019JD031477 by Nat Prov Indonesia, Wiley Online Library on [14/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1029/2019JD031477

4. Conclusions and Discussions


In this study, extreme high TWs in China during 1960–2015 are investigated, taking account of both dry‐bulb
temperatures and moisture. Daily maximum TW is calculated based on the observed daily maximum
temperature and daily mean relative humidity from 1,710 sites in China. Thermodynamic and dynamic
variables from the National Center for Environment Prediction/National Center for Atmospheric
Research reanalysis data are used to characterize the synoptic environments during extreme TWs.
Analysis of the exceedances of q/T during extreme TWs highlights the significant role of moisture in extreme
TWs. The main conclusions are summarized as follows:
• Extreme TWs show high occurrences in Southern China, and they are longer lasting over Southeastern
and Northeastern China than those in the arid and semiarid Northern China. Moreover, the magnitude
of extreme TWs is higher over the northern part of China than those in the southern part, which may
be related to the low climatology of TW in the former region.
• Anomalous T, q, and TW generally co‐occur in Southeastern China, while extreme TW and T overlap less
often in the arid and semiarid North and Northwest China. Extreme TWs in North and Northwest China
are mostly accompanied by extreme q but nonextreme T. Therefore, focusing on only high dry‐bulb
temperatures will miss most of extreme TWs in Northwest and North China.
• Extreme TWs are classified into q dominated and T dominated based on the exceedance of q and T. In
general, q shows higher and lower dominance in the arid and semiarid regions and the Southern and
Northeastern China, respectively.
• Areas of extreme TWs are generally accompanied by extreme T2m and influenced by anomalous high‐
pressure systems. Moreover, q is also increased during extreme TWs, accompanied by obvious water vapor
flux convergence.
• Notable differences are seen in the atmospheric environment during q‐dominated extreme TWs and
T‐dominated extreme TWs. That is, the atmospheric environment during q‐dominated extreme TWs
tends to favor convection with ascending motions, decreased DSR, and increased precipitable water as
well as relative humidity. In contrast, the atmospheric environment tends to inhibit convection during
T‐dominated extreme TWs with descending motion and increased DSR, decreased precipitable water,
and decrease relative humidity, which resembles the synoptic conditions during extreme warm dry‐
bulb temperatures.
T‐dominated extreme TWs are likely to last longer than q‐dominated extreme TWs. The maximum durations
of extreme TWs show longer duration in Southeastern and Southern China than those in the arid and semi-
arid regions where extreme TWs are mainly q dominated.
Given the significant role of moisture in extreme TWs, more research is needed to understand its impacts on
extreme heat events now and in the future. Based on AT, Russo et al. (2017) revealed that the magnitudes of
observed heat waves, such as the 1995 Chicago heat wave and the 2003 China heat wave, were strongly
amplified by humidity. Anthropogenic activities such as urbanization and irrigation can modify the thermo-
dynamic environment and influence extreme heat events. Fischer et al. (2012) simulated contrasting urban
and rural heat stress based on WBGT, driven strongly by the urban‐rural relative humidity contrast. Local
irrigation was found to contribute to the deadly high TWs projected for the North China Plain (Kang &
Eltahir, 2018). Current projections of extreme heat events are mainly based on projected changes of dry‐bulb
temperature (e.g., Dosio et al., 2018; Guirguis et al., 2017; Wang et al., 2019). Understanding how moisture
will change in the future is also important to evaluate future changes in heat stress. Sherwood et al. (2010)
projected the relative humidity changes in a warming climate and concluded that on global scales and over
open water bodies, the near‐surface humidity of air roughly follows the temperatures according to the
Clausius‐Clapeyron relationship. Thus, relative humidity will remain roughly constant with warming,
which implies an increasing trend of TWs. By middle to late century, extreme TWs will potentially reach
and, in some cases, exceed the postulated theoretical limits of human tolerance and affect more than half
of the world population (Coffel et al., 2017).
Intensified by humidity, AT peaks in highly populated regions can be greater than 55 °C with global
warming at 1.5 and 2.0 °C, breaking the present climate record (Russo et al., 2017). Li et al. (2018)
argued that AT has increased faster than dry‐bulb temperature based on both reanalysis data and model
simulations. Our analysis suggests that the mean TW in the warm season shows a greater increase than

WANG ET AL. 11,958


21698996, 2019, 22, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2019JD031477 by Nat Prov Indonesia, Wiley Online Library on [14/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1029/2019JD031477

the dry‐bulb temperatures over the North China Plain over the last five decades (not shown).
Considering the weak relation between T and q over arid and semiarid regions such as the northwes-
tern part of China and the western United States, how humidity will change over such regions due
to changes in water vapor transport and local processes deserves further investigations. The East
Asian summer monsoon plays an important role in water supply in China (Jin et al., 2013; Zhou &
Yu, 2005). Correlated with a southward retreat of the East Asian summer monsoon, aridity over the
semiarid regions in Northern China shows a drying trend in the recent decades (Zhang et al., 2016).
This suggests the potential for a larger role of moisture changes in extreme heat events in the recent
decades, motivating a need to investigate how summer monsoon changes in the future may impact
heat extremes.

Acknowledgments References
This study is supported by the U.S.
Department of Energy Office of Science Alam, M., Sanjayan, J., Zou, P. X., Stewart, M. G., & Wilson, J. (2016). Modelling the correlation between building energy ratings and heat‐
Biological and Environmental Research related mortality and morbidity. Sustainable cities and society, 22, 29–39.
through the Regional and Global Budd, G. M. (2008). Wet‐bulb globe temperature (WBGT)—Its history and its limitations. Journal of Science and Medicine in Sport, 11(1),
Modeling and Analysis program. 20–32.
P. Wang is jointly supported by the Buzan, J., Oleson, K., & Huber, M. (2015). Implementation and comparison of a suite of heat stress metrics within the Community Land
National Key Research and Model version 4.5. Geoscientific Model Development, 8(2), 151–170.
Development Program of China Byrne, M. P., & O'Gorman, P. A. (2018). Trends in continentaltemperature and humidity directly linked to ocean warming. Proceedings of
(2018YFC1505803 and the National Academy of Sciences, 115, 4863–4868.
2018YFA0606003), the National Cassou, C., Terray, L., & Phillips, A. S. (2005). Tropical Atlantic influence on European heat waves. Journal of climate, 18(15), 2805–2811.
Natural Science Foundation of China Cheng, X., & Wallace, J. M. (1993). Cluster analysis of the Northern Hemisphere wintertime 500‐hPa height field: Spatial patterns. Journal
(41875124), and the China Scholarship of the Atmospheric Sciences, 50(16), 2674–2696.
Council during her visit at the Pacific Chiriaco, M., Bastin, S., Yiou, P., Haeffelin, M., Dupont, J. C., & Stéfanon, M. (2014). European heatwave in July 2006: Observations and
Northwest National Laboratory modeling showing how local processes amplify conducive large‐scale conditions. Geophysical Research Letters, 41, 5644–5652. https://
(PNNL) in Richland, WA. PNNL is doi.org/10.1002/2014GL060205
operated for the Department of Energy Coffel, E. D., Horton, R. M., & de Sherbinin, A. (2017). Temperature and humidity based projections of a rapid rise in global heat stress
by Battelle Memorial Institute under exposure during the 21st century. Environmental Research Letters, 13(1).
contract DE‐AC05‐76RL01830. In this Conti, S., Meli, P., Minelli, G., Solimini, R., Toccaceli, V., Vichi, M., et al. (2005). Epidemiologic study of mortality during the Summer 2003
study, the daily station records of tem- heat wave in Italy. Environmental Research, 98(3), 390–399. https://doi.org/10.1016/j.envres.2004.10.009
perature, specific humidity, and surface Davies‐Jones, R. (2008). An efficient and accurate method for computing the wet‐bulb temperature along pseudoadiabats. Monthly Weather
pressure are taken from the China Review, 136(7), 2764–2785.
Meteorological Data Center (http:// Della‐Marta, P. M., Luterbacher, J., von Weissenfluh, H., Xoplaki, E., Brunet, M., & Wanner, H. (2007). Summer heat waves over western
data.cma.cn/en/). NCEP Reanalysis Europe 1880–2003, their relationship to large‐scale forcings and predictability. Climate Dynamics, 29(2‐3), 251–275.
data are provided by the NOAA/OAR/ Dosio, A., Mentaschi, L., Fischer, E. M., & Wyser, K. (2018). Extreme heat waves under 1.5 °C and 2 °C global warming. Environmental
ESRL PSD, Boulder, Colorado, USA, Research Letters, 13(5). https://doi.org/10.1088/1748‐9326/aab827
from their website (https://www.esrl. Fischer, E. M., & Knutti, R. (2013). Robust projections of combined humidity and temperature extremes. Nature Climate Change, 3(2), 126.
noaa.gov/psd/). Fischer, E. M., Seneviratne, S., Vidale, P., Lüthi, D., & Schär, C. (2007). Soil moisture‐atmosphere interactions during the 2003 European
summer heat wave. Journal of Climate, 20(20), 5081–5099.
Fischer, E. M., Oleson, K. W., & Lawrence, D. M. (2012). Contrasting urban and rural heat stress responses to climate change. Geophysical
Research Letters, 39, L03705. https://doi.org/10.1029/2011GL050576
Freychet, N., Tett, S., Wang, J., & Hegerl, G. (2017). Summer heat waves over Eastern China: Dynamical processes and trend attribution.
Environmental Research Letters, 12(2).
Frossard, L., Rieder, H. E., Ribatet, M., Staehelin, J., Maeder, J. A., Di Rocco, S., et al. (2013). On the relationship between total ozone and
atmospheric dynamics and chemistry at mid‐latitudes—Part 1: Statistical models and spatial fingerprints of atmospheric dynamics and
chemistry. Atmospheric Chemistry and Physics, 13, 147–164. https://doi.org/10.5194/acp‐13‐147‐2013
Garzon‐Villalba, X. P., Mbah, A., Wu, Y., Hiles, M., Moore, H., Schwartz, S. W., & Bernard, T. E. (2016). Exertional heat illness and acute
injury related to ambient wet bulb globe temperature. American journal of industrial medicine, 59(12), 1169–1176. https://doi.org/
10.1002/ajim.22650
Gershunov, A., Cayan, D. R., & Iacobellis, S. F. (2009). The great 2006 heat wave over California and Nevada: Signal of an increasing trend.
Journal of Climate, 22(23), 6181–6203.
Goulden, C. H. (1939). Methods of statistical analysis. Oxford, England: Wiley.
Grotjahn, R., Black, R., Leung, R., Wehner, M. F., Barlow, M., Bosilovich, M., et al. (2016). North American extreme temperature events
and related large scale meteorological patterns: A review of statistical methods, dynamics, modeling, and trends. Climate Dynamics, 46
(3‐4), 1151–1184.
Guirguis, K., Gershunov, A., Cayan, D. R., & Pierce, D. W. (2017). Heat wave probability in the changing climate of the Southwest US.
Climate Dynamics, 1–12.
Huang, R., & Li, L. (1989). Numerical simulation of the relationship between the anomaly of subtropical high over East Asia and the
convective activities in the western tropical Pacific. Advances in Atmospheric Sciences, 6(2), 202–214.
ISO. Hot environments—Estimation of the heat stress on working man, based on the WBGT‐index (wet bulb globe temperature). ISO
Standard 7243. Geneva: International Standards Organization. 1989
Jin, Q., Yang, X. Q., Sun, X. G., & Fang, J. B. (2013). East Asian summer monsoon circulation structure controlled by feedback of con-
densational heating. Climate Dynamics, 41, 1885–1897.
Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D., Gandin, L., et al. (1996). The NCEP/NCAR 40‐year reanalysis project.
Bulletin of the American meteorological Society, 77(3), 437–472.
Kang, S., & Eltahir, E. A. (2018). North China Plain threatened by deadly heatwaves due to climate change and irrigation. Nature
Communications, 9(1), 2894.

WANG ET AL. 11,959


21698996, 2019, 22, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2019JD031477 by Nat Prov Indonesia, Wiley Online Library on [14/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1029/2019JD031477

Knibbe, J. S., VanderA, R. J., & de Laat, A. T. J. (2014). Spatial regression analysis on 32 years of total column ozone data. Atmospheric
Chemistry and Physics, 14, 8461–8482.
Li, J., Chen, Y. D., Gan, T. Y., & Lau, N. C. (2018). Elevated increases in human‐perceived temperature under climate warming. Nature
Climate Change, 8(1), 43.
Mo, K., & Michael, G. (1988). Cluster analysis of multiple planetary flow regimes. Journal of Geophysical Research, 93(D9), 10,927–10,952.
National Institute for Occupational Safety and Health 1986. Occupational exposure to hot environments. Department of Health and
Human Services, Washington, DC: Report No. DHHS86‐113, Washington, DC.
Pal, J. S., & Eltahir, E. A. (2016). Future temperature in southwest Asia projected to exceed a threshold for human adaptability. Nature
Climate Change, 6(2), 197.
Peng, D., & Zhou, T. (2017). Why was the arid and semiarid northwest China getting wetter in the recent decades? Journal of Geophysical
Research: Atmospheres, 122, 9060–9075. https://doi.org/10.1002/2016JD026424
Peng, J. B., & Bueh, C. (2011). The definition and classification of extensive and persistent extreme cold events in China. Atmospheric and
Oceanic Science Letters, 4(5), 281–286
Raymond, C., Singh, D., & Horton, R. (2017). Spatiotemporal patterns and synoptics of extreme wet‐bulb temperature in the contiguous
United States. Journal of Geophysical Research: Atmospheres, 122, 13,108–13,124. https://doi.org/10.1002/2017JD027140
Ren, F., Cui, D., Gong, Z., Wang, Y., Zou, X., Li, Y., et al. (2012). An objective identification technique for regional extreme events. Journal
of Climate, 25(20), 7015–7027.
Rohini, P., Rajeevan, M., & Srivastava, A. (2016). On the variability and increasing trends of heat waves over India. Scientific reports, 6.
Russo, S., Sillmann, J., & Sterl, A. (2017). Humid heat waves at different warming levels. Scientific reports, 7(1), 7477. https://doi.org/
10.1038/s41598‐017‐07536‐7
Sherwood, S. C., & Huber, M. (2010). An adaptability limit to climate change due to heat stress. Proceedings of the National Academy of
Sciences, 107(21), 9552–9555.
Sherwood, S. C., Ingram, W., Tsushima, Y., Satoh, M., Roberts, M., Vidale, P. L., & O'Gorman, P. A. (2010). Relative humidity changes in a
warmer climate. Journal of Geophysical Research, 115, D09104. https://doi.org/10.1029/2009JD012585
Stefanon, M., D'Andrea, F., & Drobinski, P. (2012). Heatwave classification over Europe and the Mediterranean region. Environmental
Research Letters, 7(1), 14,023–14,031(14029).
Wang, P., Tang, J., Wang, S., Dong, X., & Fang, J. (2017). Regional heatwaves in China: A cluster analysis. Climate Dynamics, 1‐17.
Wang, P., Tang, J., Sun, X., Wang, S., Wu, J., Dong, X., & Fang, J. (2017). Heatwaves in China: Definitions, leading patterns and connections
to large‐scale atmospheric circulation and SSTs. :Journal of Geophysical Research Atmospheres, 122, 10,679–10,699. https://doi.org/
10.1002/2017JD027180
Wang, P., Hui, P., Xue, D., & Tang, J. (2019). Future projection of heat waves over China under global warming within the CORDEX‐EA‐II
project. Climate Dynamics, 53(1‐2), 957–973. https://doi.org/10.1007/s00382‐019‐04621‐7
Wespes, C., Hurtmans, D., Clerbaux, C., & Coheur, P. F. (2017). O3 variability in the troposphere as observed by IASI over 2008–2016:
Contribution of atmospheric chemistry and dynamics. Journal of Geophysical Research: Atmospheres, 122, 2429–2451. https://doi.org/
10.1002/2016JD025875
Xue, D., Lu, J., Sun, L., Chen, G., & Zhang, Y. (2017). Local increase of anticyclonic wave activity over northern Eurasia under amplified
Arctic warming. Geophysical Research Letters, 44, 3299–3308. https://doi.org/10.1002/2017GL072649
Yaglou, C. P., & Minaed, D. (1957). Control of heat casualties at military training centers. Arch Indust Health, 16(4), 302–316.
Zhang, H., Zhang, Q., Yue, P., Zhang, L., Liu, Q., Qiao, S., & Yan, P. (2016). Aridity over a semiarid zone in northern China and responses to
the East Asian summer monsoon. Journal of Geophysical Research: Atmospheres, 121, 13,901–13,918. https://doi.org/10.1002/
2016JD025261
Zhou, T. J., & Yu, R. C. (2005). Atmospheric water vapor transport associated with typical anomalous summer rainfall patterns in China.
Journal of Geophysical Research Atmospheres, 110, D08104. https://doi.org/10.1029/2004JD005413

WANG ET AL. 11,960

You might also like