Content uploaded by Yali Luo
Author content
All content in this area was uploaded by Yali Luo on Sep 19, 2019
Content may be subject to copyright.
Observed Link of Extreme Hourly Precipitation Changes to Urbanization
over Coastal South China
MENGWEN WU
State Key Laboratory of Severe Weather, Chinese Academy of Meteorological Sciences, Beijing, and Institute of Meteorological
Sciences, Zhejiang Meteorological Bureau, Hangzhou, and University of Chinese Academy of Sciences, Beijing, China
YALI LUO
State Key Laboratory of Severe Weather, Chinese Academy of Meteorological Sciences, Beijing, and Collaborative
Innovation Center on Forecast and Evaluation of Meteorological Disasters, Nanjing University of Information
Science and Technology, Nanjing, China
FEI CHEN
State Key Laboratory of Severe Weather, Chinese Academy of Meteorological Sciences, Beijing, China, and
National Center for Atmospheric Research, Boulder, Colorado
WAI KIN WONG
Hong Kong Observatory, Hong Kong, China
(Manuscript received 22 October 2018, in final form 7 April 2019)
ABSTRACT
Understanding changes in subdaily rainfall extremes is critical to urban planners for building more sus-
tainable and resilient cities. In this study, the hourly precipitation data in 1971–2016 from 61 rain gauges are
combined with historical land-use change data to investigate changes in extreme hourly precipitation (EXHP)
in the Pearl River delta (PRD) region of South China. Also, 120 extreme rainfall events (EXREs) during
2011–16 are analyzed using observations collected at densely distributed automatic weather stations and radar
network. Statistically significant increase of hourly precipitation intensity leads to higher annual amounts of both total
and extreme precipitation over the PRD urban cluster in the rapid urbanization period (about 1994–2016) than
during the preurbanization era (1971 to about 1993), suggesting a possible link between the enhanced rainfall and the
rapid urbanization. Those urbanization-related positive trends are closely related to more frequent occurrence of
abrupt rainfall events with short duration (#6 h) than the continuous or growing rainfall events with longer duration.
The 120 EXREs in 2011–16 are categorized into six types according to the originating location and movement of the
extreme-rain-producing storms. Despite the wide range of synoptic backgrounds and seasons, rainfall intensification
by the strong urban heat island (UHI) effect is a clear signal in all the six types, especially over the inland urban cluster
with prominent UHIs. The UHI thermal perturbation probably plays an important role in the convective initiation
and intensification of the locally developed extreme-rain-producing storms during the daytime.
1. Introduction
More than half of the global population now resides in
urban areas (Grimm et al. 2008), and that number is
expected to increase to 60% by 2030 and 70% by 2050
(UN Department of Economic and Social Affairs,
Population Division 2011). Urban areas, especially
coastal cities, are vulnerable to heavy rainfall-induced
flood exposure that is increasing in the changing climate
Denotes content that is immediately available upon publica-
tion as open access.
Corresponding author: Dr. Yali Luo, ylluo@cma.gov.cn
Publisher’s Note: This article was revised on 14 August 2019 to
correct the first author’s affiliations, which were not correctly
presented when originally published.
AUGUST 2019 W U E T A L . 1799
DOI: 10.1175/JAMC-D-18-0284.1
Ó2019 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright
Policy (www.ametsoc.org/PUBSReuseLicenses).
(Hallegatte et al. 2013). The majority of previous studies
suggested that daily extreme rainfall has increased in
more regions than it has decreased in a warming world
(Minetal.2011;Seneviratne et al. 2012;Westra et al.
2013;Donatetal.2016;Seneviratne et al. 2016). How-
ever, engineering practices and urban infrastructure de-
sign demand information about changes in subdaily (e.g.,
hourly or shorter) rainfall.
Subdaily precipitation extremes are often pro-
duced by convective events that are affected by
complex interactions among mesoscale dynamics,
cloud microphysical processes, and underlying sur-
face forcing. Current understanding of the changes
in subdaily extreme precipitation is very limited
(Zhang et al. 2017) because of lack of high-resolution,
long-term observations and limited knowledge about
the physical mechanisms that govern the evolution of
convective events. Climate models are not able to sim-
ulate such events well (Siler and Roe 2014;Z. H. Jiang et
al. 2017;Pfahl et al. 2017), making it difficult to attribute
past changes and to assess future changes in short-
duration precipitation extremes. Moreover, factors
affecting changes in short-duration precipitation ex-
tremes over urban environments are even less known
because of complex and sometimes compensating effects
of cities, namely, the destabilization caused by urban
heat island (UHI)-induced thermal perturbation and its
downstream rainfall enhancements translation (Huff and
Vogel 1978;Hjelmfelt 1982;BornsteinandLin2000;Craig
and Bornstein 2002;Niyogi et al. 2011), the building-
barrier and thermal effects of urban canyons (Bornstein
and Lin 2000;Guo et al. 2006;Zhang et al. 2009;Miao et al.
2011), and the anthropogenic aerosol emissions for cloud
condensation nuclei (CCN) sources (Rosenfeld 2000;Bell
et al. 2009;Jin and Shepherd 2008;Ntelekos et al. 2009).
Guangdong province, located in coastal South
China (Fig. 1a), is exposed to high flood-induced risk
(Hallegatte et al. 2013) because of the high frequency
of heavy rainfall at hourly (Luo et al. 2016) to longer
scales (Zheng et al. 2016). The heavy rainfall in South
FIG. 1. (a) The urban region (color shading) in Guangdong province identified by the
DMSP/OLS nighttime lights data in five different years. The bold black line represents the
PRD urban region in 2013, which is also used in most of the following figures. The dashed
rectangle denotes the key region defined in this study (21.78–23.78N, 112.68–114.68E). The
locations of the major cities in the PRD are labeled out. (b) Evolution of urbanization re-
flected by time series of the population density (km
2
) and built-up area (10
3
km
2
)in
Guangdong and the urban ratio (%) within the key region based on the land-use data.
1800 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 58
China is influenced by large-scale to synoptic-scale
weather systems including the Asian summer mon-
soon (Ding 1994), subtropical ridge of high pressure
in western North Pacific and heat low over southern
and southwestern parts of China (Ramage 1952),
cyclonic vortex anomalies moving from downstream
of the Tibetan Plateau (Huang et al. 2018), and
tropical cyclones (Li and Zhou 2015), as well as
mesoscale forcing associated with an underlying
surface (such as land–sea contrast and coastal
mountains) cold pool generated by significant con-
vective storms (Wang et al. 2014;Wu and Luo 2016;
Liu et al. 2018).
The rapid expansion of urban areas in Guangdong
province during the past two decades (Fig. 1b) led to
the formation of city clusters in the Pearl River delta
(PRD) (Fig. 1a). In 2016, the PRD covered an area of
about 42 200 km
2
with a population over 57 million,
accounting for 52.0% of the total population in
Guangdong (Guangdong Statistical Bureau 2016). It
encompasses a cluster of big cities, including the cap-
ital city of Guangdong province (Guangzhou) at its
northern apex, Hong Kong and the Special Economic
Zone (SEZ) of Shenzhen at the southeast, while Ma-
cao and Zhuhai SEZ over the southwestern part of
PRD. Based on satellite data in 1998–2009, Li et al.
(2011) suggest that the urban areas in PRD experience
more (less) occurrences of heavy (light) precipitation
compared to surrounding suburban regions. Under-
standing changes in subdaily precipitation extremes
FIG. 2. (a) Land-use map over Guangdong in 2015, overlaid with the PRD urban region
boundary based on the DMSP/OLS nighttime lights data in 2013 (solid black). (b) Spatial dis-
tributions of 61 national-level stations (red dots) in the analysis domain. Each station is assigned a
number, which is also used in many of the following figures. The key region is outlined by dotted
rectangle. Dots of smaller sizes represent AWSs within the key region, with orange/lavender ones
denoting stations inside/outside of the PRD urban region. Shading represents the topography.
AUGUST 2019 W U E T A L . 1801
and their relationship with urbanization is critical to
urban policy makers for developing more sustainable
and resilient cities in this region.
This study aims to investigate changes of extreme
hourly precipitation in coastal South China with a
focus on its possible relationship with the urbaniza-
tion in PRD. To achieve this objective, the long-term
changes in extreme hourly precipitation (EXHP; de-
fined by the 95th percentile) are analyzed using hourly
precipitation observations at 61 national-level surface
stations from 1971 to 2016. Then a total of 120 extreme
rainfall events (EXREs) (with at least one record of
hourly rainfall at 60 mm or more in the PRD region)
during 2011–16 are further examined using observa-
tions collected by densely distributed automatic
weatherstations(AWSs)andradarnetwork.The
paper is organized as follows: The next section de-
scribes the data and analysis methods. Section 3
presents the changes of hourly precipitation extremes
during 1971–2016. Section 4 classifies the EXREs
from2011to2016intosixtypesaccordingtothe
characteristics in the movement of rainstorms, and
compares between the strong- and weak-UHI sub-
groups for each type. A summary and conclusions are
provided in section 5.
2. Data and analysis methods
a. Estimating the PRD urbanization
This study combines historical land-use data,
satellite-based nighttime lights dataset, and statisti-
cal data of population density and built-up area
from the Guangdong Statistical Bureau (2016;http://
www.gdstats.gov.cn/) to assess the urbanization in
PRD. The historical land-use data of horizontal reso-
lution at 1 km are available from the Resources
and Environment Scientific Data Center, Chinese
Academy of Sciences (http://www.resdc.cn/data.aspx?
DATAID598) for 1980, 1990, 1995, 2000, 2005, and
2015basedonvisualinterpretation and digitalization
of the 30-m Landsat TM/ETM satellite images (Gong
et al. 2013). They include six land-use types: crop-
lands, forest, grasslands, water bodies, built-up lands,
and others. Because of its high accuracy in monitoring
land-use change in China, this dataset has been ex-
tensively used in the fields of land resource surveys,
hydrology, and ecology (e.g., Li et al. 2015;Liu et al.
2014). Moreover, the fourth version composite satellite-
based nighttime lights data gathered during 1992–2013
derived from the Defense Meteorological Satellite
Program’s Operational Linescan System (DMSP/OLS)
of the United States (https://ngdc.noaa.gov/eog/dmsp/
downloadV4composites.html) are utilized. The digi-
tal numbers (DN) values of 50, 56, and 57 are used
as the urban thresholds for the years of 1992, 1995,
and in and after 2000, respectively, based on Wang
et al. (2013).
The aforementioned three independent data sources
all reveal a rapid urbanization in the PRD region since
the early to mid-1990s, as suggested by the continuous
increases of the built-up area, population density, and
ratio of urban land use, as well as the expansion of
nighttime light (Figs. 1a,b). The boundary of the PRD
urban region is represented hereafter by the con-
tour line derived from the DMSP/OLS data in 2013
(Fig. 1a), which agrees well with the land-use data in
2015 (Fig. 2a).
b. Analysis of the long-term changes in hourly
precipitation extremes
This study utilizes the long-term, gauge-based
hourly precipitation dataset provided by the National
Meteorological Information Center (NMIC) of the
China Meteorological Administration (CMA; http://
data.cma.cn/data/online.html?t51)toanalyzethe
TABLE 1. Classification of the hourly extreme precipitation in 1971–2016 according to the temporal evolution of hourly precipitation in the
3 h prior to the hourly extreme.
Temporal
evolution type
Hourly rainfall rates in 3 h prior to the extreme rainfall hour (R
21
,R
22
,
R
23
, respectively) compared with the extreme hourly rainfall (R
0
)
Contribution to total occurrence of the
non-TC hourly precipitation extremes (%)
Abrupt type R
21
,0.1R
0
and R
22
,0.1R
0
and R
23
,0.1R
0
43.3
Growing type (R
21
.R
22
or R
21
.R
23
or R
22
.R
23
) and at least one of
(R
21
,R
22
,R
23
).10% but ,R
0
40.7
Continuous type At least one of (R
21
,R
22
,R
23
).R
0
16.0
TABLE 2. Number of the categorized 2011–16 extreme
rainfall events.
Type Strong-UHI events Weak-UHI events Total
Local/SW wind 18 20 38
Local/shear line 17 8 25
Migratory-NW 20 14 34
Migratory-SW 3 10 13
Migratory-NE 6 0 6
Migratory-S 4 0 4
Total 68 52 120
1802 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 58
long-term changes of hourly extreme precipitation
over the analysis domain (Fig. 2b). A strict quality-
control procedure has been applied to the dataset
by NMIC consisting of a climatological limit value
test, a station extreme value test, an internal consis-
tency test, and a comparison with manually checked
daily rainfall data. This dataset has been extensively
used to investigate the characteristics of subdaily pre-
cipitation over China (e.g., Yu et al. 2007;Li et al. 2008;
Yuan et al. 2012;Luo et al. 2016;Guo et al. 2017). This
study utilizes 60 stations in Guangdong and one station
in Hong Kong that have continuous records from 1971 to
2016 (distribution shown in Fig. 2b). Each of the 61
stations has over 95% of valid hourly data each year.
The EXHP at each station is defined using the 95th
percentile of the distribution function during 1971–2016
as the threshold. The threshold values increase south-
ward from about 6 to 13 mm h
21
(not shown). The
rainfall amount and occurrence frequency of the EXHP
are calculated as the annual rainfall amount and the
total number of hours of the EXHP, respectively.
The EXHP intensity is calculated as the rainfall amount
divided by its occurrence frequency. The rainfall
amount, occurrence frequency, and intensity of hourly
precipitation ($0.1 mm h
21
) are defined similarly. Their
long-term trends are estimated using a linear ten-
dency method. The significances of the trends are ex-
amined utilizing the Mann–Kendall nonparametric test
(Mann 1945;Kendall 1975). The test has been applied
extensively including in studies that test for changes
in rainfall extremes (Alexander and Arblaster 2009;
Westra et al. 2013). To make the trends more compa-
rable among the various variables (i.e., precipitation
amount, occurrence frequency, and intensity), relative
percentage changes are calculated using the equation
below:
D510 3S
P(100%), (1)
where Dis the relative change in precipitation
(% decade
21
), Sis the slope of the linear model, and Pis
the precipitation value in the start year estimated using
the linear regression equation. The significances of the
linear regression equation are examined using the Ftest
(Lomax and Hahs-Vaughn 2007).
Moreover, rainfall produced by tropical cyclones
(TCs) is identified with an objective synoptic analysis
technique (OSAT; Ren et al. 2006,2007,2011). This
method uses the distance from TC center and the
closeness and continuity between neighboring raining
stations to trace TC-influenced rain belts. An extreme
hourly precipitation record in this study is classified
asaTC-inducedEXHRrecordifitoccurswithinthe
identified TC-influenced rain belts; otherwise it is a
non-TC record. Previous studies have suggested that
rainfall trends over China are complicated by TCs
FIG. 3. (a),(c) Scatterplot of DTand its corresponding DTat each of the 6 h prior to the onset of the 2011–16
EXREs with (a) a strong-UHI and (c) a weak-UHI, respectively. (b),(d) Box-and-whisker plots of UHII for the
2011–16 EXREs with (b) a strong-UHI and (d) a weak-UHI, respectively, showing the interquartile range
(rectangle), outliers (i.e., the 10th and 90th percentile as whiskers), and mean (solid line).
AUGUST 2019 W U E T A L . 1803
(Chang et al. 2012;Li et al. 2017), and the TC-induced
hourly precipitation extremes contribute about 20% to
the total occurrence frequency of the extremes over
coastal South China (Luo et al. 2016). The precipitation
excluding the TC-produced is termed as non-TC pre-
cipitation hereafter. Furthermore, the hourly extreme
precipitation records are classified into the abrupt,
growing, and continuous types according to the tempo-
ral evolution features of the hourly rainfall series
(Table 1;Liang and Ding 2017), and into the short-
(1–6 h), moderate- (7–12 h), and long-duration (.12 h)
types based on the time span of continuous rainfall
(.0.1 mm h
21
with at most 1-h intermittence).
c. Analysis of the 2011–16 extreme rainfall events
To complement the 46-yr linear tendency analysis,
the possible relationship between a strong UHI ef-
fect and the 2011–16 EXREs are examined. Both the
quality-controlled hourly records at densely distrib-
uted AWSs and 10-min mosaic radar reflectivity in
Guangdong are used to provide detailed information
about the convective events. The EXRE is defined
as a rain system that produces at least one rainfall
record of .60 mm h
21
observed by the AWSs within
the key region (defined as a 28328region in the
analysis domain; dotted rectangle in Fig. 2)covering
the entire PRD urban cluster and the adjacent forest,
cropland, and water. The 10-min radar images are
adopted to verify the reliability of the extreme
rainfall records, as well as to characterize the life
cycle of the extreme-rain-producing convective sys-
tems. A total of 133 EXREs are found. Among them
13 cases are featured by an extensive cyclonic vortex
at 850 hPa over South China that helps produce a
persistent (duration .1 day), extensive precipitation
region (.1000 km on the radar image). The large-scale
forcing, rather than the UHI-influenced local envi-
ronment, is believed to play a dominant role in mod-
ulating the 13 EXHRs. Contrastingly the other 120
EXREs are mainly produced by MCSs lasting less than
24 h. As the focus of this study is the possible link be-
tweentheUHIandEXHP,the13EXREswithsuch
large-scale vortices are excluded from the following
analysis.
For the remaining 120 EXREs, the intensity of the
PRD UHI during the onset of each EXRE (UHII) is
calculated using the equations below:
UHII 5DT2DT, (2)
DT5Tu2Tr, and (3)
DT5
å
2016
2011
å
day15
day25
DT
6311 , (4)
where T
u
and T
r
, respectively, are the hourly surface air
temperatures averaged over the densely distributed
AWSs within the PRD urban region (small orange dots
in Fig. 2b) and outside the PRD urban region (small
FIG. 4. The pre-event potential temperature perturbation (Du) for the 2011–16 EXREs with (a) a strong-UHI and
(b) a weak-UHI, respectively. The hourly observations in the 3-h period prior to each EXRE onset are used in the
calculation.
1804 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 58
purple dots in Fig. 2b); thus DTroughly represents the
hourly temperature difference between the PRD urban
cluster and its surrounding areas. To remove the sea-
sonal and diurnal variations of DT, UHII is represented
by the difference between DTand DT. The value of DT
is calculated by averaging DTat the same local solar
time (LST) during 11 days (from 25to15 days) in
2011–16. For example, if the DTvalue is calculated at
1500 LST 12 June 2013, its corresponding DTis the av-
erage of 66 DTvalues at 1500 LST from 7 to 17 June in
2011–16. Thus, UHII .0 means that the UHI intensity
prior to each EXRE onset is stronger than its corre-
sponding multiyear average intensity. The pre-event
hourly UHII
i
(i51, 2, ... , 24; denoting the hour prior
to the EXRE onset) values are calculated for each
EXRE. Existence of a strong PRD UHI for an EXRE
is defined as at least four out of six hours with UHII
i
.0
(i51–6); these EXREs are referred to as the strong-
UHI events. The rest EXREs have at most three out of
six hours with UHII
i
.0(i51–6) and are classified
into the weak-UHI events. Existence (lack) of such a
strongUHIisfoundfor68(52)EXREs(Table 2).
Composites of the UHII for the strong- and weak-
UHI events are shown in Fig. 3. The mostly (99.25%)
positive values of the multiyear average DTindicate
thepresenceofthePRDUHIwithvaryingin-
tensities from nearly zero to about 18C(Figs. 3a,c),
probably in association with the seasonal and diur-
nal variations of UHI intensity (Landsberg 1981;
Shepherd 2005;Ren and Zhou 2014). The hourly
UHII values are mostly positive in the 4-h preonset
periods and increases rapidly 6-h before the onset of
the strong-UHI events (Figs. 3c,d). For the weak-
UHI events, about 68.7% of the DThas positive
values, reflecting presence of an UHI prior to the
events’ onset. A certain fraction (31.3%) of negative
DTcan be contributed by rain evaporative cooling
due to previous convection in the urban cluster that
might occur occasionally prior to the EXREs. Mostly
negative values of UHII are observed in the 6-h preonset
periods (Fig. 3d), suggesting UHI intensities (DT) are
weaker than the multiyear average.
To examine the spatial distribution of UHI, the sur-
face potential temperature perturbation (Du) at each
station is calculated as the deviation from the spatial-
average value of uwithin the key region at each hour.
This method largely removes the seasonal and diurnal
variations of surface potential temperature (u). The
distribution of Duaveraged during 1–3 h prior to each
EXRE is compared between the strong- and weak-
UHI EXREs (Fig. 4). A prominent surface warm
center over the inland PRD urban region is noticed in
the average Dufield of the strong-UHI events (Fig. 4a).
Such a prominent warm center does not exist in the
FIG. 5. (a) Annual precipitation amount (mm yr
21
) averaged during 1971–2016 and (c) its difference between 1994–2016
and 1971–93. (b),(d) As in (a),(c), respectively, but for the extreme hourly precipitation (.95th percentile).
AUGUST 2019 W U E T A L . 1805
weak-UHI events, as smaller positive Duvalues are
widely distributed in the southwest part of the key re-
gion (Fig. 4b).
The 2011–16 EXREs are then categorized into six
types mainly based on the rain storms’ initiation location
and movement observed by the radar network. Specifi-
cally, the EXREs with extreme-rain-producing storms
initiating and developing within the key region are
selected first, and classified as either local/southwest
(SW) wind type or local/shear line type according to
their synoptic patterns on 925 hPa using the ERA-
Interim reanalysis (Dee et al. 2011). The former has
prevailing southwesterly airflows and the latter has
a shear line over the key region. Then, all the other
EXREs with extreme-rain-producing storms moving
from outside the key region are classified into four
types based on the direction the storms move into the
key region, namely, the migratory-northwest (NW),
migratory-northeast (NE), migratory-south (S), and
migratory-southwest (SW) types.
To examine possible UHI impacts, rainfall distribu-
tions of the strong- and weak-UHI EXREs in the same
type are compared. For this purpose, the normalized
rainfall (NR) is calculated for each AWS in the key re-
gion following Eqs. (5)–(7):
Rj5å
n
1
Rjk(j51, m;k51, n), (5)
R5
å
m
1
Rj
m, and (6)
NRj5Rj2R
R, (7)
where R
jk
is event-accumulated rainfall amount of
EXRE k(k51, n; where nis the total number of
EXREs in a subgroup) at site j(j51, m; where mis the
total number of AWSs in the key region), R
j
is rainfall
accumulation of an EXRE subgroup at site j,Ris the
subgroup rainfall accumulation averaged over all the
AWSs in the key region, and NR
j
is the normalized
rainfall anomaly at site jrelative to the domain average
of the EXRE subgroup. This method is able to highlight
local features of the rainfall distribution (Yu 2007;
Dou et al. 2015).
3. Changes in hourly precipitation extremes
during 1971–2016
Figures 5a and 5b shows the spatial distributions of
total precipitation amount and extreme precipitation
amount over the analysis domain averaged during
1971–2016. The total precipitation amount has three
peaks in the southwest coastal, central, and south-
east coastal Guangdong, respectively, with the annual
maximal precipitation amounts exceeding 2100 mm yr
21
in the southwest center and of about 1900–50 mm yr
21
in the other two centers. The inland PRD urban re-
gion, with an annual precipitation amount of about
1800 mm yr
21
, is located between the southwest
and central precipitation centers. Such a distribution
shows a largely similar pattern to that of the hourly
extremes, except that the southwest and central
FIG. 6. The observed 1971–2016 changes (% decade
21
)inthe
hourly precipitation ($0.1 mm h
21
): (a) amount, (b) occurrence
frequency, and (c) intensity. Dots (circles) denote signifi-
cant (insignificant) trends at the 95% confidence level using
the Mann–Kendall test. Size of the dots and circles represents
the various magnitudes of the changes. Shadings represent the
topography.
1806 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 58
centers tend to connect crossing the east PRD urban
region while the southeast center is less evident. Of in-
terest is that the difference in the annual precipita-
tion amount between the two 23-yr periods, that is,
1994–2016 minus 1971–93, reveals a prominent increase
over the PRD region and the northern Guangdong re-
gion (Figs. 5c,d).
Figure 6 shows an increasing trend in the precipitation
amount for most stations over the PRD region. Never-
theless, only the Guangzhou station (number 35) passes
the significant test at 95% confidence level. Positive and
negative trends are scattered outside the PRD region,
but they are statistically insignificant. Negative trends in
the occurrence frequency of hourly precipitation are ob-
served at most stations over Guangdong, with only
a few passing the significant test. By contrast, the hourly
precipitation intensity exhibits an increasing trend
at most stations. Importantly, the most significant in-
creasing trends in the hourly precipitation intensity are
mainly confined in the PRD region. These results suggest
that more intense hourly precipitation, which is statisti-
cally significant over the PRD urban region (Fig. 6c),
leads to the higher annual amounts of both total pre-
cipitation and extreme precipitation over the PRD in
1994–2016 than 1971–93 (Figs. 5c,d), suggesting a pos-
sible link to the rapid urbanization since mid-1990s.
FIG. 7. (a)–(c) The observed 1971–2016 changes (% decade
21
) in the non-TC extreme hourly precipitation:
(a) amount, (b) occurrence frequency, and (c) intensity. (d)–(f) As in (a)–(c), but for all the extreme precipitation
including the TC-induced records. Dots (circles) denote significant (insignificant) trends at the 95% confidence
level using the Mann–Kendall test. Size of the dots and circles denotes different magnitudes of the changes.
Shadings represent the topography.
AUGUST 2019 W U E T A L . 1807
Changes in the amount, frequency, and intensity of
the non-TC extreme hourly precipitation during 1971–
2016 are shown in Figs. 7a–c. Data from 11 stations
reveal statistically significant (95% level) positive
trends in the extreme rainfall amount, with six stations
located in the PRD urban region while the other five
stations scattered in the western, northern and eastern
parts of Guangdong, respectively. The significant pos-
itive trends of the extreme rainfall frequency are ob-
served only at seven stations over the analysis domain,
mostly (five out of seven) in the PRD urban region,
especially its inland portion. The extreme hourly pre-
cipitation intensity also increases at most of the PRD
urban stations. However, the trends are insignificant
at 95% level. These results suggest that the in-
creased amount of non-TC precipitation extremes
may mainly be caused by their more frequent oc-
currence over the PRD urban region. The inclusion
of TC-induced extreme hourly rainfall results in a
smaller number of stations with statistically signifi-
cant positive trends of the amount (8 vs 11) (Fig. 7d
vs Fig. 7a)andfrequency(4vs7)(Fig. 7e vs Fig. 7b),
which is qualitatively consistent with the finding of a
reduction in landfalling TC occurrence over South
China during 1975–2014 by Li et al. (2017). Note that
the significant positive trends in the occurrence fre-
quency and amount of the hourly precipitation ex-
tremes, including the TC-induced ones, are still
concentrated over the PRD urban region, again
suggesting a possible rainfall enhancement by the
urban cluster.
The seven stations with a significant increasing
trend of the EXHP’s occurrence frequency (Fig. 7b) are
selected to further examine their time series and
change rates during 1971–2016. Similar analyses are also
conducted for the pre- and post-rapid-urbanization
FIG. 8. (a)–(g) Time series (gray bars) and linear trends (solid lines) of the non-TC extreme hourly rainfall
occurrence frequency at seven stations, with the station number labeled at the upper-left corner. The locations of
the seven stations could be seen in Fig. 1a. The annual occurrence frequency is normalized by subtracting the mean
value of the study period from the annual value of each year, and then divided by the corresponding standard
deviation. Blue lines denote the linear trends for the period of 1971–2016, while the orange/red lines are for
1971–93/1994–2016. The solid (dashed) line denotes that the linear regression equation is significant (insignificant)
at the 95% confidence level using the Ftest. The corresponding slope values of the linear model for the three
periods are shown in the bottom sequentially.
1808 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 58
periods, respectively, to further examine possible in-
fluence of the urban effect on the increasing EXHP.
The comparison between the two periods at each
station remains at least qualitatively unchanged when
any year of the 1991–97 period is used as the de-
marcation point. Therefore, only the results using
year 1994 as the demarcation point are presented
herein. There are five stations (numbered 33, 34, 35,
39, 40) are in the PRD urban region, and the other two
(numbered 04 and 27) are in the north and east
Guangdong, respectively. The five stations in the PRD
urban region have either increase or decrease rates
during the preurbanization era of 1971–2016 (orange
lines; Figs. 8a–e). They all have increase rates during
the rapid urbanization period (1994–2016) (red lines;
Figs. 8a–e). The later 23-yr increase rates are about 1.44
to 2.36 times of the 46-yr (1971–2016) change rates. In
contrast, at the other two stations away from the core
PRD urban region the increase rates in the later 23-yr
period are nearly the same (station 04; Fig. 8f) or even
smaller (station 27; Fig. 8g) compared to their coun-
terparts in the preurbanization era. These results con-
sistently support possible contribution of the PRD
urbanization to more occurrences of hourly precipita-
tion extremes.
Moreover, it is found that the statistically significant
trends in thehourly precipitation extremes over the PRD
urban region are more closely related to the abrupt type
FIG. 9. Spatial distributions of the occurrence frequency (shading, %) and trends (line, decade
21
) of the cate-
gorized non-TC extreme rainfall events in 1971–2016: (a) abrupt type, (b) growing type, (c) continuous type,
(d) short-duration type, (e) middle-duration type, and (f) long-duration type. Gray dots represent that the trends
are significant at the 95% confidence level using the Mann–Kendall test. The occurrence frequency of an extreme
precipitation type is the fractional contribution of this type to the total number of the extreme rainfall events.
AUGUST 2019 W U E T A L . 1809
extremes than the growing or continuous types (defini-
tions given in Table 1). The abrupt type accounts for 44%
of the total occurrence of the non-TC hourly extremes
over the 61 stations in 1971–2016, the growing type is a
close second (40%), and the continuous type contributes
the least (16%). The spatial distributions of their fractional
contributions (shadings in Figs. 9a–c) show that the abrupt
type contributes the most in west Guangdong but rela-
tively less along the coastline and in north Guangdong,
while both the growing and the continuous types similarly
show a largely opposite pattern to the abrupt type. Sig-
nificant increase trends are observed only in the abrupt
FIG. 10. Composite geopotential height (gpm; green lines) and wind barbs on 925 hPa for the six types of the
2011–16 extreme rainfall events: (a) local/SW wind type, (b) local/shear line type, (c) migratory-NW type,
(d) migratory-SW type, (e) migratory-NE type, and (f) migratory-S type. A full barb is 5 m s
21
. The gray dots and
bold wind barbs denote the values that are over the 90% confidence level using the Student’s ttest. A typical rain
storm of each extreme rainfall type is illustrated by radar reflectivity (color shadings), overlaid with a black arrow to
denote its movement direction. The contribution of each type to the total number of the 2011–16 extremes (120) is
labeled in parentheses. The gray-shaded regions denote the portions of isobaric surfaces underneath the ground.
1810 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 58
type, mainly over the PRD urban region, with only a few
scattered in other places of Guangdong (Fig. 9a).
Furthermore, when the non-TC extremes are clas-
sified by the time span of continuous rainfall (.0.1 mm h
21
with at most 1-h intervals), the short-, moderate-, and long-
duration types contribute 33.5%, 34.0%, 32.5%, re-
spectively, to the total occurrence of the extremes. Their
corresponding spatial patterns (shadings in Figs. 9d–f)
suggest that the short-duration type contributes more
in west Guangdong (about 40%–48%) than in other
places of Guangdong especially along the coastline
(mostly ,32%); the long-duration ones show a nearly
opposite feature; and the moderate-duration type ac-
counts for about 32%–36% evenly over the entire
analysis domain (Figs. 9d–f). Qualitatively similar
to the abrupt type (Fig. 9a), significant increase
trends of the short-duration events are observed and
concentrated in the PRD urban region (Fig. 9d),
while the longer-duration types show a lack of posi-
tive trends that are significant at the 95% confidence
level.
The abovementioned results collectively suggest that
the statistically significant increasing trends of hourly
extreme precipitation in the PRD urban cluster are
likely attributed to the rapid urbanization. Such trends
are closely related to increasing occurrence frequency of
short-duration, abrupt rainfall events, which may pose a
greater threat to flash floods in the urban cluster and
raise more challenge for accurate prediction.
4. The 2011–16 EXREs
As described in section 2c, totally 120 EXREs
during 2011–16, each with at least one rainfall record
FIG. 11. Seasonality of the 2011–16 extreme rainfall events: (a) local/SW wind type, (b) local/shear line type,
(c) migratory-NW type, (d) migratory-SW type, (e) migratory-NE type, and (f) migratory-S type. Pink and blue
bars, respectively, represent the strong- and weak-UHI events. The number of events within each subtype is shown
in parentheses.
AUGUST 2019 W U E T A L . 1811
of .60 mm h
21
within the key region, are classified
into six types (Table 2). This section first describes
synoptic circulation and seasonality of each EXRE
type. Then the two subgroups with the strong- and
weak-UHI that belong to the same EXRE type are
compared in terms of the temporal and spatial distri-
butions of the precipitation.
a. Synoptic circulation and seasonality
Among the EXREs, 63 events (52.5%) have locally
developed rain systems, including 38 (31.7%) under the
influence of prevailing southwesterly winds in the lower
troposphere and the planetary boundary layer (PBL)
and 25 (20.8%) accompanied by a synoptic shear line
in the PBL over South China (Figs. 10a,b). These two
types of locally developed rain systems are catego-
rized into the local/SW wind and local/shear line types,
respectively. The other 57 EXREs originated outside of
the key region and moved from the northwest (28.4%),
southwest (10.8%), northeast (5.0%), or south (3.3%) to
influence the PRD urban region. These are categorized
into the migratory-NW, migratory-SW, migratory-NE,
or migratory-S types, respectively. The migratory-NW
type is usually accompanied by a northeast–southwest
oriented shear line in South China (Fig. 10c). In the
migratory-NE type, a significant TC or its remnants
is present around the Taiwan Strait, accompanied by
a large precipitation area from where the migratory
storm is separated (Fig. 10d). The migratory-SW and
migratory-S types take place in prevailing southerly and
southeasterly winds in the PBL (Figs. 10e,f).
Figure 11 shows the monthly occurrence frequency
of the 2011–16 EXRE types, with a strong or weak
PRD UHI, respectively. The two local types occur
FIG. 12. (a) The timing of the local/SW wind type events with a strong UHI. The beginning time of each event is
marked with a cross. The number of events and their average duration are also shown. (b) Composite spatial
distribution of region-normalized rainfall amount (shading). Blue (dark blue) dots denote hourly rainfall records
exceeding 60 mm h
21
occur once (more than once) in total among this type. (c),(d) As in (a),(b), respectively, but
for the local/SW wind type events with a weak UHI.
1812 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 58
more evenly from May to September than the migra-
tory types. However, the local/SW wind type has
about 60% (23 out of 38) in the middle-and-late stage
of the presummer rainy season (May and June), while
the local/shear line type has about 60% (15 out of 25)
in August and September. The migratory-NW type
mostly (85%) takes place in April and May with the
remaining 15% in June–August. The migratory-SW,
migratory-NE, and migratory-S types are completely
found in April-June, July–August, and May-July, re-
spectively. The former four types have both strong- and
weak-UHI events, without a remarkable difference
between the strong and weak UHI events in the sea-
sonality of each EXRE type. The migratory-NE and
migratory-S types of the hourly precipitation extremes
occur only with the strong UHI.
b. Comparison between local EXREs with a strong
and weak UHI
The local/SW wind type with the strong UHI mostly
(13 out of 18) initiates from the midday to afternoon and
ends in the evening (Fig. 12a), producing extreme hourly
rainfall .60 mm h
21
mainly in the late afternoon. In
contrast, with the weak UHI, this type of rain storms
mostly (14 out of 20) initiates from late evening to
nocturnal hours (Fig. 12c) and lasts significantly longer
by average (11.3 vs 7.0 h). This interesting contrast
suggests possible difference in in large-scale environ-
ments for the two subgroups. Therefore, the key region-
averaged values of 700-hPa vertical velocity, 850-hPa
wind speed, 925-hPa wind convergence, and precipitable
water, are compared among the local/SW wind type of
EXREs (Fig. 13). The results suggest that the strong-UHI
events tend to be associated with relatively slower upward
motion, lower wind speed, weaker PBL wind conver-
gence, and smaller amount of precipitable water (PW).
These indicate relatively less favorable large-scale con-
ditions for persistent convective development and could
partially explain the shorter duration of the strong-UHI
events. These contrasting results between the two sub-
groups suggest that the UHI thermal perturbation (e.g.,
Shepherd et al. 2002;Shepherd and Burian 2003)likely
plays an important role in the convective initiation and
development in the afternoon, when the UHIsare greater
in clear and calm conditions (Landsberg 1981;Oke
1987;Yang et al. 2017) but the PBL southwesterly flows
over South China tend to be weaker than in nocturnal
hours (Du et al. 2014). The extreme hourly rainfall
(.60 mm h
21
) records of the local/SW wind type
(blue dots in Figs. 12b,d) are observed mainly within
the urban cluster and close (within about 50 km) to its
northern boundary regardless of the UHI strength,
while some extreme rainfall records in the weak UHI
cases are located near the coastline.
The strong-UHI subgroup of the local/shear line
type mainly (13 out of 17 events) initiates and deve-
lops during the daytime particularly in the afternoon-
to-evening hours (Fig. 14a). This subgroup has the
hourly extremes and rainfall accumulation mostly
over the inland portion of the PRD urban cluster
(Fig. 14b),wheretheUHIismostintense(Fig. 4a).
In contrast the weak-UHI subgroup has 3 out of 8
events initiating in the nocturnal, 1 in the morning,
and 4 in the afternoon (Fig. 14c). The weak-UHI
subgroup produces rainfall mostly near the coastline
(Fig. 14d), attributable to more (less) favorable
FIG. 13. (a) Vertical velocity (210
22
Pa s
21
) at 700 hPa, (b) horizontal wind speed (m s
21
) at 850 hPa, (c) wind
convergence (210
26
s
21
) at 925 hPa, and (d) precipitable water (mm) averaged over the key region using the
ERA-Interim reanalysis data. The red (blue) symbols represent the 18 (20) local/SW wind type EXREs with a
strong (weak) UHI, and the red (blue) lines denote the corresponding average values.
AUGUST 2019 W U E T A L . 1813
thermodynamic conditions [i.e., more (less) moisture
and larger (smaller) convective available potential
energy] to the south/north of the shear line (Fig. 10b).
The different features in the temporal and spatial
distributions of the two subgroups support possible
convective intensification and rainfall enhancement
by the strong UHI over the inland PRD urban region
for the local/shear line type EXREs.
In addition to the UHI effect, sea breezes could
also play an important role in convection development
and rainfall production over the PRD region; for ex-
ample, a close association between sea breezes and in-
land propagation of warm-season rainfall over the PRD
coastal region during the daytime was noticed (Chen
et al. 2016;Z. N. Jiang et al. 2017). Moreover, extreme
precipitation events over coastal South China can be
influenced by the trumpet-shaped topography of the
PRD (Huang et al. 2019) and the coastal mountains
(Wang et al. 2014;Wu and Luo 2016). Interactions be-
tween sea breezes, UHI-induced circulation, larger-
scale southwesterly air flows (Fig. 10a), and topography,
as well as their influence on the timing, location, and
evolution of rain storms in the local/SW wind type of
EXREs deserve further study.
c. Comparison between migratory EXREs with a
strong and weak UHI
ForthemigratoryEXREtypes,astrongPRDUHI
is expected to impact the rainfall intensity and distri-
bution over the key region rather than the timing
of these EXRE types. Visual examination of radar
reflectivity animation indeed suggests notable inten-
sification of radar reflectivity when the rain storms
approach the strong-UHI region especially over the
inland PRD, which is not observed in the weak-UHI
cases. This expectation is also confirmed by examin-
ing the rainfall distribution. Therefore, this subsection
will discuss mostly the spatial distributions of ex-
treme hourly rainfall records and accumulated rain-
fall amounts, although the timing of the EXREs will
still be shown.
For the migratory-NW type, which accounts for
60% of the total number of the migratory EXREs,
both the hourly precipitation extremes and rainfall
FIG. 14. As in Fig. 12, but for the local/shear line type of the 2011–16 EXREs.
1814 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 58
accumulation tend to concentrate over the inland portion
of the PRD urban cluster in the strong-UHI subgroup
(Fig. 15b), while those of the weak-UHI subgroup are
contrastingly situated near the coastline with smaller
rainfall amounts in the inland urban region (Fig. 15d).
The larger rainfall amounts near the coastline could
be attributed to higher equivalent potential tempera-
ture (u
e
) of air masses near the coastline in the weak-
UHI cases (not shown).
The migratory-SW type consists of only 3 EXREs
with strong UHI. Despite of the small number of events,
it is still noted that both the hourly precipitation ex-
tremes and rainfall accumulation are mainly concen-
trated over the urban cluster (Fig. 16c). In contrast
the 10 migratory-SW EXREs with weak UHI have the
hourly extremes mainly distributed over the southwest
coastal area (Fig. 16d).
All the 6 migratory-NE EXREs are accompanied
by a strong UHI and their extreme hourly precipita-
tion records are mostly located over the urban cluster
(Fig. 17b). The rain storms in these cases intensify im-
mediately when enter the urban cluster and weaken
quite rapidly when move out from the southwest bound-
ary of the urban region. Moreover, three extra rainfall
events that similarly have the rain storm moving out
from a TC or its remnant centered around the Taiwan
Strait to influence the PRD region are found. These
three events have weak UHI and do not produce
hourly precipitation .60 mm h
21
. The radar animation
(not shown) suggests that the rain storms weaken
or dissipate when they pass the PRD urban region,
opposite to the evolution of rain storms in the 6 strong-
UHI cases.
The 4 migratory-S EXREs are also accompanied by
a strong UHI. Examination of the radar anima-
tion (not shown) suggests convective intensification
when the storms move over the UHI. However, the
storms produce accumulative rainfall mainly over
the coastal region, with a secondary rainfall center
over the northwest PRD urban region (Fig. 17d).
FIG. 15. As in Fig. 12, but for the migratory-NW type of the 2011–16 EXREs. The arrow in (c) and (d) denotes the
movement direction of the extreme-rain-producing storm.
AUGUST 2019 W U E T A L . 1815
Stronger evidence of rainfall intensification by the UHI
is limited by the small number of this EXRE type.
5. Summary and conclusions
This study investigates the changes in extreme hourly
precipitation in coastal areas of South China, focusing
on its relationship with the PRD urbanization through:
1) exploring long-term changes of hourly extremes
(defined by the 95th percentile) using gauge-based
hourly precipitation observations at 61 national-level
surface stations over 46 years (1971–2016); and 2) ana-
lyzing the 120 extreme rainfall events (with at least one
record of .60 mm h
21
hourly rainfall in the key region)
in 2011–16 using observations collected by densely dis-
tributed AWSs and radar network in coastal South
China. The major findings are summarized as follows:
1) Statistically significant increase of hourly precipitation
intensity leads to higher annual amounts of both total
and extreme precipitation over the PRD urban cluster
in the rapid urbanization period (about 1994–2016)
than during the preurbanization era (1971–about
1993), suggesting a possible link of rainfall enhance-
ment to the rapid urbanization. This tends to change
the pattern of climatological-mean precipitation dis-
tribution as the urban cluster is situated among the
three centers of climatological-mean precipitation
amount over coastal South China and tends to connect
the centers.
2) Those urbanization-induced positive trends of the
extreme hourly rainfall amount and frequency
over PRD are found in the majority of stations
over the PRD when the TC-produced extreme
rainfall records are excluded. The trends are more
closely related to more frequent occurrence of the
short-duration (#6 h), abrupt rainfall events, than
the longer-duration, continuous, or growing rain-
fall events.
3) The 120 EXREs in 2011–16 consist of 67 (53) cases
with a strong (weak) UHI. They are categorized into
six types according to the originating location and
movement of the extreme-rain-producing storms,
namely, the local/SW wind (31.7%), local/shear
FIG. 16. As in Fig. 12, but for the migratory-SW type of the 2011–16 EXREs. The arrow in (c) and (d) denotes the
movement direction of the extreme-rain-producing storm.
1816 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 58
line (20.8%), and migratory-NW, migratory-SW,
migratory-NE, or migratory-S types (28.3%, 10.8%,
5.0%, 3.3%). Irrespective of synoptic conditions and
seasons (April–September) in which the EXREs
take place, rainfall intensification by the strong UHI
is noticeable in all the six types, especially over
the inland urban region where the UHI intensity is
strongest, because of the strong UHI-induced larger
CAPE to feed the rain storms. The UHI thermal
perturbation probably plays an important role in the
convective initiation and intensification of the locally
developed rain storms in the afternoon.
This study provides observed evidence of hourly
rainfall intensification over the PRD urban cluster,
especially its inland portion, in a large ensemble of
EXREs under various synoptic conditions, consistent
with the findings from the long-term trend analysis.
The results build a scientific basis for future in-depth
analysis and modeling studies to better understand the
individual and combined impacts of local urban envi-
ronment and global warming on the change of hourly
precipitation extremes over the coastal South China
during the past decades. Physical mechanisms about
the impacts of urban cluster on evolution of extreme-
rain-producing storms (e.g., the UHI-induced destabi-
lization, anthropogenic aerosol emissions) can also
be better understood using high-resolution modeling
experiments.
Acknowledgments. This research is supported by
National (Key) Basic Research and Development
Program of China (2018YFC1507400), The National
Natural Science Foundation of China (41775050), and
the Basic Research and Operation Funding of Chi-
nese Academy of Meteorological Sciences (CAMS)
(2017Z006). We also acknowledge the support from
the National Center for Atmospheric Research
(NCAR) Water System and the USDA-NIFA Agri-
culture and Food Research Initiative (Awards 2015-
67003-23508 and 2015-67003-23460). We thank Dr.
FuminRen(CAMS)forhelponusingOSATto
identify the TC-induced precipitation and Prof. Da-
Lin Zhang (University of Maryland) for helpful dis-
cussions. The land-use data were downloaded from
the Resources and Environment Scientific Data Cen-
ter, Chinese Academy of Sciences (http://www.resdc.cn/
data.aspx?DATAID598).
FIG. 17. As in Fig. 12, but for the (a),(b) migratory-NE type and (c),(d) migratory-S type of the 2011–16 EXREs.
Both types only have strong-UHI events. The arrow in (c) and (d) denotes the movement direction of the extreme-
rain-producing storm.
AUGUST 2019 W U E T A L . 1817
REFERENCES
Alexander, L. V., and J. M. Arblaster, 2009: Assessing trends in
observed and modelled climate extremes over Australia in
relation to future projections. Int. J. Climatol.,29, 417–435,
https://doi.org/10.1002/joc.1730.
Bell, T. L., D. Rosenfeld, and K.-M. Kim, 2009: Weekly cycle of
lightning: Evidence of storm invigoration by pollution. Geophys.
Res. Lett.,36, L23805, https://doi.org/10.1029/2009GL040915.
Bornstein, R. D., and Q. Lin, 2000: Urban heat islands and sum-
mertime convective thunderstorms in Atlanta: Three case
studies. Atmos. Environ.,34, 507–516, https://doi.org/10.1016/
S1352-2310(99)00374-X.
Chang, C. P., Y. Lei, C.-H. Sui, X. Lin, and F. Ren, 2012: Tropical
cyclone and extreme rainfall trends in East Asian summer
monsoon since mid-20th century. Geophys. Res. Lett.,39,
L18702, https://doi.org/10.1029/2012GL052945.
Chen, X., F. Zhang, and K. Zhao, 2016: Diurnal variations of the
land–sea breeze and its related precipitation over South
China. J. Atmos. Sci.,73, 4793–4815, https://doi.org/10.1175/
JAS-D-16-0106.1.
Craig, K., and R. Bornstein, 2002: MM5 simulations of urban in-
duced convective precipitation over Atlanta. Preprints, Fourth
Conf. on the Urban Environment, Norfolk, VA, Amer. Meteor.
Soc., 1.3, https://ams.confex.com/ams/AFMAPUE/techprogram/
paper_38803.html.
Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Con-
figuration and performance of the data assimilation system. Quart.
J. Roy. Meteor. Soc.,137, 553–597, https://doi.org/10.1002/qj.828.
Ding, Y. H., 1994: Monsoons over China. Kluwer Academic Pub-
lishers, 419 pp.
Donat, M. G., A. L. Lowry, L. V. Alexander, P. A. O’Gorman, and
N. Maher, 2016: More extreme precipitation in the world’s dry
and wet regions. Nat. Climate Change,6, 508–513, https://
doi.org/10.1038/nclimate2941.
Dou, J., Y. Wang, R. Bornstein, and S. Miao, 2015: Observed
spatial characteristics of Beijing urban climate impacts on
summer thunderstorms. J. Appl. Meteor. Climatol.,54, 94–105,
https://doi.org/10.1175/JAMC-D-13-0355.1.
Du, Y., Q. Zhang, Y. Chen, Y. Zhao, and X. Wang,2014: Numerical
simulations of spatial distributions anddiurnal variations of low-
level jets in China during early summer. J. Climate,27, 5747–
5767, https://doi.org/10.1175/JCLI-D-13-00571.1.
Gong, P., and Coauthors, 2013: Finer resolution observation and
monitoring of global land cover: First mapping results with
Landsat TM and ETM1data. Int. J. Remote Sens.,34, 2607–
2654, https://doi.org/10.1080/01431161.2012.748992.
Grimm, N. B., S. H. Faeth, N. E. Golubiewski, C. L. Redman,
J. Wu, X. Bai, and J. M. Briggs, 2008: Global change and the
ecology of cities. Science,319, 756–760, https://doi.org/10.1126/
science.1150195.
Guangdong Statistical Bureau, 2016: Statistical Yearbook of
Guangdong Province (updated yearly).Accessed 15 November
2017, http://stats.gd.gov.cn/gdtjnj/index.html.
Guo, J., and Coauthors, 2017: Declining frequency of summertime
local-scale precipitation over eastern China from 1970 to 2010
and its potential link to aerosols. Geophys. Res. Lett.,44, 5700–
5708, https://doi.org/10.1002/2017GL073533.
Guo, X., D. Fu, and J. Wang, 2006: Mesoscale convective pre-
cipitation system modified by urbanization in Beijing city. Atmos.
Res.,82, 112–126, https://doi.org/10.1016/j.atmosres.2005.12.007.
Hallegatte, S., C. Green, R. J. Nicholls, and J. Corfee-Morlot, 2013:
Future flood losses in major coastal cities. Nat. Climate
Change,3, 802–806, https://doi.org/10.1038/nclimate1979.
Hjelmfelt, M. R., 1982: Numerical simulation of the effects of
St. Louis on mesoscale boundary-layer airflow and verti-
cal air motion: Simulations of urban vs. nonurban effects.
J. Appl. Meteor.,21, 1239–1257, https://doi.org/10.1175/
1520-0450(1982)021,1239:NSOTEO.2.0.CO;2.
Huang,L.,Y.Luo,andD.-L.Zhang, 2018: The relationship
between anomalous presummer extreme rainfall over
south China and synoptic disturbances. J. Geophys. Res.
Atmos.,123, 3395–3413, https://doi.org/10.1002/2017JD028106.
Huang, Y., Y. Liu, Y. Liu, H. Li, and J. C.Knievel, 2019: Mechanisms
for a record-breaking rainfall in the coastal metropolitan city of
Guangzhou, China: Observation analysis and nested very large
eddy simulation with the WRF Model. J. Geophys. Res.,124,
1370–1391, https://doi.org/10.1029/2018JD029668.
Huff, F. A., and J. L. Vogel, 1978: Urban, topographic, and diurnal
effects on rainfall in the St. Louis region. J. Appl. Meteor.,17,
565–577, https://doi.org/10.1175/1520-0450(1978)017,0565:
UTADEO.2.0.CO;2.
Jiang, Z. H., F. Huo, H. Ma, J. Song, and A. Dai, 2017: Impact of
Chinese urbanization and aerosol emissions on the East Asian
summer monsoon. J. Climate,30, 1019–1039, https://doi.org/
10.1175/JCLI-D-15-0593.1.
Jiang, Z. N., D.-L. Zhang, R. Xia, and T. Qian, 2017: Diurnal vari-
ations of presummer rainfall over southern China. J. Climate,
30, 755–773, https://doi.org/10.1175/JCLI-D-15-0666.1.
Jin, M., and J. M. Shepherd, 2008: Aerosol relationships to warm
season clouds and rainfall at monthly scales over east China:
Urban land versus ocean. J. Geophys. Res.,113, D24S90,
https://doi.org/10.1029/2008JD010276.
Kendall, M. G., 1975: Rank Correlation Methods. Griffin & Co., 196
pp.
Landsberg, H. E., 1981: The Urban Climate. Academic Press, 275
pp.
Li, C. Y., and W. Zhou, 2015: Interdecadal changes in summertime
tropical cyclone precipitation over southeast China during
1960–2009. J. Climate,28, 1494–1509, https://doi.org/10.1175/
JCLI-D-14-00246.1.
——, ——, C. M. Shun, and T. C. Lee, 2017: Change in de-
structiveness of landfalling tropical cyclones over China in
recent decades. J. Climate,30, 3367–3379, https://doi.org/
10.1175/JCLI-D-16-0258.1.
Li, J., R. Yu, and T. Zhou, 2008: Seasonal variation of the diurnal
cycle of rainfall in southern contiguous China. J. Climate,21,
6036–6043, https://doi.org/10.1175/2008JCLI2188.1.
Li, W., S. Chen, G. Chen, W. Sha, C. Luo, Y. Feng, Z. Wen, and
B. Wang, 2011: Urbanization signatures in strong versus weak
precipitation over the Pearl River delta metropolitan regions
of China. Environ. Res. Lett.,6, 034020, https://doi.org/
10.1088/1748-9326/6/3/034020.
Li, Z., X. Deng, F. Wu, and S. S. Hasan, 2015: Scenario analysis for
water resources in response to land use change in the middle
and upper reaches of the Heihe River Basin. Sustainability,7,
3086–3108, https://doi.org/10.3390/su7033086.
Liang, P., and Y. Ding, 2017: The long-term variation of extreme
heavy precipitation and its link to urbanization effects in
Shanghai during 1916–2014. Adv. Atmos. Sci.,34, 321–334,
https://doi.org/10.1007/s00376-016-6120-0.
Liu, J., and Coauthors, 2014: Spatiotemporal characteristics, pat-
terns, and causes of land-use changes in China since the late
1980s. J. Geogr. Sci.,24, 195–210, https://doi.org/10.1007/
s11442-014-1082-6.
Liu, X., Y. Luo, Z. Guan, and D.-L. Zhang, 2018: An extreme
rainfall event in coastal South China during SCMREX-2014:
1818 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 58
Formation and roles of rainband and echo trainings. J. Geo-
phys. Res. Atmos.,123, 9256–9278, https://doi.org/10.1029/
2018JD028418.
Lomax, R. G., and D. L. Hahs-Vaughn, 2007: Statistical Concepts:
A Second Course. 532 pp.
Luo, Y., M. Wu, F. Ren, J. Li, and W. Wong, 2016: Synoptic situ-
ations of extreme hourly precipitation over China. J. Climate,
29, 8703–8719, https://doi.org/10.1175/JCLI-D-16-0057.1.
Mann, H. B., 1945: Nonparametric tests against trend. Econo-
metrica,13, 245–259, https://doi.org/10.2307/1907187.
Miao, S., F. Chen, Q. Li, and S. Fan, 2011: Impacts of urban pro-
cesses and urbanization on summer precipitation: A case study
of heavy rainfall in Beijing on 1 August 2006. J. Appl. Meteor.
Climatol.,50, 806–825, https://doi.org/10.1175/2010JAMC2513.1.
Min, S. K., X. Zhang, F. W. Zwiers, and G. C. Hegerl, 2011: Human
contribution to more-intense precipitation extremes. Nature,
470, 378–381, https://doi.org/10.1038/nature09763.
Niyogi, D., P. Pyle, M. Lei, S.P. Arya, C. M. Kishtawal, M. Shepherd,
F. Chen, and B. Wolfe, 2011: Urban modification of thunder-
storms—An observational storm climatology and model case
study for the Indianapolis urban region. J. Appl. Meteor. Cli-
matol.,50, 1129–1144, https://doi.org/10.1175/2010JAMC1836.1.
Ntelekos, A. A., J. A. Smith, L. Donner, J. D. Fast, W. I. Gustafson
Jr., E. G. Chapman, and W. F. Krajewski, 2009: The effects of
aerosols on intense convective precipitation in the northeast-
ern United States. Quart. J. Roy. Meteor. Soc.,135, 1367–1394,
https://doi.org/10.1002/qj.476.
Oke, T. R., 1987: Boundary Layer Climates. 2nd ed. Methuen Co.,
435 pp.
Pfahl, S., P. A. O’Gorman, and E. M. Fischer, 2017: Understanding
the regional pattern of projected future changes in extreme
precipitation. Nat. Climate Change,7, 423–427, https://doi.org/
10.1038/nclimate3287.
Ramage, C. S., 1952: Variation of rainfall over South China
through the wet season. Bull. Amer. Meteor. Soc.,33, 308–311,
https://doi.org/10.1175/1520-0477-33.7.308.
Ren, F., G. Wu, W. Dong, X. Wang, Y. Wang, W. Ai, and W. Li,
2006: Changes in tropical cyclone precipitation over China.
Geophys. Res. Lett.,33, L20702, https://doi.org/10.1029/
2006GL027951.
——, Y. Wang, X. Wang, and W. Li, 2007: Estimating tropical
cyclone precipitation from station observations. Adv. Atmos.
Sci.,24, 700–711, https://doi.org/10.1007/s00376-007-0700-y.
——, J. Liang, G. Wu, W. Dong, and X. Yang, 2011: Reliability
analysis of climate change of tropical cyclone activity over the
western North Pacific. J. Climate,24, 5887–5898, https://
doi.org/10.1175/2011JCLI3996.1.
Ren, G., and Y. Zhou, 2014: Urbanization effect on trends of ex-
treme temperature indices of national stations over Mainland
China, 1961–2008. J. Climate,27, 2340–2360, https://doi.org/
10.1175/JCLI-D-13-00393.1.
Rosenfeld, D., 2000: Suppression of rain and snow by urban and
industrial air pollution. Science,287, 1793–1806, https://
doi.org/10.1126/science.287.5459.1793.
Seneviratne, S. I., and Coauthors, 2012: Changes in climate ex-
tremes and their impacts on the natural physical environment.
Managing the Risks of Extreme Events and Disasters to Ad-
vance Climate Change Adaptation, C. B. Field et al., Eds.,
Cambridge University Press, 109–230.
——, M. G. Donat, A. J. Pitman, R. Knutti, and R. L. Wilby, 2016:
Allowable CO
2
emissions based on regional and impact-
related climate targets. Nature,529, 477–483, https://doi.org/
10.1038/nature16542.
Shepherd, J. M., 2005: A review of current investigations of urban-
induced rainfall and recommendations for the future. Earth
Interact.,9,https://doi.org/10.1175/EI156.1.
——, and S. J. Burian, 2003: Detection of urban-induced rainfall
anomalies in a major coastal city. Earth Interact.,7,https://
doi.org/10.1175/1087-3562(2003)007,0001:DOUIRA.2.0.CO;2.
——, H. Pierce,and A. J. Negri, 2002:Rainfall modification by major
urban areas: Observations from spaceborne rain radar on the
TRMM satellite. J. Appl. Meteor.,41, 689–701, https://doi.org/
10.1175/1520-0450(2002)041,0689:RMBMUA.2.0.CO;2.
Siler, N., and G. Roe, 2014: How will orographic precipitation re-
spond to surface warming? An idealised thermodynamic
perspective. Geophys. Res. Lett.,41, 2606–2613, https://
doi.org/10.1002/2013GL059095.
UN Department of Economic and Social Affairs, Popula-
tion Division, 2011: Comprehensive tables. Vol. I, World
Population Prospects: The 2010 Revision, United Nations,
ST/ESA/SER.A/313, 503 pp., https://www.un.org/en/development/
desa/population/publications/pdf/trends/WPP2010/WPP2010_
Volume-I_Comprehensive-Tables.pdf.
Wang, X. H., P. F. Xiao, X. Feng, and H. Li, 2013: Extraction of
large-scale urban area information in China using DMSP/OLS
nighttime light data. Remote Sens. Land Resour.,25, 159–164,
https://doi.org/10.6046/gtzyyg.2013.03.26.
Wang, H., Y. L. Luo, and B. J.-D. Jou, 2014: Initiation, mainte-
nance, and properties of convection in an extreme rainfall
event during SCMREX: Observational analysis. J. Geophys.
Res.,119, 13 206–13 232, https://doi.org/10.1002/2014JD022339.
Westra, S., L. V. Alexander, and F. W. Zwiers, 2013: Global in-
creasing trends in annual maximum daily precipitation. J. Cli-
mate,26, 3904–3918, https://doi.org/10.1175/JCLI-D-12-00502.1.
Wu, M., and Y. Luo, 2016: Mesoscale observational analysis of
lifting mechanism of a warm-sector convective system pro-
ducing the maximal daily precipitation in China mainland
during pre-summer rainy season of 2015. J. Meteor. Res.,30,
719–736, https://doi.org/10.1007/s13351-016-6089-8.
Yang, P., G. Y. Ren, and P. Yan, 2017: Evidence for a strong as-
sociation of short-duration intense rainfall with urbanization
in the Beijing urban area. J. Climate,30, 5851–5870, https://
doi.org/10.1175/JCLI-D-16-0671.1.
Yu, Q., 2007: Inter-annual variability of precipitation urbanization
effects in Beijing. Prog. Nat. Sci.,17, 632–638.
Yu, R., T. Zhou, A. Xiong, Y. Zhu, and J. Li, 2007: Diurnal
variations of summer precipitation over contiguous China.
Geophys. Res. Lett.,34, L01704, https://doi.org/10.1029/
2006GL028129.
Yuan, W., R. Yu, M. Zhang, W. Lin, H. Chen, and J. Li, 2012:
Regimes of diurnal variation of summer rainfall over sub-
tropical East Asia. J. Climate,25, 3307–3320, https://doi.org/
10.1175/JCLI-D-11-00288.1.
Zhang, C. L., F. Chen, S. G. Miao, Q. C. Li, X. A. Xia, and C. Y.
Xuan, 2009: Impacts of urban expansion and future green
planting on summer precipitation in the Beijing metropolitan
area. J. Geophys. Res.,114, D02116, https://doi.org/10.1029/
2008JD010328.
Zhang,X.,F.W.Zwiers,G.Li,H.Wan,andA.J.Cannon,2017:
Complexity in estimating past and future extreme short-duration
rainfall. Nat. Geosci.,10, 255–259, https://doi.org/10.1038/ngeo2911.
Zheng, Y., M. Xue, B. Li, J. Chen, and Z. Tao, 2016: Spatial
characteristics of extreme rainfall over China with hourly
through 24-hour accumulation periods based on national-level
hourly rain gauge data. Adv. Atmos. Sci.,33, 1218–1232,
https://doi.org/10.1007/s00376-016-6128-5.
AUGUST 2019 W U E T A L . 1819
