This article appeared in a journal published by Elsevier. The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy
Journal of Hydrology 377 (2009) 274–283
Contents lists available at ScienceDirect
Journal of Hydrology
journal homepage: www.elsevier.com/locate/jhydrol
Abrupt behaviors of the streamflow of the Pearl River basin and implications
for hydrological alterations across the Pearl River Delta, China
Qiang Zhang a,*, Chong-Yu Xu b, Yongqin David Chen c, Jianmin Jiang d
a
Department of Water Resources and Environment, Sun Yat-sen University, Guangzhou 510275, China
Department of Geosciences, University of Oslo, P.O. Box 1047 Blindern, N-0316 Oslo, Norway
c
Department of Geography and Resource Management, The Chinese University of Hong Kong, Shatin N.T., Hong Kong, China
d
Beijing Meteorological College, Beijing 100081, China
b
a r t i c l e
i n f o
Article history:
Received 28 April 2009
Received in revised form 18 August 2009
Accepted 20 August 2009
This manuscript was handled by G. Syme,
Editor-in-Chief
Keywords:
Two-phase regression scheme
Abrupt variations
Hydrological variations
Streamflow ratio
Pearl River Delta
s u m m a r y
In this study, we analyze the long streamflow series of three hydrological stations of the lower Pearl River
basin and the streamflow ratio between Makou and Sanshui stations by using statistical techniques. Furthermore, we also attempt to address influences of precipitation and human activities (human-induced
deepening of river channels) on streamflow ratio. The results indicate that: (1) the streamflow variations
show remarkable relations with precipitation changes in West and East River basins, implying tremendous influences of climate changes on hydrological processes. Decreasing precipitation was observed
in North River basin. However, the streamflow amount of the Sanshui station largely increased due to
enlarged streamflow allocation from the West River to the North River; (2) increasing streamflow ratio
of Sanshui/(Makou + Sanshui) is the result of morphological changes (downcut) of river channels in the
upper Pearl River Delta. The fast downcut of river channels is mainly due to intensive sand mining. Larger
magnitude of increase in streamflow ratio corresponds well to the higher intensity of in-channel sand
dredging; (3) after late-1990s, decreasing precipitation of the Pearl River basin abates the streamflow
amount and also the streamflow ratio. The influences of human activities and climate changes are varying
in different time intervals and in different river basins. Due to tremendous impacts of increased streamflow ratio between Sanshui and Makou station, relations between streamflow and precipitation relations
in the North River basin are not statistically good. This study helps to improve understandings of the
causes underlying altered streamflow variations in the lower Pearl River basin and the hydrological alterations within the Pearl River Delta region.
Ó 2009 Elsevier B.V. All rights reserved.
Introduction
River deltas are heavily populated though contribute much to
the socioeconomic development of human society (Ericson et al.,
2006; Pont et al., 2002). Accordingly, river deltas receive rising
concerns from hydrologists, fluvial geomorphologists, policymakers, ecologists especially during last decades. Various aspects
of the river deltas, such as wetland ecological environment, are
considerably altered as a result of booming economic development. Intensifying human activities in the river deltas also greatly
influenced the morphological properties of the deltas, altered both
hydrological processes and ecological environment (e.g. Bott et al.,
2006). This is particularly the case for the Pearl River Delta (PRD)
(Chen et al., 2008). The Pearl River Delta is characterized by the
complicated crisscross river network with density of 0.68–
1.07 km/km2, being the fastest developing region in China since
* Corresponding author. Tel./fax: +86 20 84113730.
E-mail address: zhangqnj@gmail.com (Q. Zhang).
0022-1694/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.jhydrol.2009.08.026
the country adopted the ‘‘open door and market reform” policy
in the late 1970s. Because of the booming economy and intensifying human activities, new environmental problems emerged in
recent decades in the Pearl River Delta, such as floods, salinity
intrusion and storm surge. All these environmental problems could
be partly attributed to altered hydrological processes within the
river channels across the Pearl River Delta. The altered hydrological
conditions are mainly represented by abnormally high flood stage
in flooding seasons and more frequent salinity intrusion events in
the dry seasons. The hydrological alterations are the integrated
consequences of intensive channel dredging, sand mining and
levee construction, etc. which took place mainly after 1980s (e.g.
Zeng et al., 1992; Chen and Chen, 2002; Liu et al., 2003), and posed
new challenges for the regional water resource management.
The water level variations and underlying causes have been
widely discussed and related reports can be found in Chinese
journals (e.g. Zeng et al., 1992; Huang et al., 2000; Luo et al.,
2000; Yang et al., 2002; Liu et al., 2003; Chen et al., 2004). Many
studies (e.g. Chen and Chen, 2002) indicated that significant
Author's personal copy
Q. Zhang et al. / Journal of Hydrology 377 (2009) 274–283
hydrological alteration occurred around 1992. The dynamic balance between scouring and filling processes of the river channel
was broken after mid-1980s because of intensive human activities
such as in-channel dredging, construction of hydraulic facilities, as
the direct results of these human activities are fast down-cutting of
riverbeds of the upper Pearl River Delta. Changes of the geometrical shapes of the river channels brought about an altered streamflow ratio between North River and West River. Larger
magnitude of cutting-down of the river channel of the lower North
River when compared to the lower West River leads to more
streamflow transferring from the West River to the North River
as measured by the Sanshui/(Makou + Sanshui) streamflow ratio.
In general, the Sanshui/(Makou + Sanshui) ratio is increasing,
which directly raise the amount of water in the hinterland of the
Pearl River Delta and result in higher risk of flood inundation in
the flooding season. The changed streamflow ratio was seen as
one of the major factors causing alterations of the water levels
across the PRD (Zhou et al., 2001; Chen and Chen, 2002; Luo
et al., 2007).
Huang and Zhang (2004) also suggested that riverine streamflow from the upper PRD heavily influenced water level behaviors
within the river network of the PRD. It should be accepted that
former researches benefit a good understanding of the water level
changes and associated underlying reasons in the study region. The
hydrological series analyzed in the former studies, however, are
usually short and cannot reflect hydrological conditions of recent
years. Besides, short and intermittent streamflow series can not
well reflect dynamic processes of the streamflow variations. Therefore, human influences on streamflow ratio and hydrological alterations are by no means well addressed or understood. Besides,
most of former reports attached overwhelming importance to
human activities, and less attention has been paid on influences
of climate changes on hydrological processes. Furthermore, in
recent years, hydrological alterations caused abnormally high
water level and more frequent salinity intrusion events in the
PRD, posing new challenges for the management of water resource
and natural hazards. Knowledge of hydrological processes from the
upper Pearl River Delta and the underlying causes in terms of
human activities and climate changes will be greatly helpful for
the development of improved understanding of water level alterations within the PRD, which has the potential to enhance human
275
mitigation to natural hazards and effective water resource management. In this sense, we will investigate hydrological properties
of the upper PRD with updated long hydrological series and
improved statistical techniques. This is the major motivation of
this study.
Therefore, the objectives of this paper are: (1) to investigate
abrupt behaviors and trends of streamflow series of three hydrological stations, i.e. Makou station for the West River basin, Sanshui
station for the North River basin and Boluo for the East River basin;
(2) to explore the abrupt behaviors of streamflow ratio of the
Sanshui station to the total streamflow amount of Sanshui and
Makou stations and also the underlying causes; (3) to analyze
implications of these hydrological properties for the water
resource management and the hydrological alterations within the
PRD. The novelty of this study lies in the following points: (1) we
will analyze the longest possible monthly hydrological series; (2)
we will quantitatively evaluate abrupt behaviors of streamflow
variations and relate these changes to human activities and climate
changes; (3) besides, we will further improve the simple twophase linear regression scheme based on the work by Lund and
Reeves (2002). The modified method can help to investigate abrupt
changes of the considered time series on different time scales,
providing the potential to differentiate various factors exerting
influences on hydrological processes on different time scales. As
such, this study will be helpful to improve our understandings of
statistical properties of hydrological processes from the upper
PRD under both human activities and climatic changes, to uncover
associated implications for the hydrological alterations, water resource management and human mitigation to natural hazards in
the study region.
Data
The monthly hydrological data (unit: m3/s) analyzed in this
study were extracted from three hydrological stations of the West
River basin, the North River basin and the East River basin, i.e.
Makou station, Sanshui station and Boluo station. The hydrological
data of Makou and Sanshui station cover January 1959 to December 2005 and that of the Boluo station cover January 1954 to
December 2002. The hydrologic data before 1989 are extracted
Fig. 1. Location of the study region and the hydrological stations.
Author's personal copy
276
Q. Zhang et al. / Journal of Hydrology 377 (2009) 274–283
from the Hydrological Year Book (published by the Hydrological
Bureau of the Ministry of Water Resources of China) and those after
1989 are provided by the Water Bureau of Guangdong Province.
The quality of the hydrological data was firmly controlled before
its release. The hydrological series are free of missing data. Our
previous studies of the precipitation changes in the Pearl River
basin indicated a relatively homogeneous spatial distribution of
precipitation variations (e.g. Zhang et al., 2008a). To match the
time interval the hydrological data covers, we extracted precipitation data covering the period of 1951–2005 from the 160 rain
gauging stations (Gemmer et al., 2004; Zhang et al., 2008b). The
data are collected from the National Climatic Centre (NCCC) of
the China Meteorological Administration (CMA). Locations of these
stations have been indicated in Fig. 1. The quality of the hydrometeorological dataset used in this study is firmly controlled
before its release (see Zhang et al., 2008b).
The denominators in (2) can be explicitly evaluated as:
j1
X
ðt t 1 Þ2 ¼
t¼jn
¼
ðj 1Þjðj 2Þ
12
and
jþn1
X
ðt t 2 Þ2
t¼j
ðn j þ 1Þðn j þ 2Þðn jÞ
12
ð4Þ
Under the null hypothesis of no change points, the regression
parameters during the two phases must equal, i.e. a1 = a1 and
^1 a
^ 2 should be close to zero for each
^1 l
^ 2 and a
l1 = l2. If so, l
sub-samples divided by j.
Rescaling this to a regression F statistic merely states that (Lund
and Reeves, 2002)
Fc ¼
ðSSERed SSEFull Þðn 4Þ
2SSEFull
ð5Þ
In (5), SSEFull is the ‘full model’ sum of squared errors computed
from
Methodology
First, we will briefly introduce the methodology used in the
study (Solow, 1987; Easterling and Peterson, 1995; Vincent,
1998; Lund and Reeves, 2002).
The model is written as:
Xt ¼
l1 þ a1 t1 þ et
l2 þ a2 t2 þ et
ð1Þ
Pj1
t¼jn ðt t 1 ÞðX t Pj1
2
t¼jn ðt t 1 Þ
j1
X
^ 1 tÞ2 þ
^1 a
ðX t l
jþn1
X
t¼jn
^ 2 tÞ2
^2 a
ðX t l
ð6Þ
t¼j
SSERed is the ‘reduced model’ sum of squared errors, which was
formulated as
SSERed ¼
jþn1
X
^ Red tÞ2
^ Red a
ðX t l
ð7Þ
t¼jn
In the work by Lund and Reeves (2002), t1 and t2 are defined as:
1 6 t1 6 c and c < t2 6 n respectively, where c is the possible change
point to be tested. l1 and l2 are the mean values for the hydrological series defined by t1 and t2, respectively. et is the error term.
With this original method, only the time when the abrupt change
occurs is decided. The modification we have made in this study
mainly lies in the ranges of t1 and t2. In our study, t1 = t2 is defined
as t1 = [jn, j1], t2 = [j, j + n1] respectively. The sub-sample size
n varies as n = 2, 3, . . . , < N/2, or may be selected at suitable intervals. The quantity j = n + 1, n + 2, . . . , Nn + 1 is the reference time
point. N is the length of the time series. Therefore, t1 and t2 are
changing. t1 and t2 here are the size of the window shifting along
the entire time series and are also taken as the time scales as mentioned in the following sections. So, we can see that this modified
method behaves in the similar way as the wavelet transform technique does. In this sense, the modification of the method is enlightened by the wavelet transform method. In so doing, a time point
when possible abrupt change occurs is decided with respect to a
certain time scale, i.e. t1, t2. Thus, comparatively, the original method just decides when the change point occurs.
The improved method, however, can both tell the time when
the change point occurs and also the related time scales the change
point occurs. In this way, mechanisms behind the abrupt behaviors
of the streamflow variations are expected to be exposed, as different influencing factors exert their impacts on the streamflow
changes on different time scales. The least squares estimates of
the trend parameters in (1) are:
a^ 1 ¼
SSEFull ¼
1Þ
X
Pjþn1
^2 ¼
and a
2Þ
ðt t 2 ÞðX t X
Pjþn1
2
ðt t 2 Þ
t¼j
t¼j
ð2Þ
1 and X
2 are the average series values before and after
In (2), X
time j, respectively. t 1 and t 2 are the average time observations
before and after time j, respectively. Least squares estimates of
the location parameters l1 and l2 in equation (1) are:
l^ 1 ¼ X 1 a^ 1t1 and l^ 2 ¼ X 2 a^ 2t2
ð3Þ
^ Red are estimated under the constraints a1 = a1
^ Red and a
where l
^ Red and l1 = l2 l
^ Red (Lund and Reeves, 2002). If a change point
=a
is present at time j1, Fc should be statistically larger when compared to the threshold value by F test. The effective degree of freedom (EffD) after the correction of dependence and in a normalized
distribution for the time series (Storch and Zwiers, 1999; Jiang
et al., 2007) can be estimated by
EffD ¼
INTð1 þ 2
2n
PINTðn=2Þ
s¼1
r X ðsÞr t ðsÞÞ
ð8Þ
where INT denotes the integer part of the number. r(s) is the autocorrelation with time lag of s. rX(s) is the autocorrelation function of
the entire hydrological series and rt(s) is the autocorrelation function of the hydrological series considered. After the effective degree
of freedom is known, the threshold value (Fth) can be obtained via
the F test table (Lund and Reeves, 2002). If Fc > Fth, then we can
say that the change point is statistically significant. This method
is used to detect significant change points and linear trends between the change points in time series. The figures in this study
were made by using the Surfer software package.
Results and discussion
Hydrological variations of the Pearl River basin
Fig. 2 illustrates the abrupt behaviors of streamflow series of the
Makou station (the upper panel) and trends of sub-series between
change points (the lower panel). X axis shows the time scales, i.e. t
(t1 or t2, because t1 = t2). Dashed contours denote turning points
from upward to downward trends and vice versa. Thick dashed
or solid contours indicate significant change points at >95% confidence level. More detailed descriptions can be found in the caption
of Fig. 2. The upper panel of Fig. 2 shows the shifts of change points
on different time scales. At time scales of <32 months, there are
many upward and downward trends interrupted by change points.
Only two change points are significant at >95% confidence level.
One change point occurs in about 1962, followed by downward
trend. Another significant change point occurs in about 1964,
Author's personal copy
277
Q. Zhang et al. / Journal of Hydrology 377 (2009) 274–283
Time scales (months)
256
Makou station
1.35
1.1
0.85
0.6
0.2
0
-0.15
-0.4
-0.65
-0.9
-1.15
-1.4
128
64
32
16
8
1960
6
Standardized streamflow
5
1965
1970
1975
1980
1985
1990
1995
2000
2005
1980
1985
1990
1995
2000
2005
Makou station
4
3
2
1
0
−1
−2
−3
1960
1965
1970
1975
Time (years)
Fig. 2. Trends estimation of streamflow variations of the Makou station. The upper panel shows change points on different time scales; and lower panel shows linear trends of
time intervals separated by change points. Dashed lines show decreasing trend after the change point, and solid lines indicate increasing trend after the change point. The
meanings of the line styles are the same for subsequent figures.
followed by upward trend. For the sake of better understanding of
abrupt changes (change points and trends between change points)
of stream series, we also plotted the standardized streamflow
series and trends interrupted by change points in the lower panel
of Fig. 2. Slight increasing trend can be identified between 1959
and 1962. After a decreasing trend in a shorter time interval,
increasing trend was observed between 1964 and 1966. On time
scales of >64 months, more than two change points with centers
circled by thick lines can be found.
The upper panel of Fig. 2 indicates that after 1983, the streamflow of the Makou station turns to be decreasing and increasing
trend can be detected after about 1990. The streamflow series
come to be decreasing after about 1994. To clearly demonstrate
these statistical properties, we illustrate the standardized streamflow series and also the trends between change points (see lower
panel of Fig. 2). After about 1966, the streamflow is increasing
and then turns into decreasing trend in 1983. After 1990, the
streamflow is increasing, and is decreasing after 1994. Therefore,
four time intervals were identified marked by increasing streamflow trends, while two time intervals by decreasing trends. These
time intervals can be observed from lower panel of Fig. 2. Generally, before mid-1980s, the streamflow variations of the Makou
station are dominated by increasing trends; after mid-1980s the
streamflow changes are characterized mainly by decreasing trends
(lower panel of Fig. 2). In addition, it can be found in the lower
panel of Fig. 2 that an abnormally high streamflow occurred in
1983 which was due to heavy rainfall induced by strong typhoon
activity. The flood event caused by this typhoon affected
67,000 ha of agricultural land.
To demonstrate the influence of precipitation variations on
streamflow changes, herein we will also analyze statistical proper-
ties of standardized areal average precipitation changes of the
West River basin. Spatial patterns of change points of precipitation
in the ‘time scale vs. time’ space (upper panel of Fig. 3) show similar characteristics to those of the streamflow variations of the
Makou station (upper panel of Fig. 2). F test results show no significant changes within the standardized areal average precipitation
series of the West River basin. Moreover, we also analyzed the
trends between time points identified by relatively higher threshold values for the sake of an easy comparison between streamflow
and precipitation changes. The visual comparison of the lower panels of Figs. 2 and 3 indicates that the streamflow variations largely
follow the precipitation changes during the last decades. Particularly the abnormally high precipitation in 1983 (lower panel of
Fig. 3) well corresponds to the high streamflow in 1983 observed
in the lower panel of Fig. 2. Thereby, the variations of streamflow
amount in Makou station are mainly influenced by precipitation
changes. Quantitative evaluation of relations between precipitation and streamflow for three river basins considered in this study
is conducted in the following section.
As for abrupt changes of streamflow of the Sanshui station
(Fig. 4), on time scales of <32 months, only one significant change
point is found. On time scales of >32 months, two regions are characterized by positive and negative threshold values, showing different trends after the change point. As shown in the lower panel
of Fig. 4, after 1964, the streamflow is increasing and turns to be
decreasing after 1984. After about 1990, the streamflow is increasing and turns to be decreasing after 1998. A comparison between
the lower panels of Figs. 2 and 4 implies similar changing properties in terms of streamflow of the Makou and the Sanshui stations,
i.e. increasing trends during mid-1960s to early 1980s and 1990 to
late 1990s; decreasing trends during early 1980s to 1990 and after
Author's personal copy
278
Q. Zhang et al. / Journal of Hydrology 377 (2009) 274–283
Precipitation in the West River basin
Time scales (months)
256
128
64
32
16
8
1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005
Standardized precipitation
4
0.55
0.45
0.35
0.25
0.15
0.05
-0.05
-0.15
-0.25
-0.35
-0.45
-0.55
-0.65
Precipitation in the West River basin
3
2
1
0
−1
−2
−3
1950
1960
1970
1980
1990
2000
2010
Time (years)
Fig. 3. Trends estimation of precipitation variations of the West River basin. The upper panel shows change points on different time scales; and lower panel shows linear
trends of time intervals separated by change points.
Time scales (months)
256
Sanshui station
128
64
32
16
8
1960
6
1965
1970
1975
1980
1985
1990
1995
2000
2005
1980
1985
1990
1995
2000
2005
2.7
2.4
2.1
1.8
1.5
1.2
0.9
0.6
0.3
0
-0.3
-0.6
-0.9
-1.2
-1.5
-1.8
Standardized streamflow
Sanshui station
4
2
0
−2
−4
1960
1965
1970
1975
Time (years)
Fig. 4. Trends estimation of streamflow variations of the Sanshui station. The upper panel shows change points on different time scales; and lower panel shows linear trends
of time intervals separated by change points.
late 1990s. Fig. 5 illustrates different changing patterns when compared to streamflow variations of the Sanshui station. The upper
panel of Fig. 5 indicates two significant change points on time
scales of <32 months, i.e. 1983 and 1992, followed by decreasing
precipitation. The lower panel of Fig. 5 clearly indicates general
decreasing trends of areal average precipitation of the North River
Author's personal copy
279
Q. Zhang et al. / Journal of Hydrology 377 (2009) 274–283
Precipitation in the North River basin
0.5
Time scales (months)
256
0.3
0.1
128
-0.1
64
-0.3
-0.5
32
-0.7
-0.9
16
-1.1
8
-1.3
1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005
6
Standardized precipitation
Precipitation in the North River basin
4
2
0
−2
−4
1960
1970
1980
1990
2000
Time (years)
Fig. 5. Trends estimation of precipitation variations of the North River basin. The upper panel shows change points on different time scales; and lower panel shows linear
trends of time intervals separated by change points.
basin. Two changes points occur in early 1980s and early 1990s.
Anti-phase relations between streamflow of the Sanshui station
and the precipitation amount in the North River basin were further
elucidated in the next sections.
Fig. 6 shows abrupt behaviors of streamflow of the Boluo station
on different time scales. At time scales of <32 months, the streamflow series is subdivided mostly by short intervals of decreasing
trends (the upper panel of Fig. 6). Four significant change points
are identified in mid-1960s, 1975, early 1980s, 1997 and 1999.
On time scales of >32 months, 1975–1985 can be taken as the transitional time interval between increasing and decreasing streamflow. After early 1990s, the streamflow changes are characterized
by increasing trends. The standardized streamflow series is subdivided into seven segments by significant changes points (the lower
panel of Fig. 6). It can be observed from the lower panel of Fig. 6
that the streamflow variations before mid-1980s are characterized
mainly by increasing trends, and decreasing trends can be found
during two time intervals, i.e. mid-1980s-early 1990s and 2000–
2005. From the perspective of large time scales of >64 months,
the streamflow variations of the Boluo station are featured by
increasing trends. As for the precipitation changes of the East River
basin, no abrupt changes are identified on the time scales of
>64 months. Two significant change points occur in early 1980s
and late 1990s (upper panel of Fig. 7). It can be seen from lower panel of Fig. 7 that general increasing trends are found during 1990–
1997. After 1997, the precipitation turns to be decreasing. More
complicated changing properties of streamflow variations when
compared to those of precipitation changes might be the result
of intensifying human interferences with the hydrological processes besides the influences of climatic changes (Chen et al.,
2008; Zhang et al., 2009a). To further understand the influences
of human activities and climate changes on the streamflow
changes, we analyze the statistical relations between areal average
precipitation of the West River (Fig. 8A), the North River (Fig. 8B)
and the East River (Fig. 8C). Fig. 8 demonstrates the good correlations between precipitation and streamflow for the West River basin and East River basin when compared to the East River basin.
Bad precipitation vs. streamflow relations in the North River basin
could be attributed mainly to the increasing streamflow ratio between Sanshui and Makou station as a result of human activities.
Therefore, the streamflow changes of the Makou station and the
Boluo station are mainly controlled by climate changes; streamflow changes of the Sanshui stations are mainly influenced by altered streamflow ratio as a result of deepened river channel
induced by intensive sand dredging.
Streamflow ratio between Makou and Sanshui stations
Changes of streamflow ratio between Makou and Sanshui station (Sanshui/(Makou + Sanshui)) have exerted tremendous influences on hydrological processes and were seen as an important
factor causing hydrological alteration across the PRD (Chen et al.,
2004, 2008). The upper panel of Fig. 9 indicates abrupt changes
of streamflow ratio on different time scales. Only one significant
change point is identified in 1960s on time scales of <32 months.
On time scales of >32 months, four time intervals can be seen as
transitional periods when the streamflow ratio changes from one
pattern to another: late 1970s, mid-1980s, mid-1990s and after
2000. We analyzed trends of streamflow ratio between change
points and the results are shown in the lower panel of Fig. 9.
Decreasing streamflow ratio can be found during 1959–1964,
1980–1983 and 1986–1999. Increasing streamflow ratio occurred
to the rest time segments within the study period. It can be observed that, during the time intervals considered in this study,
Author's personal copy
280
Q. Zhang et al. / Journal of Hydrology 377 (2009) 274–283
Time scales (months)
256
2
1.7
1.4
1
0.8
0.5
0.2
-0.1
-0.4
-0.7
-1
-1.3
-1.6
-1.9
-2.2
Boluo station
128
64
32
16
8
1955
6
1960
1965
1970
1975
1980
1985
1990
1995
2000
1965
1970
1975
1980
1985
1990
1995
2000
Boluo station
Standardized streamflow
5
4
3
2
1
0
−1
−2
1955
1960
Time (years)
Fig. 6. Trends estimation of streamflow variations of the Boluo station. The upper panel shows change points on different time scales; and lower panel shows linear trends of
time intervals separated by change points.
Precipitation in the East River basin
Time scales (months)
256
128
64
32
16
8
1
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
-1.2
-1.4
-1.6
1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005
5
Precipitation in the East River
Standardized precipitation
4
3
2
1
0
−1
−2
1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005
Time (years)
Fig. 7. Trends estimation of precipitation variations of the East River basin. The upper panel shows change points on different time scales; and lower panel shows linear
trends of time intervals separated by change points.
Author's personal copy
281
Q. Zhang et al. / Journal of Hydrology 377 (2009) 274–283
4
A
4
3
B
y=0.50x
R2=0.25
3
2
2
Standardized streamflow
1
1
y=0.86x
R2=0.74
0
0
−1
−1
6
0
1
2
3
−1
0
1
2
3
C
4
2
y=0.77x
R2=0.60
0
−1
0
1
2
3
4
5
Standardized precipitation
Fig. 8. Correlations between streamflow and areal average precipitation in the
West River basin (A), the North River basin (B) and the East River basin (C).
two longer time intervals are characterized by increasing streamflow flow ratio: 1964–1979 and 1984–1997. The increase magnitude of streamflow ratio during 1964–1979 is smaller than that
during 1984–1997. Visual evaluation of curves during 1984–1997
indicates even larger increase magnitude during 1993–1997 when
compared to the interval of 1984–1992.
Thereby, increasing streamflow of Sanshui station (the lower
panel of Fig. 4) can be largely attributed to the increasing streamflow ratio between Sanshui and Makou, causing anti-phase
Time scales (m onths)
256
relations between precipitation changes in the North River basin
(Fig. 5) and the streamflow changes of the Sanshui station
(Fig. 4). Decreasing streamflow of Makou station is the result of
increasing streamflow ratio after mid-1990s (Fig. 2). Increasing
streamflow flow ratio is largely attributed to down-cutting behaviors of riverbed in the upper PRD. Research results (Chen and Chen,
2002) indicated that fast and intensive down-cutting processes of
the river channels in the upper PRD occurred after mid-1980s.
Since 1985, the sand sediments dredged annually are about
0.05–0.06 billion cubic meters, which is the direct cause for the
intensive downcut of riverbed of the upper PRD (Chen and Chen,
2002). However, the amount of sediments dredged during 1990–
1993 accounted for more than half of the total sediment mined
during recent decades, which heavily influenced morphological
properties of the river channels of the upper PRD.
Study by Luo et al. (2007) indicated that, from 1986 to 2003,
about 0.87 billion cubic meters of sand were excavated, which
caused average down-cutting depths of 0.59–1.73 m, 0.34–
4.43 m, and 1.77–6.48 m in the main channels of the West River,
North River and East River, the major water systems in the PRD.
Before mid-1980s however, the scouring and filling processes of
the river channels are in dynamic balances or in slight deposition
processes. Sediment load dredged during 1980–1998 accounts for
total net sedimentation within 70–125 years (Huang and Zhang,
2006). The river channel in the upper PRD was greatly altered
due to in-channel dredging and levee construction after about
mid-1980s, resulting in decreasing water level (Lu et al., 2007),
which can be seen as the major reason for a significant increasing
streamflow ratio during early 1980s and mid-1990s (Fig. 9). The
lower panel of Fig. 9 indicates moderate variations of streamflow
ratio before mid-1980s. Larger magnitude of increase of streamflow ratio was observed after mid-1980s as a result of fast and
intensive downcut of river channels, showing considerable
influences of human activities, such as sand dredging, upon hydrological processes. Even larger magnitude of increase in streamflow
5.5
Sanshui/(Sanshui+Makou)
4.5
3.5
2.5
128
1.5
0.5
64
-0.4
-1
32
-2
16
-3
-4
8
1960
4
1965
1970
1975
1980
1985
1990
1995
2000
2005
1980
1985
1990
1995
2000
2005
-5
Sanshui/(Sanshui+Makou)
Streamflow ratio
3
2
1
0
−1
−2
−3
1960
1965
1970
1975
Time (years)
Fig. 9. Trends estimation of Sanshui/(Makou + Sanshui) streamflow ratio variations. The upper panel shows change points on different time scales; and lower panel shows
linear trends of time intervals separated by change points.
Author's personal copy
282
Q. Zhang et al. / Journal of Hydrology 377 (2009) 274–283
ratio changes was observed after 1993. After 1996, the amount of
sand sediments dredged was decreasing. Fig. 9 also indicates
decreasing streamflow ratio after 1996. Therefore, these results
and observations imply the influences of riverbed downcut on
streamflow ratio between Sanshui and Makou stations. It should
be noted that, after late-1990s, the streamflow amount from the
Pearl River basin is decreasing, which is largely because of decreasing precipitation of the Pearl River basin in the period (Figs. 3 and
5).
The streamflow ratio of Sanshui/(Sanshui + Makou) is also
decreasing, partly for the reasons of the reduced sand sediment
dredged from the river channel and partly because of the decreasing streamflow of the West River basin. Changed streamflow ratio
between Sanshui and Makou greatly altered the water allocation
between hinterland of the Pearl River basin and West River channel. Our previous studies (e.g. Chen et al., 2009) indicated that
the time when the change points of water level changes occur
match well those of the streamflow ratio revealed in this study,
i.e. early 1980s and early 1990s. Lower panel of Fig. 9 shows larger
changing magnitude of streamflow after early 1990s, which can
well explain the hydrological alterations starting at early 1990s.
With ‘range of variability approach’ (RVA) approach, we also advocated the considerable influences of changed streamflow ratio on
hydrological processes, particularly the water level variations
across the Pearl River Delta (Zhang et al., 2009b).
Generally, human activities such as in-channel sand dredging
and climatic changes work together to trigger altered streamflow
allocation between Sanshui and Makou station by changing hypsographical properties of the river channel, and the changed streamflow ratio or water allocation further lead to or intensify the
hydrological alterations across the Pearl River Delta region (Chen
et al., 2008; Zhang et al., 2009b). This is the ripple effect in terms
of the influences of human activities and climate changes on the
hydrological processes. The aforementioned analyses can be seen
as the implication of the changing streamflow ratio between Sanshui and Makou station for the hydrological alterations within
the Pearl River Delta region.
Conclusions
In recent years, altered hydrological processes across the Pearl
River Delta were observed which are represented by abnormal
high water level in the hinterland in flooding season and lower
water level in winters. The direct consequences of these hydrological alterations are higher risk of flood inundation in flooding seasons and more difficult human withdrawal of fresh water resource
due to more frequent salinity intrusion in the dry seasons. Changed
streamflow ratio between Sanshui and Makou stations could be
seen as the major cause for the hydrological alterations. In this
study, we quantitatively evaluated abrupt changes and trends of
streamflow variations and also those of streamflow ratio of Sanshui/(Sanshui + Makou) based on long monthly hydrological data
at the Makou, Sanshui and Boluo stations. We obtained some interesting and important conclusions based on the aforementioned
analysis:
(1) The modified simple two-phase linear regression scheme in
this study can well reveal abrupt changes of hydrological
series on different time scales. The hydrological processes
are influenced by more than one factor and these influencing
factors tend to alter the changing properties of hydrological
series in terms of trend and abrupt behaviors. The improved
method, when compared to the original two-phase linear
regression scheme, is greatly helpful to deeply investigate
mechanisms behind hydrological alterations, holding the
potential to differentiate various influencing factors having
impacts on hydrological changes. This point is also one of
the main contributions of this study.
(2) Generally, the streamflow changes are heavily controlled by
the precipitation variations except the North River basin.
The precipitation of the North River basin is decreasing
and the streamflow of the Sanshui station is increasing. This
is mainly due to changed allocation of water between Sanshui and Makou stations. Therefore, in the lower Pearl River
basin, intensifying human activities have more influences, to
a certain degree, on streamflow changes than the climate
changes do. This conclusion comes up with challenges for
the water resource management under the fast changing
environment, and it is particularly the case for the lower
Pearl River Delta, one of the highly developed regions in
China.
(3) Massive sand sediments dredged from the river channels
resulted in considerable downcut of riverbed, leading to
increased streamflow ratio of Sanshui/(Makou and Sanshui),
and it is particularly true for the period of 1980–1998. Intensity of sand dredging in different periods is in good line with
changing properties of streamflow ratio. Before mid-1980s,
the scouring and filling processes are nearly in balance.
The streamflow ratio is in moderate variations. Larger magnitude of increase of streamflow ratio occurred after mid1980 when compared to that before mid-1980s which is
mainly due to intensified sand mining, and this is particularly the case after 1993. After mid-1990s, reduced sand
mining caused decreasing streamflow ratio. Decreasing
streamflow of Makou station and Sanshui station may also
be due to the reduced streamflow ratio. Therefore, we can
conclude that the hydrological alterations within the Pearl
River basin are mostly the results of human activities such
as sand minding, hydraulic facilities and so on. Precipitation
changes may also make its contribution to the hydrological
alterations. Even so, our study implies that influences of
human activities on changing properties of streamflow series in the lower Pearl River basin are larger.
(4) The precipitation of the Pearl River basin is decreasing after
late 1990s, so does the streamflow amount and streamflow
allocation between the West River and the North River.
Altered streamflow allocation was seen as a major cause
for the hydrological alterations across the Pearl River Delta,
and alteration of streamflow allocation was the integrated
consequence of human activities and climate variations.
These two factors exerted varying influences in different
time intervals and on different time scales. Altered streamflow allocation between Sanshui and Makou intensifies the
hydrological alterations of water level within the Pearl River
Delta. Furthermore, the impacts various influencing factors
have on hydrological processes in the lower Pearl River
basin, including the Pearl River Delta region, are varying in
different river channels because of extremely complicated
hydrological dynamic processes in the river networks in
the Pearl River Delta. Our previous studies (Chen et al.,
2008; Zhang et al., 2009b) investigated hydrological alterations in terms of water level changes and associated relations to hydrological processes from the upper Pearl River
Delta and human activities (mainly the in-channel sand
dredging). This study further addressed the changing properties of streamflow allocation between North River and
the West River and underlying causes such as climate
changes and human activities, which greatly helps to deeply
understand the mechanisms behind the hydrological alterations within the Pearl River Delta. This posed a new challenge for policy-making aiming to enhance human
Author's personal copy
Q. Zhang et al. / Journal of Hydrology 377 (2009) 274–283
mitigation to water hazards and water resource management within the Pearl River Delta. Thus, reasonable adjustments of human activities were in urgent need to satisfy
natural evolution of environment, and the final objective is
to satisfy the sustainable socioeconomic development of
the Pearl River Delta.
Acknowledgements
The work described in this paper was supported by the ‘‘985
Project” (Grant No.: 37000-3171315) and Research Grants Council
of the Hong Kong Special Administrative Region, China (Project
No.: CUHK405308), the National Natural Science Foundation of
China (Grant No.: 40701015), and by the 111 Project under Grant
B08048, Ministry of Education and State Administration of Foreign
Experts Affairs, PR China. Cordial thanks should be extended to the
editor, Prof. Dr. Geoff Syme, and two anonymous reviewers for
their invaluable comments and suggestions which greatly helped
to improve the quality of this paper.
References
Bott, J.S., Pye, K., van der Wal, D., Neal, A., 2006. Long-term morphological change
and its causes in the Mersey Estuary, NW England. Geomorphology 81, 185–
206.
Chen, X.H., Chen, Y.D., 2002. Hydrological change and its causes in the river network
of the Pearl River Delta. Acta Geographica Sinica 57 (4), 430–436 (in Chinese).
Chen, X.H., Zhang, L., Shi, Z., 2004. Study on spatial variability of water levels in river
net of Pearl River Delta. Journal of Hydraulic Engineering 10, 36–42 (in Chinese).
Chen, Y.D., Zhang, Q., Xu, C.-Y., Yang, Tao., 2008. Change-point alterations of
extreme water levels and underlying causes in Pearl River Delta, China. River
Research and Application. doi:10.1002/rra.1212.
Chen, Y.D., Yang, T., Xu, C.-Y., Zhang, Q., Chen, X., Hao, Z.C., 2008. Hydrologic
alteration along the middle and upper East River (Dongjiang) basin, South
China: a visually enhanced mining on the results of RVA method. Stochastic
Environmental Research and Risk Assessment. doi:10.1007/s00477-008-02947.
Easterling, D.R., Peterson, T.C., 1995. A new method for detecting undocumented
discontinuities in climatological time series. International Journal of
Climatology 15, 369–377.
Ericson, P.J., Vörösmarty, J.C., Dingman, L.S., Ward, G.L., Meybeck, M., 2006. Effective
sea-level rise and deltas: causes of change and human dimension implications.
Global and Planetary Change 50, 63–82.
Gemmer, M., Becker, S., Jiang, T., 2004. Observed monthly precipitation trends in
China 1951–2002. Theoretical and Applied Climatology 77, 39–45.
Huang, Z.G., Zhang, W.Q., Wu, H.S., Fan, J.C., Jiang, P.L., Chen, T.G., Li, Z.H., Huang,
B.S., 2000. Prediction of the increasing magnitude of the sea level in the Pearl
River Delta in 2030 and possible mitigation measures. Science in China (Series
D) 30 (2), 202–208 (in Chinese).
283
Huang, Z.G., Zhang, W.Q., 2004. Impacts of artificial factors on the evolution of
geomorphology during recent thirty years in the Zhujiang Delta. Quaternary
Science 24 (4), 394–401 (in Chinese).
Huang, Z.G., Zhang, W.Q., 2006. The recent change of distributive ratios of runoff
and stream load and its impacts and countermeasures in Zhujiang River Delta.
Yunnan Geographic Environment Research 18 (2), 21–27 (in Chinese).
Jiang, J.M., Gu, X.Q., Ju, J.H., 2007. Significant changes in subseries means and
variances in an 8000-year precipitation reconstruction from tree rings in the
southwestern USA. Annales Geophysicae 25, 1–12.
Liu, Y.H., Chen, X.H., Chen, Y.Q., Zeng, C.H., 2003a. Correlation analysis on abnormal
change of flood level in the central area of the Pearl River Delta. Tropical
Geography 23 (3), 204–208 (in Chinese).
Liu, Y.H., Chen, X.H., Chen, Y.Q., Zeng, C.H., 2003b. Correlation analysis on abnormal
change of flood level in the central area of the Pearl River Delta. Tropical
Geography 23 (3), 204–208 (in Chinese).
Lu, X.X., Zhang, S.R., Xie, S.P., Ma, P.K., 2007. Rapid channel incision of the lower
Pearl River (China) since the 1990s. Hydrology and Earth System Sciences 4,
2205–2227.
Lund, R., Reeves, J., 2002. Detection of undocumented changepoints: a revision of
the two-phase regression model. Journal of Climate 15, 2547–2554.
Luo, X.L., Zeng, E.Y., Ji, R.Y., Wang, C.P., 2007. Effects of in-channel sand excavation
on the hydrology of the Pearl River Delta, China. Journal of Hydrology 343, 230–
239.
Luo, Z.R., Yang, S.Q., Luo, X.L., Yang, G.R., 2000. Dredging at Pearl River mouth and its
dynamical and geomorphologic effects. Tropical Geomorphology 21(1, 2), 15–
20 (in Chinese).
Pont, D., Day, J.W., Hensel, P., Franquet, E., Torre, F., Rioual, P., Ibanez, C., Coulet, E.,
2002. Response scenarios for the deltaic plain of the Rhone in the face of an
accelerated rate of sea-level rise with special attention to Salicornia-type
environments. Estuaries 25, 337–358.
Solow, A.R., 1987. Testing for climate change: an application of the two-phase
regression model. Journal of Climate Applied Meteorology 26, 1401–1405.
Storch, H.V., Zwiers, F., 1999. Statistical Analysis in Climate Research. Cambridge
University Press, Cambridge, p. 116.
Vincent, L.A., 1998. A technique for the identification of inhomogeneities in
Canadian temperature series. Journal of Climate 11, 1094–1104.
Yang, Q.S., Shen, H.T., Luo, X.L., Luo, Z.R., Yang, G.R., Ou, S.Y., 2002. The secular trend
of water level changes in the network channels of the Zhujiang River (Pearl
River) Delta. Acta Oceanologica Sinica 24 (2), 30–37 (in Chinese).
Zeng, Z.X., Liu, N.W., Qiu, S.J., Wu, N., Huang, S., 1992. The tendency of sea level
variation in Pearl River Estuary. Tropic Oceanology 11 (4), 56–62 (in Chinese).
Zhang, Q., Xu, C.-Y., Gemmer, M., Chen, YD., Liu, C.-L., 2008a. Changing properties of
precipitation concentration in the Pearl River basin, China. Stochastic
Environmental Research and Risk Assessment. doi:10.1007/s00477-008-02257.
Zhang, Q., Xu, C.-Y., Gemmer, M., Chen, YD., Zhang, Z.X., Liu, C.-L., 2008b. Spatial and
temporal variability of precipitation over China. Theoretical and Applied
Climatology 1951–2005. doi:10.1007/s00704-007-0375-4.
Zhang, Q., Xu, C.-Y., Yu, Z.G., Liu, C.-L., Chen, Y.D., 2009. Multifractal analysis of
streamflow records of the East River basin (Pearl River), China. Physica A:
Statistical Mechanics and its Applications 388, 927–934.
Zhang, Q., Xu, C.-Y., Chen, Y.D., Yang, T., 2009b. Spatial assessment of hydrologic
alteration across the Pearl River Delta, China, and possible underlying causes.
Hydrological Processes. doi:10.1002/hyp.7268.
Zhou, Z.F., Luo, X.L., Luo, Z.R., Yang, Q.S., 2001. The analysis on the major reason
about local exceptional rise of flood water level in the Pearl River Delta in recent
years. Tropical Geography 21 (4), 319–322 (In Chinese).