Theor. Appl. Climatol. (2009) 95: 53–68 DOI 10.1007/s00704-007-0375-4 Printed in The Netherlands

Theor. Appl. Climatol. (2009) 95: 53–68 DOI 10.1007/s00704-007-0375-4 Printed in The Netherlands
Theor. Appl. Climatol. (2009) 95: 53–68
DOI 10.1007/s00704-007-0375-4
Printed in The Netherlands
1
Department of Geography and Resources Management, Institute of Space and Earth Information Science,
The Chinese University of Hong Kong, Hong Kong, China
2
Chinese Academy of Sciences, Nanjing Institute of Geography and Limnology, Nanjing, China
3
Laboratory for Climate Studies, National Climate Center, China Meteorological Administration, Beijing, China
4
Department of Geosciences, University of Oslo, Oslo, Norway
5
Jiangsu Key Laboratory of Forestry Ecological Engineering, Nanjing Forestry University, Nanjing, China
Spatial and temporal variability of precipitation
over China, 1951–2005
Q. Zhang1;2;3 , C.-Y. Xu4 , Z. Zhang5 , Y. D. Chen1 , C.-L. Liu1
With 10 Figures
Received 26 July 2007; Accepted 8 December 2007; Published online 11 April 2008
# Springer-Verlag 2008
Summary
Annual, winter and summer precipitation records for the
period 1951–2005 from 160 stations in China were analysed using the rotated empirical orthogonal function
(REOF), the Mann–Kendall trend test and the Continuous
Wavelet Transform (CTW) method. The REOF method
was used to analyse the annual and seasonal variability of
precipitation patterns over China, the Mann–Kendall
method was used to detect the temporal trend of the rotated
principal components time series, and the continuous wavelet method was used to explore the periodicity of precipitation changes. In general, six coherent regions across
China are identified using the REOF method: north-east
China; the middle and lower Yangtze River basin; the Haihe
River and the Liaohe River; north-west China; the middle
Yellow River and the South-east Rivers (rivers in south-east
China). Continuous wavelet transform results indicate that
the significant 2–4 year and 6–9 year bands are the major
period components. Precipitation in China is uneven in
space and time, and its complex temporal structure and
spatial variations are different in each season. The Mann–
Kendall test results show that, in general, the middle and
the lower sections of the Yangtze River are dominated
by increasing annual, summer and winter precipitation.
Correspondence: Dr. Qiang Zhang, Department of Geography
and Resource Management, The Chinese University of Hong Kong,
Shatin, NT, Hong Kong, China, e-mail: [email protected]
Increasing annual precipitation can be observed in northwest China. Increasing summer precipitation is found in
north-east China and the Pearl River basin, and the South-east
Rivers are dominated by increasing winter precipitation. The
availability of water as a resource is in close association with
precipitation changes; therefore, this research will be helpful
to watershed-based water resource managers in China.
1. Introduction
Climate change is being influenced heavily by
the effects of greenhouse gases, which will alter
the regional hydrological cycle and subsequently
precipitation changes. The evidence for humaninduced global warming is now ‘‘unequivocal’’,
(IPCC, 2007: http:==en.wikipedia.org=wiki=
IPCC_Fourth_Assessment_Report). The tremendous importance of water in both society and
nature underscores the necessity of understanding how changes in climate could affect regional
water supplies (Xu and Singh 2004). Labat et al.
(2004) indicated that present global warming has
led to changes in the global hydrological cycle
and to the increased magnitude of global and
continental runoff. This global trend should be
quantified at the regional scale where both in-
54
Q. Zhang et al.
creasing and decreasing trends are identified.
Since the change in water resources is different
from region to region under the changing climate
(e.g., Zhang et al. 2006a), climatic changes and
possible impacts on global=regional water resources are receiving increasing focus from the
scientific community (e.g., Loukas et al. 2002;
Camilloni and Barros 2003; Labat et al. 2004).
Booij (2005) studied the impact of climate change
on floods in the River Meuse (western Europe)
using spatially and temporally changed climate
patterns and a hydrological model with three
different spatial resolutions, to investigate the
variability and uncertainty of impacts of climate
changes on river floods obtained from simulations of the hydrological model. Many studies
have shown that the impacts of climatic changes
on global=regional water resources hinge on the
influences of climatic changes on the spatial and
temporal distribution of precipitation (e.g., Gao
et al. 2007). Global warming will alter regional
hydrological cycles, and these alterations are different from region to region, which makes further
investigation of the spatial and temporal variability of precipitation under the changing climate critical (e.g., Domroes et al. 1998; Lana et al. 2001).
The spatial distribution and seasonal variation
of precipitation in China have been discussed
widely (e.g., Qian et al. 2002; Gemmer et al.
2004). Liu et al. (2005) examined the spatial
and temporal variation in daily precipitation from
1960 to 2000 observed at 272 weather stations.
Their results indicated that precipitation in China
had increased by 2% over the study period.
Seasonally, they found that precipitation had increased in winter and summer but decreased in
spring and autumn for the same period. Chen
et al. (1991) studied the climate variation in
China during 1951–1989, indicating that most
of China was dominated by decreased precipitation, especially in northern and north-western
China. Zhai et al. (1999a) reported no significant
trend in annual precipitation over China between
1951 and 1995. Zhai et al. (1999b) investigated
changing trends in annual precipitation and annual extreme precipitation in China during 1951–
1995, and found no significant trends in annual
precipitation, and 1-day and 3-day maximum
precipitation. Wang et al. (2004) showed an increasing trend in precipitation variation during
the second half of the 20th century in western
China, while a similar trend was not found in
eastern China, where the 20- to 40-year periodicities were predominant in the precipitation variations. Ren et al. (2000) studied the trend in
spatial patterns of rainfall in China during 1951–
1996, showing an increasing trend in summer
precipitation over the mid-lower reaches of the
Yangtze River and a decreasing trend over the
Yellow River basin, but almost no change in
the high latitude areas. Zhai et al. (2005), investigated the trends in annual and seasonal total
precipitation and in extreme daily precipitation
for the year, summer, and winter half years using
a daily precipitation dataset of 740 stations across
China for the period 1951–2000. Their study
showed that precipitation changes display different properties in terms of specific regions and
seasons in China. In general, precipitation in
summer and winter increased, while in spring
and autumn it decreased. Precipitation in west and
north-west China decreased before the 1980s and
increased after the 1980s. However, some conclusions are in disagreement. The different conclusions, or even contradictions, reported in the
aforementioned studies are caused by the differences in the time period of the data used in the
analyses, since the temporal trends depend on the
time period of the data series and in which part of
the time series the outliers (most wet years) appear.
Uneven seasonal and spatial distributions of
precipitation usually result in the uneven distribution of flood and drought occurrence. For example, in the summer of 1997, south China was
flooded with excessive rainfall, while one of the
most severe droughts on record occurred in north
China (e.g., Huang et al. 2000). Furthermore,
water resource management usually refers to river
basins. Therefore, an investigation of the spatial
and temporal variability of precipitation based on
long precipitation series in China at the river
basin scale will be useful for the management
of fluvial systems and water resources in China.
Much attention (e.g., Zhai et al. 2005) has
been focussed on the variability of summerand winter-time precipitation due to the fact that
more than 40% of the total annual precipitation
falls in summer, and the winter usually has the
least precipitation of all seasons. To partially redress this inbalance, this study has chosen to focus an annual precipitation as well as on summer
and winter precipitation.
As mentioned above, previous studies are limited in their research scope and data used. The
Spatial and temporal variability of precipitation over China, 1951–2005
spatial and temporal variability of precipitation
in China needs to be re-examined with more robust methods for thorough research on trends,
periodicity and spatial patterns. The objectives
of the current research are: 1) to detect spatial
patterns of the annual, summer and winter precipitation identified using the Rotated Empirical
Orthogonal Function (REOF) method; 2) to explore the trends of major precipitation modes
identified by REOF using the Mann–Kendall
trend test and 3) to detect the periodicity features
of precipitation over China using the Morlet
Wavelet Transform method. This research will
be helpful in furthering the understanding of
the trends and periodicity of precipitation over
China, and also in watershed-based water resource management.
55
quality, continuity, homogeneity and the length
of data records are considered to be more important than the number of stations. (2) Quality
control and homogeneity testing of the meteorological stations were performed by calculating
the von Neumann ratio (N), the cumulative deviations (Q=n0.5 and R=n0.5), and the Bayesian
procedures (U and A) (Buishand 1982; Gemmer
et al. 2004). The quality control and homogeneity tests show that data from the 160 stations used
in the current study are homogeneous at the
>95% confidence level and are of good quality
(Gemmer et al. 2004). The data are from the
National Climatic Centre (NCCC) of the China
Meteorological Administration (CMA). The locations of the stations are shown in Fig. 1.
2.2 Methods
2. Data and methods
2.1 Data
This study uses 160 rain gauge stations which
have good quality, continuous data records for
the period 1951–2005 (Gemmer et al. 2004).
The 160 stations have been chosen for two reasons: (1) the main objective of the study is to
examine the changing spatial and temporal patterns of precipitation in China rather than to provide a high resolution precipitation dataset. The
Fig. 1. Meteorological stations used in this study and the
10 drainage basins (after Gao et al. 2007, with minor revision). The solid dots denote the rain gauging stations. Numbers denote the 10 drainage basins: 1: SongHuajiang River;
2: Liaohe River; 3: Haihe River; 4: Yellow River; 5: Huaihe
River; 6: Yangtze River; 7: SE Rivers (rivers in the southeast
China); 8: Pearl River; 9: SW Rivers (rivers in the southwest
China); 10: NW Rivers (rivers in the northwest China)
Principal Component Analysis is first applied to
extract the general behaviour of all series to be
analysed (Esteban-Parra et al. 1998). The empirical orthogonal function (EOF) analysis tends to
identify physically and dynamically independent
patterns (or normal modes) (Montroy 1997),
which provide important clues as to the physics
and dynamics of the system to be studied (e.g.,
Kim and Wu 1999). However, the physical interpretability of the obtained patterns is a matter
of controversy because of the strong constraints
satisfied by EOFs, namely orthogonality in both
space and time. Physical modes such as normal
modes (Simmons et al. 1983) are not, in general,
orthogonal. This limitation has led to the development of the rotated empirical orthogonal
function (REOF) (Richman 1986). Furthermore,
REOF yields localised structures by compromising some of the EOF properties such as orthogonality. In a comparison study by Kim and Wu
(1999), REOF was found to be good at dividing
climatic patterns. In this paper, the varimax rotated empirical orthogonal function (REOF) method is used, meaning that the initial EOF modes
are linearly transformed using the varimax method, which maximises the variance of the squared
correlation coefficients between the time series
of each REOF mode and each original EOF
mode. The method increases the spatial variability of the obtained modes (Wang et al. 2006).
The trend in the rotated principal components
(PC) time series has been analysed using the
Mann–Kendall test. The non-parametric rank-
56
Q. Zhang et al.
based Mann–Kendall method (MK) (Mann 1945;
Kendall 1975) is commonly used to assess the
significance of monotonic trends in hydro-meteorological time series (e.g., Helsel and Hirsch
1992; Zhang et al. 2006a, b). This test has the
advantage of not assuming any distribution form
for the data and is as powerful as its parametric
competitors (Serrano et al. 1999). Therefore, it
is recommended highly for general use by the
World Meteorological Organization (Mitchell
et al. 1966). In this study, the procedure of the
Mann–Kendall test follows Gerstengarbe and
Fig. 2. Annual rotated REOF rainfall patterns for the period 1951–2005 for A: 1st REOF; B: 2nd REOF; C: 3rd REOF; D: 4th
REOF; E: 5th REOF; F: 6th REOF
Spatial and temporal variability of precipitation over China, 1951–2005
Werner (1999) who used the method to test an
assumption regarding the beginning of the development of a trend within a sample (x1, x2, . . . , xn)
of the random variable X, based on the rank series r of the progressive and retrograde rows of
this sample. The assumption (null hypothesis) is
formulated as follows: the sample under investigation shows no start of a developing trend. In
order to prove or to disprove the assumption, the
following test is done. First a MK test statistic, dk
is calculated:
k
X
dk ¼
ri ð2 ¼ k ¼ nÞ
ð1Þ
i¼1
and
ri ¼
þ1 if xi >xj
0 otherwise
ðj ¼ 1; 2; :::; iÞ
ð2Þ
Under the null hypothesis of no trend, the statistic dk follows a normal distribution with the
expected value of E(dk) and the variance var(dk)
as follows:
E½dk ¼
nðn 1Þ
4
ð3Þ
nðn 1Þð2n þ 5Þ
ð4Þ
72
Under the above assumption, the definition of the
statistic index Zk is calculated as:
Var½dk ¼
dk E½dk Zk ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi ðk ¼ 1; 2; 3; . . . ; nÞ
var½dk ð5Þ
Zk follows the standard normal distribution. In
a two-sided test for trend, the null hypothesis is
rejected at the significance level of if jZj>
Zð1=2Þ , where Zð1=2Þ is the critical value of
the standard normal distribution with a probability exceeding =2. A positive Z value denotes a
positive trend and a negative Z value denotes a
negative trend. In this paper, the significance level of ¼ 5% is used. In contrast to the traditional
MK test which calculates the above statistic variables only once for the whole sample, the corresponding rank series for the so-called retrograde
rows are similarly obtained for the retrograde
sample ðxn ; xn1 ; . . . ; x1 Þ. Following the same
procedure as shown in Eqs. (1) to (5), the statistic
variables, dk, E(dk), var(dk) and Zk will be calculated for the retrograde sample. The Z values
calculated with progressive and retrograde series
57
are named Z1 and Z2, respectively, in this paper.
The intersection point of the two lines, Z1 and
Z2 ðk ¼ 1; 2; . . . ; nÞ gives the point in time of
the beginning of a developing trend within the
time series. The null hypothesis (the sample is
not affected by a trend) must be rejected if the
intersection point is significant at the 5% level
(i.e., outside the 95% confidence interval).
The influence of serial correlation in the time
series on the results of the MK test has been
discussed in the literature (e.g., von Storch 1995;
Yue et al. 2002). In this study, before the MK test
was applied, the meteorological series were tested for persistence using the serial correlation
method (Haan 2002). Pre-whitening has been
used to eliminate the influence of serial correlation (if significant) on the MK test (Yue and
Wang 2004).
The current study uses the continuous wavelet
transform (CWT) (Torrence and Compo 1998)
method to study the periodicity of the PC series.
We applied the Morlet wavelet in the current
study because Morlet wavelet provides a good
balance between time and frequency localisation.
The wavelet is not completely localised in time.
To ignore the edge effects, the Cone of Influence
(COI) was introduced. Here, COI is the region of
the wavelet spectrum in which edge effects become important and is defined here as the e-folding time for the autocorrelation of wavelet power
at each scale. This e-folding time is chosen so
that the wavelet power for a discontinuity at the
edge drops by a factor e2 and ensures that the
edge effects are negligible beyond this point
(Torrence and Compo 1998; Grinsted et al. 2004).
The statistical significance of wavelet power can
be assessed under the null hypothesis that the
signal is generated by a stationary process, given
the background power spectrum. A confidence
Table 1. Percentages of explained variance for each rotated
EOF for the annual precipitation data
REOFs
Eigenvalue
Explained
variance
Cumulated
explained
variance
1
2
3
4
5
6
16.8
14.1
12.7
9.8
8.6
7.6
11
9
8
6
5
5
11
2
28
34
39
44
58
Q. Zhang et al.
level of 95% was taken as the threshold at which
to classify the significance of the wavelet power.
Detailed information of the continuous wavelet
transform used in the current study was thoroughly introduced in Torrence and Compo (1998).
3. Results
3.1 Annual precipitation
Figure 2 demonstrates the annual precipitation
patterns of the (varimax) rotated EOFs for China
Fig. 3. Mann–Kendall trend of annual rotated PC series, rotated PC series and 5-year moving average (thick solid). Rotated
PC series of A: 1st REOF; B: 2nd REOF; C: 3rd REOF; D: 4th REOF; E: 5th REOF; F: 6th REOF. The Z1 and Z2 are the MK
statistic index. Positive Z value denotes increasing trend and vice versa. Detailed information about the Z1 and Z2 can be
referred to Zhang et al. (2006a)
Spatial and temporal variability of precipitation over China, 1951–2005
59
Fig. 4. Continuous wavelet transform of annual Rotated PC series of A: 1st REOF; B: 2nd REOF; C: 3rd REOF; D: 4th REOF;
E: 5th REOF; F: 6th REOF. The thick black contour designates the 95% confidence level against red noise and the cone of
influence (COI) where edge effects might distort the picture is shown as a lighter shade
by drawing the isolines of the loading factor
values. Table 1 lists the percentage of variance
explained by each REOF. The first REOF pattern
(Fig. 2A) is centred mainly on the middle-north
Yangtze River basin and middle-south Huaihe
River basin. The MK trend test of the principal
component (PC) time series of this region
(Fig. 3A) indicates that precipitation increased
after about 1962. However, no significant changing trend is detected at the >95% confidence level. The 5-year moving average identified no
obvious dry or wet periods. The wavelet power
spectra for the PCs are shown in Fig. 4A. The
95% confidence regions demonstrate that 1960–
1970 and 1975–1990 include intervals of higher
precipitation variance. Four to eight year periods
are identified in precipitation series for 1960–
1970 and a 8-year period for 1975–1990.
The second REOF pattern (Fig. 2B) dominates
the south-east Yangtze River basin and the
South-east Rivers (rivers in the southeast China).
The precipitation of this region (Fig. 3B) decreased during 1951–1970 and increased thereafter. Wet periods are identified during 1951–1956,
1970–1978 and 1993–2005; dry periods, however, are detected during 1956–1970 and 1979–
1993. Different periodicity features are shown in
Fig. 4B compared to those of PCs of the first
REOFs (Fig. 4A). Figure 4B indicates that the
higher precipitation variance in the south-east
Yangtze River basin and the South-east Rivers
occurred around 1970 and 1980. These two periods have two significant year bands: 2–4 year
band and 6–9 year band.
The middle Yellow River is associated with
the 3rd REOF pattern (Fig. 2C), and the Liaohe
River and the Haihe River are associated with the
4th REOF pattern (Fig. 2D). The MK trend indicates that the middle Yellow River basin is
characterised by increasing precipitation during
1951–1969 and by decreasing precipitation during 1970–2000 (Fig. 3C). A 5-year moving averTable 2. Percentages of explained variance for each rotated
EOF for the summer precipitation data
REOFs
Eigenvalue
Explained
variance
Cumulated
explained
variance
1
2
3
4
5
6
15.9
14.4
11.7
9.6
7.2
6.3
11
9
7
6
5
4
11
2
27
33
38
42
60
Q. Zhang et al.
age also indicates a dry period during 1970–2000.
The Liaohe River and the Haihe River basins are
characterised by slightly decreasing precipitation
(Fig. 3D) for the same period. The annual precip-
itation trend in the middle Yellow, Liaohe and
Haihe Rivers is not significant at the >95% confidence level. These two regions have similar
periodicity features, that is, the precipitation
Fig. 5. Summer REOF rainfall patterns for the period 1951–2005 for A: 1st REOF; B: 2nd REOF; C: 3rd REOF; D: 4th REOF;
E: 5th REOF; F: 6th REOF
Spatial and temporal variability of precipitation over China, 1951–2005
changes are dominated by the shorter periods of
the 2–4 year band, which is significant at the
>95% confidence level. The higher precipitation
variance occurred over the middle Yellow River
in the 2–4 year band during 1960–1970 and for
the Liaohe and Haihe Rivers in the 2–5 year band
61
during 1953, 1968–1975 and 1990–1993, respectively (see Fig. 4C, D).
Figure 2E illustrates that the 5th REOF is
centred over the upper Yangtze River and the
North-west Rivers (rivers in north-west China).
North-east China is dominated by the 6th REOF
Fig. 6. Mann–Kendall trend of summer rotated PC series, rotated PC series and 5-year moving average (thick solid). Rotated
PC series of A: 1st REOF; B: 2nd REOF; C: 3rd REOF; D: 4th REOF; E: 5th REOF; F: 6th REOF
62
Q. Zhang et al.
(Fig. 2F). The loading factor values of these two
REOF patterns are negative. The annual precipitation in north-west China has been decreasing
since 1965 (Fig. 3E). The annual precipitation
in north-east China is more variable with two
periods dominated by increasing precipitation
(1951–1960 and 1976–1985) and decreasing precipitation (1960–1975 and 1986–2005) (Fig. 3F).
Wet periods are identified during 1955–1970 and
1981–1994; and dry periods are observed during
1970–1980 and 1995–2005. Higher precipitation
variance occurred over north-west China during
1985–2000 in the 2–4 year band (Fig. 4E), and to
north-east China during 1960 and 1980–1985
in the 2–4 and 10–15 year band, respectively.
3.2 Summer precipitation
Figure 5 shows the summer precipitation patterns
of the (varimax) REOFs for China using the isolines of the loading factor values. Table 2 lists the
percentage of variance explained by each REOF.
The first and second REOFs (Fig. 5A, B) are
associated alternatively with the precipitation
regimes of the middle and lower Yangtze River
and the Huaihe River. The summer precipitation
in the middle and lower Yangtze River decreased
during 1951–1968, but increased during 1968–
2000. This increasing trend is significant at the
>95% confidence level (Fig. 6A). The 5-year
moving average also indicates a dry period during 1955–1980 and a wet period during 1980–
2002. Huaihe River precipitation has decreased
slightly, however, a relatively wet period has
been identified with the 5-year moving average,
i.e., 1969–1975 (Fig. 6B). The periodicity features of the precipitation series of the middle
and lower Yangtze River and the Huaihe River
are illustrated in Fig. 7A, B. Figure 7A indicates
that the higher precipitation variance occurred
during 1970 and 1980 in the 2–4 year band.
The higher precipitation variance in the Huaihe
River, however, occurred during 1958–1970 in
the 5–7 year band (Fig. 7B).
The third REOF pattern is centred over the
middle Yellow River and eastern part of the
North-west Rivers (Fig. 5C). The fourth REOF
pattern is associated with the precipitation regimes in the Liaohe and Haihe Rivers (Fig. 5D).
The MK trend indicates no obvious increasing
or decreasing precipitation changes in the middle
Yellow River and eastern part of the North-west
Rivers (Fig. 6C). Even so, two wet periods and
two dry periods are also identified in Fig. 6C:
Fig. 7. Continuous wavelet transform of summer rotated PC series of A: 1st REOF; B: 2nd REOF; C: 3rd REOF; D: 4th REOF;
E: 5th REOF; F: 6th REOF. The thick black contour designates the 95% confidence level against red noise and the cone of
influence (COI) where edge effects might distort the picture is shown as a lighter shade
Spatial and temporal variability of precipitation over China, 1951–2005
1951–1960 and 1975–1985 are relatively wet
periods; 1960–1975 and 1986–2005 are relatively dry periods. A 5-year moving average also
indicates decreasing summer precipitation during
63
1980–2005. It can be seen from Fig. 6D that the
Liaohe and the Haihe Rivers are characterised by
increasing precipitation during 1952–1964 and
by decreasing precipitation during 1964–1983
Fig. 8. Winter rotated EOF rainfall patterns for the period 1951–2005 for A: 1st REOF; B: 2nd REOF; C: 3rd REOF; D: 4th
REOF; E: 5th REOF; F: 6th REOF
64
Q. Zhang et al.
and 1995–2005. It can also be observed from
Fig. 6D that 1980–1990 is characterised by dryness, and 1960–1975 by wetness. The periodicity
features of the precipitation series of the regions
are shown in Fig. 7C, D. The higher level of
precipitation variance occurred to the 2–4 year
band during 1955–1965 and the 8–10 year band
during 1965–1975 for the third rotated pattern
(Fig. 7C). The higher precipitation variance occurred to the 2–4 year band during 1965–1970
(Fig. 7D).
The fifth and sixth REOFs (Fig. 5E, F) are
associated with the precipitation regimes in the
Huaihe River, the north-middle Yangtze River
and north-east China. MK trend test results indicate that the Huaihe River is characterised
by decreasing precipitation during 1951–1978.
Thereafter, increasing precipitation is detected
in 1978 and 1980 (Fig. 6E). The summer precipitation in north-east China decreased during
1955–1980 and during 1985–2005, and increased during 1980–1985. Dry periods are identified during 1970–1980, and wet periods are
identified during 1956–1965 and 1980–1998
(Fig. 6F). It can be seen from Fig. 7E that no
significant periods are detected with the help of
continuous wavelet transform method, which is
different from the periodicity features of other
PCs. As for the periodicity of precipitation in
north-east China, higher precipitation variance
occurred in the period 1990–2000 in the 3–5
year band.
3.3 Winter precipitation
Figure 8 demonstrates the winter precipitation
patterns of the (varimax) REOFs over China by
the isolines of the loading factor values. Table 3
lists the percentage of variance explained by each
Table 3. Percentages of explained variance for each REOF
for the winter precipitation data
REOFs
Eigenvalue
Explained
variance
Cumulated
explained
variance
1
2
3
4
5
6
34.8
20.1
10.7
9.1
7.3
5.6
22
13
7
6
5
4
22
35
42
48
53
57
REOF. The first REOF (Fig. 8A) is associated
with the winter precipitation regimes of the middle Yangtze River basin, the Huaihe River and
the lower Yellow River, and the second REOF
centres on the Pearl River and in the south
Yangtze River (Fig. 8B). It can be seen from
Fig. 9a that no obvious changing patterns can
be identified in the winter precipitation series
from the middle Yangtze River, the Huaihe River
and the lower Yellow River. A 5-year moving
average shows three wet periods, i.e., 1970–
1975, 1988–1993 and 2000–2002. In the Pearl
River and the south Yangtze River (Fig. 9B), a
5-year moving average shows a wetter winter
period during 1980–2000 and a dry winter period
during 1960–1980. No periodicity features of the
winter precipitation series can be detected in the
middle Yangtze River basin, the Huaihe River
and the lower Yellow River (Fig. 10A). The significant wavelet power can be found in the 2–3
and 6–8 year band around 1980–1985 and 1980–
1990 in the Pearl River and the south Yangtze
River (Fig. 10B).
The third REOF is centred on the Liaohe River
and the Haihe River (Fig. 8C). The fourth REOF
is dominant in the middle and south Yangtze River and the west Pearl River (Fig. 8D). The MK
trend test result indicates that winter precipitation decreased after 1980 and no obvious wet
or dry periods are identified in the Liaohe and
Haihe Rivers (Fig. 9C). Figure 9D indicates that
winter precipitation in the middle and the south
Yangtze River and the west Pearl River decreased during 1960–1980 and increased thereafter until 1990. A relatively wet period during
1963–1979 and a relatively dry period during
1989–1998 have been identified by a 5-year
moving average (Fig. 9D). Figure 10C demonstrates the significant wavelet power of winter
precipitation in the 2–4 year band around
1990, and in the 6–9 year band during 1970–
1990. No significant wavelet power is identified
in winter precipitation in the middle and south
Yangtze River and in the west Pearl River
(Fig. 10D).
The fifth and the sixth REOFs (Fig. 8E, F) are
associated with the winter precipitation regimes
of north-east China, the lower Yangtze River and
in the South-east Rivers. Figure 9E indicates that
winter precipitation in north-east China decreased
during 1955–1975 and increased during 1951–
Spatial and temporal variability of precipitation over China, 1951–2005
65
Fig. 9. Mann–Kendall trend of winter rotated PC series, rotated PC series and 5-year moving average (thick solid). Rotated
PC series of A: 1st REOF; B: 2nd REOF; C: 3rd REOF; D: 4th REOF; E: 5th REOF; F: 6th REOF
1955 and 1975–2005. A 5-year moving average
identifies a dry period during 1960–1990, and
wet periods during 1951–1960 and 1999–2005.
Winter precipitation in the lower Yangtze River
and the South-east Rivers increased during 1988–
2005, no obvious wet=dry periods are identified
with a 5-year moving average (Fig. 9F). Contin-
uous wavelet transform results indicate significant
wavelet power in the 2–4 years band during
1955–1963 in north-east China (Fig. 10E), and
in the 4–8 year band during 1960–1970 and
in the 3–4 year band during 1995–1998 in the
lower Yangtze River and the South-east Rivers
(Fig. 10F).
66
Q. Zhang et al.
Fig. 10. Continuous wavelet transform of winter rotated PC series of A: 1st REOF; B: 2nd REOF; C: 3rd REOF; D: 4th REOF;
E: 5th REOF; F: 6th REOF. The thick black contour designates the 95% confidence level against red noise and the cone of
influence (COI) where edge effects might distort the picture is shown as a lighter shade
4. Conclusion and discussion
As shown in Sect. 3, no individual method can
reveal the different statistical properties of precipitation variability, and each method has its own
strength and weakness. The results of the three
methods complement each other. Long series of
the annual, winter and summer precipitation data
for 1951–2005 from 160 stations in China were
analysed using the rotated empirical orthogonal
function (REOF) method, Mann–Kendall method, and continuous wavelet transform (CWT)
method. The coherent regions of China vary with
season as defined by annual, winter and summer
time series. In general, six coherent regions in
China are identified based on the REOF method:
north-east China, the middle and lower Yangtze
River basin, the Haihe River and the Liaohe River,
the north-west China, the middle Yellow River,
and the Sourth-east Rivers.
Precipitation in China presents a complicated
spatial and temporal structure. It is uneven in
space and time and this complex spatial and temporal structure is different in different seasons. In
reality, several factors such as sea surface temperatures and Tibetan Plateau heating (Zhang et al.
2007) exert a tremendous influence on spatial
and temporal precipitation changes over China.
All these factors combine to result in the complex spatial and temporal structure of precipita-
tion changes over China. Research results of the
current study suggest that different change patterns occurred over the different regions of China
for different seasons. For annual precipitation,
north-east China is dominated by decreasing precipitation, especially after 1970. Decreasing precipitation also occurred over the middle and
lower Yellow River and over the Huaihe River.
Annual precipitation in the middle and lower
Yangtze River and in the South-east Rivers has
been increasing. North-west China is dominated
by significant increasing precipitation after 1970.
The start time of this increase is similar, namely,
1970. The precipitation in the Haihe and Liaohe
River exhibits no obvious changing patterns.
Summer precipitation in north-east China has
been decreasing since after 1980. The middle
Yellow River and Huaihe River are dominated
by decreasing precipitation. The South-east Rivers
and north-west China are characterised by increasing precipitation after 1980. Summer precipitation in the middle and lower Yangtze River
increased significantly during 1968–2005 and increased slightly over the Haihe and Liaohe Rivers.
Almost all these changes have a similar start time
of 1980–1985. In north-east China, annual precipitation has decreased since 1986; however,
summer precipitation has increased since 1980.
The interesting results are that winter precipitation
Spatial and temporal variability of precipitation over China, 1951–2005
has increased over the middle and lower Yellow
River and Huaihe River, and that these places are
characterised by decreased summer precipitation.
North-west China, the Haihe and Liaohe Rivers,
the South-east Rivers and the Pearl River are
dominated by increasing winter precipitation after about 1980, 1990, respectively. Increasing
summer precipitation is found in north-east China.
The Pearl River basin and the South-east Rivers
are dominated by increasing winter precipitation.
The results of this research demonstrate that
winter precipitation has increased and that summer and annual precipitation has decreased in
some places in China, e.g., the middle and lower
Yellow River and north-east China, respectively.
Increasing precipitation is identified in northwest China. Zhai et al. (2005) and Gong and
Ho (2002) also indicated significant positive
trends in winter precipitation over Tibet, and significant increasing summer precipitation in the
lower Yangtze River. Continuous wavelet transform results indicate that the periods of all the
PCs share the similarly significant 2–4 year and
8 year periods. The climatic changes in China
are controlled mainly by the winter and summer
monsoon (Domroes and Peng 1988). In general,
precipitation in south-west China is greater than
in north-west China, and these precipitation patterns are determined mainly by the monsoon system and effects of topography (Zhai et al. 2005).
Rainy seasons in eastern China depend on the
progress and retreat of the East Asian summer
monsoon. Detailed information on the evolution
of the summer Asian monsoon and the associated
propagation of the rain belt can be found in Ding
(1994). In general, the rain belt propagates northward in early May and June and reaches north
China between 5–25 August. After midsummer,
the rain belt retreats southward rapidly, which
directly determines precipitation changes across
China. The MK trend test indicates a crossover of
annual and seasonal precipitation changes during
the 1970s and 1980s. These results suggest that
change in precipitation are attributed to changes
in the East Asian summer monsoon system. Wang
(2001) indicated that the weakening of the Asian
monsoon circulation after the 1970s is not beneficial for the northward propagation of the rain belt.
Some Chinese scholars attributed precipitation
anomalies to sea surface temperature anomalies
in the equatorial eastern Pacific, the sub-tropical
high across the north-west Pacific, the monsoon
67
system and snow cover over Tibet in winter (e.g.,
Gong and Ho 2002; Zhao and Xu 2002). These
studies indicate that various, complex influencing
factors are responsible for changes in precipitation
in China. The availability of water is closely associated with precipitation changes; therefore, this
research will be helpful for watershed-based water
resource management in China.
Acknowledgement
This research was supported financially by the Laboratory for
Climate Studies, National Climate Center, China Meteorological Administration, China (Grant No: CCSF2007-35), the
National Natural Science Foundation of China (Grant No.:
40701015), the Outstanding Oversea Chinese Scholars Fund
from the Chinese Academy of Sciences and by the Direct
Grant from the Faculty of Social Science, The Chinese
University of Hong Kong (Project No. 4450183). Wavelet software was provided by C. Torrence and G. Compo, and is available from: http:==paos.colorado.edu=research=wavelets=.
Cordial thanks are extended to the two anonymous reviewers
and the managing editor of theorectical and Applied Climatology, Prof. Dr. Hartmut Grassl, for their invaluable comments which improved greatly the quality of this paper.
References
Booij MJ (2005) Impact of climate change on river flooding
assessed with different spatial model resolutions. J Hydrol
303: 176–198
Buishand TA (1982) Some methods for testing the homogeneity of rainfall records. J Hydrol 58: 11–27
Camilloni IA, Barros VR (2003) Extreme discharge events
in the Paraná River and their climate forcing. J Hydrol
278: 94–106
Chen LX, Shao YN, Dong M, Ren ZH, Tian GS (1991)
Preliminary analysis of climatic variation during the last
39 years in China. Adv Atmos Sci 8: 279–288
Ding Y (1994) Monsoons over China. Kluwer Academic
Publishers, Dordrecht, 419 pp
Domroes M, Peng G (1988) The climate of China. Springer,
Berlin Heidelberg New York, 361 pp
Domroes M, Kaviani M, Schaefer D (1998) An analysis of
regional and intra-annual precipitation variability over
Iran using multivariate statistical methods. Theor Appl
Climatol 61: 151–159
Esteban-Parra JM, Rodrigo FS, Castro-Diez Y (1998)
Spatial and temporal patterns of precipitation in Spain
for the period 1880–1992. Int J Climatol 18: 1557–1574
Gao G, Chen DL, Xu C-Y, Simelton E (2007) Trend of
estimated actual evapotranspiration over China during
1960–2002. J Geophys Res – Atmosph 112: D11120,
doi: 10.1029=2006JD008010
Gemmer M, Becker S, Jiang T (2004) Observed monthly
precipitation trends in China 1951–2002. Theor Appl
Climatol 77: 39–45
Gerstengarbe FW, Werner PC (1999) Estimation of the
beginning and end of recurrent events within a climate
regime. Climate Res 11: 97–107
68
Q. Zhang et al.: Spatial and temporal variability of precipitation over China, 1951–2005
Gong DY, Ho CH (2002) Shift in the summer rainfall over
the Yangtze River valley in the late 1970s. Geophys Res
Lett 29: 1436, doi: 10.1029=2001GL014523
Grinsted A, Moore JC, Jevrejeva S (2004) Application of the
cross wavelet transform and wavelet coherence to geophysical time series. Nonlinear Proc Geophy 11: 561–566
Haan CT (2002) Statistical methods in hydrology (second
Ed.). Blackwell, Iowa State Press
Helsel DR, Hirsch RM (1992) Statistical methods in water
resources. Studies in Environmental Science. Elsevier,
Amsterdam, 522 pp
Huang R, Zhang R, Zhang Q (2000) The 1997=98 ENSO
cycle and its impact on summer climate anomalies in East
Asia. Adv Atmos Sci 17: 348–362
IPCC (2007) Climate change 2007: The Physical Science
Basis. In: Solomon S, Qin D, Manning M, Chen Z,
Marquis M, Averyt KB, Tanor M, Miller HL (eds)
Working Group I Contribution to the Intergovernmental
Panel on Climate Change Fourth Assessment Report.
Cambridge University Press, Cambridge, UK, p 996
Kendall MG (1975) Rank correlation methods. Griffin,
London, UK
Kim KY, Wu QG (1999) A comparison study of EOF
techniques: analysis of nonstationary data with periodic
statistics. J Climate 12: 185–199
Labat D, Godderis Y, Probst JL, Guyot JL (2004) Evidence
for global runoff increase related to climate warming. Adv
Water Resour 27: 631–642
Lana X, Serra C, Burgue~
no A (2001) Patterns of monthly
rainfall shortage and excess in terms of the standardized
precipitation index for Catalonia (NE Spain). Int J
Climatol 21: 1669–1691
Liu B, Xu M, Henderson M, Qi Y (2005) Observed trends of
precipitation amount, frequency, and intensity in China,
1960–2000. J Geophys Res 110: D08103, doi: 10.1029=
2004JD004864
Loukas A, Vasiliades L, Dalezios NR (2002) Potential climate change impacts on flood producing mechanisms in
southern British Columbia, Canada using the CGCMA1
simulation results. J Hydrol 259: 163–188
Mann HB (1945) Nonparametric tests against trend. Econometrica 13: 245–259
Mitchell JM, Dzerdzeevskii B, Flohn H, Hofmeyr WL,
Lamb HH, Rao KN, Wallen CC (1966) Climate change,
WMO Technical Note No. 79, World Meteorological
Organization, 79 pp
Montroy DL (1997) Linear relation of central and eastern
North American precipitation to tropical Pacific sea surface temperature anomalies. J Climate 10: 541–558
Qian W, Kang HS, Lee DK (2002) Distribution of seasonal
rainfall in the East Asian monsoon region. Theor Appl
Climatol 71: 151–168
Ren GY, Wu H, Chen ZH (2000) Spatial patterns of change
trend in rainfall of China. Quart J Appl Meteor 11(3):
322–330 (in Chinese)
Richman MB (1986) Rotation of principal components.
J Climatol 6: 293–335
Serrano VL, Mateos VL, Garcı́a JA (1999) Trend analysis of
monthly precipitation over the Iberian Peninsula for the
period 1921–1995. Phys Chem Earth (B) 24(2): 85–90
Simmons AJ, Branstator GW, Wallace JM (1983) Barotropic
wave propagation, instability and atmospheric teleconnection patterns. J Atmos Sci 40: 1363–1392
Torrence C, Compo GP (1998) A practical guide to wavelet
analysis. Bull Amer Meteor Soc 79: 61–78
von Storch VH (1995) Misuses of statistical analysis in
climate research. In: Storch HV, Navarra A (eds) Analysis
of climate variability: application of statistical techniques.
Berlin, Springer, pp 11–26
Wang HJ (2001) The weakening of the Asian Monsoon circulation after the end of the 1970s. Adv Atmos Sci 18: 376–386
Wang SW, Zhu JH, Cai JN (2004) Interdecadal variability of
temperature and precipitation in China since 1880. Adv
Atmos Sci 21: 307–313
Wang SY, Mcgrath R, Semmler T, Sweeney C (2006)
Validation of simulated precipitation patterns over Ireland
for the period 1961–2000. Int J Climatol 26: 251–266
Webster PJ, Magana VO, Palmer TN, Shukla J, Tomas RA,
Yanai M, Yasunari T (1998) Monsoons: processes,
predictability, and the prospects for prediction. J Geophys
Res 103: 14451–14510
Xu C-Y, Singh VP (2004) Review on regional water
resources assessment models under stationary and changing climate. Water Resour Manag 18: 591–612
Yang FL, Lau KM (2004) Trend and variability of China
precipitation in spring and summer: linkage to sea-surface
temperatures. Int J Climatol 24: 1625–1644
Yatagai A, Yasunari T (1995) Interannal variations of summer precipitation in the arid=semi-arid regions in China
and Mongolia: Their regionality and relation to the Asian
summer monsoon. J Meteor Soc Jpn 73: 909–923
Yue S, Wang CY (2004) The Mann-Kendall test modified by
effective sample size to detect trend in serially correlated
hydrological series. Water Resour Manag 18: 201–218
Yue S, Pilon P, Phinney B, Cavadias G (2002) The influence
of autocorrelation on the ability to detect trend in hydrological series. Hydrol Process 16: 1807–1829
Zhai PM, Sun A, Ren F, Liu X, Gao B, Zhang Q (1999a)
Changes of climate extremes in China. Clim Change 42:
203–218
Zhai PM, Ren FM, Zhang Q (1999b) Detection of trends in
China’s precipitation extremes. Acta Meteor Sinica 57:
208–216 (in Chinese)
Zhai PM, Zhang XB, Wan H, Pan XH (2005) Trends in total
precipitation and frequency of daily precipitation
extremes over China. J Climate 18: 1096–1108
Zhang Q, Gemmer M, Chen JQ (2007) Flood=drought
changes during past 1000 years in the Yangtze Delta region and possible connections with Tibetan climatic
changes. Global Planet Change 57: 213–221
Zhang Q, Liu CL, Xu C-Y, Xu Y-P, Jiang T (2006a)
Observed trends of annual maximum water level and
streamflow during past 130 years in the Yangtze River
basin, China. J Hydrol 324: 255–265
Zhang Q, Xu C-Y, Becker S, Jiang T (2006b) Sediment and
runoff changes in the Yangtze River basin during past 50
years. J Hydrol 331: 511–523
Zhao ZG, Xu L (2002) Potential impact on El Ni~
no events on
the circulation and climate variation in China. Wea Climate 1: 109–118
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement