Journal of Hydrology (2007) 333, 265– 274 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jhydrol Possible influence of ENSO on annual maximum streamflow of the Yangtze River, China Qiang Zhang a,b,* , Chong-yu Xu a,c , Tong Jiang a, Yijin Wu d a Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences, 73 East Beijing Road, Nanjing 210008, PR China b Geographical Institute, Giessen University, Giessen 35390, Germany c Department of Geosciences, University of Oslo, Norway d School of Urban and Environmental Sciences, Huazhong Normal University, Wuhan 430079, China Received 12 January 2006; received in revised form 17 July 2006; accepted 30 August 2006 KEYWORDS Annual maximum streamflow; El Niño/Southern Oscillation (ENSO); Wavelet approach; Yangtze River basin Summary Variability and possible teleconnections between annual maximum streamflow from the lower, the middle and the upper Yangtze River basin and El Niño/Southern Oscillation (ENSO) are detected by continuous wavelet transform (CWT), cross-wavelet and wavelet coherence methods. The results show that: (1) different phase relations are found between annual maximum streamflow of the Yangtze River and El Niño/Southern Oscillation (ENSO) in the lower, the middle and the upper Yangtze River basin. In-phase relations are detected between annual maximum streamflow of the lower Yangtze River and anti-phase relations are found in the upper Yangtze River. But ambiguous phase relations occur in the middle Yangtze River, showing that the middle Yangtze River basin is a transition zone. Different climatic systems control the upper and the lower Yangtze River. The upper Yangtze River is mainly influenced by the Indian summer monsoon and the lower Yangtze is mainly influenced by the East Asian summer monsoon; (2) as for the individual stations, different phase relations are found in the longer and the shorter periods, respectively. In the longer periods, the annual maximum streamflow is more influenced by climatic variabilities, while in the shorter periods, it is influenced by other factors, e.g. human activities. The results of the study provide valuable information for improving the long-term forecasting of the streamflow using its relationship with ENSO and the Indian Monsoon. ª 2006 Elsevier B.V. All rights reserved. * Corresponding author. Address: Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences, 73 East Beijing Road, Nanjing 210008, PR China. Tel./fax: +86 25 86882125. E-mail address: zhangq@niglas.ac.cn (Q. Zhang). Introduction Flood hazards cause enormous economical, social and environmental damages and loss of lives. Floods usually include 0022-1694/$ - see front matter ª 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.jhydrol.2006.08.010 266 three factors: peak flood discharge, water level and flood duration; extreme flood discharge usually plays the key role in the occurrence of flood hazards and is likely to have a greater potential to impact water resources in many regions than the mean annual discharge does. More frequent or larger floods could lead to increased expenditures for flood management. It is why more and more researchers draw concerns on the study of extreme flood events (Jain and Lall, 2001; Camilloni and Barros, 2003). The intensifying human activities (e.g. urbanization, forestation/deforestation, construction of water reservoir) will exert tremendous influences on flood frequency, and temporal and spatial distributions of water resources. Furthermore, climatic variability combined with human-induced emission of green-house gases result in an increase in mean global temperature (IPCC, 2001), which in turn, leads to higher evaporation rates and makes the atmosphere transport larger amounts of water vapor. The global hydrological cycle is accelerated (Menzel and Bürger, 2002). Influence of the slowly changing climate on flood frequency has attracted interest (e.g. Robson et al., 1998; Jain and Lall, 2000, 2001; Olsen et al., 1999; Zhang et al., 2005). The El Niño/Southern Oscillation (ENSO) represents the dominant coupled ocean–atmosphere mode of the tropical Pacific (Cane, 1992). On inter-annual timescales the significant part of the global climatic changes can be linked to ENSO (Trenberth et al., 1998). The ENSO extreme phases are usually in linkage with major episodes of floods and droughts (e.g. Barlow et al., 2001) in many locations worldwide (Jain and Lall, 2001; Aceituno, 1988; Amarasekera et al., 1997). Many scholars try to detect possible connections between ENSO and precipitation and streamflow. Lan et al. (2002) suggested that ENSO contributed to the runoff in the upper reaches of the Yellow River in China; the occurrence of El Niño is usually accompanied by high probability of low flow, while flood events in the Yellow River usually accompanied by the occurrence of La Niña event. Cardoso and Silva Dias (2006) also investigated the relationship between the Paraná River (27.36S, 55.90W) flow and the ENSO mode, and statistical forecasts of river flow are made using the relationship. An evaluation of the relationship between the Pacific sea surface temperature and the Paraná River flow indicates an ENSO pattern over the equatorial Pacific. Gong and Wang (1999), however, studied the teleconnection between ENSO and precipitation in China with the help of statistical analysis (v2 test), suggesting that the decreasing precipitation in China usually matches the El Niño events and there exists a significant relationship between winter and autumn rainfall and the ENSO in eastern China. Some other scientists have also detected strong correlations between flood events and ENSO events (e.g. Chang and King, 1999; Dilley and Heyman, 1999). Many researches were performed on streamflow changes of the Yangtze River. Zhang et al. (2006) analyzed the changes of trends and periodicity of the annual maximum streamflow and water level at different stations along the Yangtze River during the past 130 years, indicating that the annual maximum streamflow in the upper Yangtze River is in a decreasing trend while the opposite is true in the middle and the lower Yangtze River. Annual maximum stream- Q. Zhang et al. flow in the middle Yangtze River has a significant upward trend, which shows that the flood hazard in the middle Yangtze River is of a serious concern. Jiang et al. (2006) analyzed the teleconnections between flood/drought events in the Yangtze River basin and ENSO events during 1868–2003 with the help of v2 test and spectral analysis, suggesting that ENSO events and flood/drought variations are significantly correlated at a 5.04-year period and a 10- to 12-year period. These researches are greatly helpful for understanding and controlling the floods and droughts problems in the Yangtze River basin. However, what are the possible connections between ENSO and annual maximum streamflow of the Yangtze River, especially in terms of periods? To what degree does the ENSO impact the annual maximum streamflow? These questions are remaining unanswered and are seldom studied, especially with the help of the powerful cross and coherence wavelet analysis methods. The main objectives of the present study are: (1) to explore the changes of variance and in-phase linkages between Niño3 (sea surface temperature) and annual maximum streamflow of the three major monitoring hydrologic stations along the main Yangtze River, i.e. Datong station, Hankou station and Yichang station; and (2) to evaluate the possible impacts of ENSO on flood hazards in the Yangtze River Basin. This study uses the wavelet transform (WT) approach, the cross-wavelet power and coherence wavelet analyses methods to detect the relations between Niño3 SST and annual maximum streamflow of the Yangtze River. Yangtze River: climate and hydrology The Yangtze River (Changjiang), being the longest river in China and the third longest river in the world, lies between 91E and 122E and 25N and 35N. It has a drainage area of 1,808,500 km2 with the mean annual discharge of 23,400 m3 s1 measured at Hankou Station. The river originates in the Qinghai-Tibet Plateau and flows about 6300 km eastwards to the East China Sea (Zhang et al., 2006). The Yangtze River Basin is located in the monsoon region of East Asia subtropical zone, and has a mean annual precipitation of about 1090 mm (Zhang et al., 2005; Jiang et al., 2006). Climatically, the southern part of the basin is close to the tropical zone and the northern part is close to the temperate zone, making it an ideal place for studying the influence of climate changes on hydrological conditions. The mean annual temperature in the southern and northern parts of the middle and the lower Yangtze basin is 19 and 15 C, respectively. Summer is the main flooding season for the Yangtze River basin due to the heavy monsoon rainfall. Temporal and spatial distributions of the rain zone are closely related to monsoon activities and seasonal motion of subtropical highs. Flood or drought events happed nearly every year. The river reach between Yichang and Wuhan (Fig. 1) is the most dangerous river section in the Yangtze River basin concerning the flood events. In 1998, the entire Yangtze River Basin suffered from tremendous flooding – the largest flood since 1954, which led to the economic loss of 166 billion Chinese Yuan (or 20 billion US$) (Yin and Li, 2001). Possible influence of ENSO on annual maximum streamflow of the Yangtze River, China Figure 1 267 Location of the study region and hydrological stations. Data and method Data Annual maximum streamflow data from the three main gauge stations of the Yangtze River are analyzed in this study: Yichang station (controlling 1,005,501 km2), Hankou station (controlling 14,488,036 km2) and Datong station (controlling 1,705,383 km2) representing the upper, the middle and the lower reaches of the river, respectively (CWRC, 2000; Zhang et al., 2006) (Table 1). The streamflow from the upper station Yichang and the large tributary of the middle Yangtze River – Hanjiang River – is passing through Hankou station, which is the key reference station for flood mitigation and flood control in the basin. Datong station is the monitoring station at the lower Yangtze River, receiving the streamflow from Hankou and tributary Poyang water system. It can be seen from Table 1 that there exist long data series of annual maximum streamflow for all three stations and the longest one has a 135-year record from 1865 to 2000. Table 1 Detailed information on the extreme hydrological records of Yichang, Hankou and Datong gauging stations (revised after Zhang et al., 2006) Station name Max. runoff (m3/s) Occurrence time of max. runoff Time series of data Yichang station Hankou station Datong station 71,100 76,100 92,600 1896.09.04 1954.08.14 1954.08.01 1877–2000 1865–2000 1922–2000 Niño3 sea surface temperature (SST) is used as a measure of the amplitude of the El Niño3-Southern Oscillation (ENSO). The Niño3 SST index is defined as the seasonal SST averaged over the central Pacific (5S–5N, 90–150W). The sea surface temperature fields are blended from ship, buoy and bias-corrected satellite data (Reynolds and Smith, 1994). These data (1864–1950) are available from http:// ingrid.ldgo.columbia.edu/SOURCES/.Indices/.nino/.KAPLAN, while the data for January 1951–December 2000 are obtained from the Climate Prediction Center (CPC). Methods The normality of the data series is first tested in the study by applying the Kolmogorov–Smirnov test (Xu, 2001). The method first compares the specified theoretical cumulative distribution function (in our case normal distribution) with the sample cumulative density function based on observations, then calculates the maximum deviation, D, of the two. If, for the chosen significance level, the observed value of D is greater than or equal to the critical tabulated value of the Kolmogorov–Smirnov statistic, the hypothesis of normal distribution is rejected. After this step, continuous wavelet transform, wavelet coherence and cross-wavelet transform were performed on annual maximum streamflow of Yichang, Hankou and Datong stations and Nino3 SST series. The continuous wavelet transform (CWT) (Torrence and Compo, 1998) is used in this study. We assume that xn is a time series with equal time spacing dt and n = 0, . . . , N 1. wo (g) is a wavelet function which depends on a dimensionless ‘time’ parameter g with zero mean and localized in both frequency and time (Farge, 1992; Torrence and Compo, 1998). Because Morlet wavelet provides a good 268 Q. Zhang et al. balance between time and frequency localizations, we applied the Morlet wavelet that is defined as wo ðgÞ ¼ p1=4 eixo g eg 2 =2 ð1Þ ; where xo is the nondimensional frequency, here taken to be 6 to satisfy the admissibility condition (Farge, 1992; Torrence and Compo, 1998). The continuous wavelet transform of a discrete sequence xn is defined as the convolution of xn with a scaled and translated version of wo(g): 0 N1 X ðn nÞdt ; ð2Þ W n ðsÞ ¼ xn0 w s n0 where the asterisk indicates the complex conjugate. Because the wavelet is not completely localized 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 (Grinsted et al., 2004; Torrence and Compo, 1998). 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 (Pk). It is assumed that the time series has a mean power spectrum, given by (3); if a peak in the wavelet power spectrum is significantly above this background spectrum, then it can be assumed to be a true feature with a certain confidence level. The ‘‘95% confidence interval’’ refers to the range of confidence about a given value. To determine the 95% confidence level (significant at 5%), one multiplies the background spectrum (3) by the 95th percentile value for v2 (Torrence and Compo, 1998). Many geophysical series have the red noise characteristics which can be modeled by a first-order autoregressive (AR(1)) process. The Fourier power spectrum of an AR(1) process with lag-1 autocorrelation a (estimated from the observed time series, e.g. Allen and Smith, 1996) is given by (Grinsted et al., 2004) Pk ¼ 1 a2 j1 a e2ipk j2 ð3Þ ; where k is the Fourier frequency index. Torrence and Compo (1998) used the Monte Carlo method to show that the probability that the wavelet power of a process with a given power spectrum (Pk) is greater than p is ! jW Xn ðsÞj2 1 < p ¼ pk v2v ðpÞ; ð4Þ P 2 r2X where v is equal to 1 for real and 2 for complex wavelets. We use the circular mean of the phase over regions with >95% confidence level which is outside the COI to quantify the phase relationship. The circular mean of a set of angles (ai, i = 1, . . . , n) is defined as (Zar, 1999; Grinsted et al., 2004): am ¼ argðX; YÞ; where X ¼ n X i¼1 cosðai Þ and Y ¼ n X sinðai Þ i¼1 ð5Þ Cross-wavelet power reveals areas with a high common power. As for the covariance of two time series, Torrence and Compo (1998) defined the cross-wavelet spectrum of two time series X and Y with wavelet transform WX and WY as W XY ðs; tÞ ¼ W X ðs; tÞW Y ðs; tÞ; ð6Þ where the asterisk denotes complex conjugation. The phase angle of WXY describes the phase relationship between X and Y in time-frequency space. Statistical significance is estimated against a red noise model (Torrence and Compo, 1998). Another useful tool is the wavelet coherence. Coherence is a measure of the intensity of the covariance of the two series in time-frequency space, unlike the cross-wavelet power which is a measure of the common power. Again, beginning with the approach of Torrence and Webster (1999), the coherence was defined as R2n ðsÞ ¼ 2 jSðs1 W XY n ðsÞÞj Sðs1 jW Xn ðsÞj2 Þ Sðs1 jW Yn ðsÞj2 Þ ; ð7Þ where S is a smoothing operator. The scales in time and frequency over which S is smoothing define the scales at which the coherence measures the covariance. We write the smoothing operator S as (Jevrejeva et al., 2003) SðWÞ ¼ Sscale ðStime ðWðs; tÞÞÞ; ð8Þ where Sscale denotes smoothing along the wavelet scale axis and Stime smoothing in time, which are given by (Torrence and Webster, 1998): t2 Stime ðWÞjs ¼ W n ðsÞ c12s2 ; s Y ð9Þ Stime ðWÞjs ¼ W n ðsÞ c2 ð0:6sÞ ; n where c1 and c2 are normalization constants, and is the rectangle function. The factor of 0.6 is the empirically determined scale decorrelation length for the Morlet wavelet (Torrence and Compo, 1998). Monte Carlo method is used with a red noise to determine the 95% statistical confidence level of the coherence (Torrence and Webster, 1999; Jevrejeva et al., 2003). Results and discussions The results of the Kolmogorov–Smirnov test and the serial correlation analysis (not shown) reveal that the annual maximum streamflow at the three stations in the Yangtze River are normally distributed and serial correlations are either nonsignificant at 95% confidence level or relatively small. This means that the use of cross-wavelet analysis and wavelet coherence is warranted. Wavelet power spectra for Niño3 SST The wavelet power spectra for the Niño3 SST (December– February) are shown in Fig. 2, which reveal that the power is broadly distributed with peaks in the 2- to 8-year ENSO band. The 95% confidence regions demonstrate that 1875– 1920 and 1960–1990 include intervals of higher ENSO variance, while lower ENSO variance is found during 1920– Possible influence of ENSO on annual maximum streamflow of the Yangtze River, China 269 Datong station runoff Figure 2 Continuous wavelet power spectrum for the normalized time series of Niño3 SST (December–February). 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. The normalized Niño3 SST (December–February) has the AR(1) coefficient of 0.023. 1960. Similar changing patterns were also discovered in east Pacific SST and tropical zonal winds (Wang and Wang, 1996; Gu and Philander, 1995; Torrence and Webster, 1998). Therefore, only Niño3 SST is used in the study as a measure of the amplitude of the El Niño3-Southern Oscillation (ENSO). Continuous wavelet power spectra for the runoff time series of annual maximum streamflow of Datong station show a high wavelet power in the 3- to 8-year band around 1970–1985 (upper graph of Fig. 3). The wavelet power of the annual maximum streamflow of Datong station is not significant at >95% confidence level during 1970–1985. The El Niño3 SST has a significant wavelet power in 1970–1985 (Fig. 2). The lower left graph of Fig. 3 demonstrates a significant common power in the 3- to 8-year band from 1975 to 1988. It can also be seen from crosswavelet transform that the annual maximum streamflow of Datong station and Niño3 SST are in the same phase in the sectors with a significant common power (significant at >95% confidence level). The wavelet coherence (the lower right graph of Fig. 3) shows how coherent the cross-wavelet transform is in the time-frequency space (Torrence and Compo, 1998; Grinsted et al., 2004). The squared wavelet coherence (WTC) is shown in the lower right graph of Fig. 3. A relatively larger region (in the 3- to 7-year band during 1975–1985) is prominent and is significant at >95% confidence level. This region shows the in-phase relationship between annual maximum streamflow of Datong station and the Niño3 SST. Part of Figure 3 Continuous wavelet power spectrum for the normalized time series of annual maximum streamflow of Datong station (upper graph). 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. The normalized annual maximum streamflow of Datong station has the AR(1) coefficient of 0.029. The lower left graph is the cross-wavelet transform and the lower right graph is squared wavelet coherence result, showing the relations between annual maximum streamflow of Datong station and Niño3 SST (December– February). The relative phase relationship is shown as arrows (with in-phase pointing right, anti-phase pointing left). 270 the regions covered by COI shows an anti-phase relation. However, the in-phase relation is the dominant one. Hankou station runoff The continuous wavelet power spectra for the time series of annual maximum streamflow of Hankou station (upper graph in Fig. 4) show a high wavelet power in the 2- to 8-year band around 1920–1960 and 1970–1980. The peak power value occurred in the 2- to 8-year band, which is in good agreement with the continuous wavelet power analysis of Niño3 SST and COI (Torrence and Webster, 1999). The wavelet power spectra for annual maximum streamflow indicate a significant (at 95% confidence level) nonstationarity of variance in the 2- to 8-year band, especially during 1920–1930 and 1950–1960, indicating the strongest fluctuation occurred in about 1923 and 1947. In the 9- to 16-year band, there also exist regions with a higher wavelet power, but not significant at >95% confidence level. The cross-power spectra results (lower left graph of Fig. 4) show a significant common power in the 2- to 8-year band during 1875–1882 and 1960–1980. However, the phase changes in these regions with the significant wavelet power show ambiguous changing patterns as compared with that of Datong station. But the anti-phase in these sectors is still relatively obvious. The squared wavelet coherence (WTC) (lower right graph of Fig. 4) demonstrates that more regions (in the 2- to 16-year Q. Zhang et al. band during 1880–1920 and 1960–1985) are prominent and are significant at >95% confidence level. The phase relations within these regions between annual maximum streamflow of Hankou station and Niño3 SST are not stable. The phase relations in the 2- to 4-year band and the 5- to 8-year band during 1880–1885 have in-phase and anti-phase simultaneously, showing the changing and ambiguous phase relationships between annual maximum streamflow of Hankou station and Niño3 SST. Yichang station runoff Fig. 5 (upper graph) shows the continuous wavelet power spectra for the time series of annual maximum streamflow of Yichang station. The significant wavelet power spectra are in the 2- to 7-year band during 1885–1905 and 1940– 1950, and in the 8- to 16-year band during 1910–1950 and 1970–1980. There are common features in the wavelet power of the two time series (annual maximum streamflow of Yichang station and Niño3 SST) such as the significant peak in the 4- to 8-year band around 1935–1945; they also have a high power in the 2- to 4-year band around 1900 and the 4- to 8-year band in 1980 (Fig. 2 and upper graph of Fig. 5). Cross-wavelet power spectra (lower left graph of Fig. 5) show common features with significant wavelet power spectra at >95% confidence level, and these significant common Figure 4 Continuous wavelet power spectrum for the normalized time series of annual maximum streamflow of Hankou station (upper graph). 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. The normalized annual maximum streamflow of Hankou station has the AR(1) coefficient of 0.054. The lower left graph is the cross-wavelet transform and the lower right graph is squared wavelet coherence result, showing the relations between annual maximum streamflow of Hankou station and Niño3 SST (December– February). The relative phase relationship is shown as arrows (with in-phase pointing right, anti-phase pointing left). Possible influence of ENSO on annual maximum streamflow of the Yangtze River, China 271 Figure 5 Continuous wavelet power spectrum for the normalized time series of annual maximum streamflow of Yichang station (upper graph). 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. The normalized annual maximum streamflow of Yichang station has the AR(1) coefficient of 0.151. The lower left graph is the cross-wavelet transform and the lower right graph is squared wavelet coherence result, showing the relations between annual maximum streamflow of Yichang station and Niño3 SST (December– February). The relative phase relationship is shown as arrows (with in-phase pointing right, anti-phase pointing left). power occurred in the 2- to 4-year band during 1890–1900 and the 8- to 16-year band during 1920–1930. These phase relations in the regions that are significant at >95% confidence level show clear anti-phase relations between annual maximum streamflow and Niño3 SST. The changes of phase relations are relatively complex. The squared wavelet coherence (WTC) (lower right graph of Fig. 5) shows that in 1920–1950 there are two regions with high coherency peaks at the 8- to 16-year band and the 20- to 32-year band, respectively. These regions correspond to the significant period of Niño3 SST and annual maximum streamflow of Yichang station during 1920–1950. A visual comparison of annual maximum streamflow of Yichang station and Niño3 SST suggests that higher Niño3 SST usually corresponds to a smaller annual maximum streamflow at Yichang station. The phase changes in the regions that are significant at 95% confidence level are dominated by anti-phase relations. It should be noted that the phase is changed from 180 to 90 in the regions in the 8- to 16-year band during 1900– 1940. The phase changes in the regions in the 20- to 32-year band during 1920–1950 are also changed. These phase changes are related to the time lag between Niño3 SST and annual maximum streamflow of Yichang station. Therefore, it can be said that the annual maximum streamflow of the upper Yangtze River basin is not only influenced by one factor of Niño3 SST, but also by East summer monsoon in the upper Yangtze Basin. Research results (Torrence and Web- ster, 1999) indicated that during El Nino the Indian monsoon tends to be weaker, yet the weak monsoon actually occurs approximately 4 months before the peak Niño3 SST (Webster et al., 1998). The strong monsoon is usually associated with cold Niño3 SST (Torrence and Webster, 1999). This may be the reason that the phase relations between annual maximum streamflow and Niño3 SST are changing in anti-phase relation. Summary and discussion Cross-wavelet analysis and wavelet coherence are powerful methods for testing a proposed linkage between two time series (Grinsted et al., 2004). Relationships between annual maximum streamflow of three monitoring stations of the Yangtze River and ENSO are detected by cross-wavelet and wavelet coherence. The following points can be concluded from the study: 1. The relationship between annual maximum streamflow and ENSO is changing from the lower Yangtze River Basin to the upper Yangtze River Basin. In the lower Yangtze Basin, the in-phase relationship occurred between annual maximum streamflow and ENSO. In the upper Yangtze Basin, however, the anti-phase relationship is dominant. The phase relation of annual maximum streamflow and 272 ENSO is ambiguous in the middle Yangtze, demonstrating that the middle Yangtze River is the transition zone. It can be seen from Fig. 5 that from the 8-year band to the 16-year band the phase angle is changing, showing that more than one factor influences the annual maximum streamflow from the upper Yangtze Basin. 2. Continuous wavelet transform analysis results indicate that the annual maximum streamflow of the middle and the lower Yangtze River is dominated by the 2- to 8-year periods. However, the annual maximum streamflow of the upper Yangtze River is dominated by both the 2- to 4-year and the 8- to 16-year periods. Wavelet power relations and phase relations between annual maximum streamflow of the Yangtze River (at Datong, Hankou and Yichang stations) and ENSO are relatively stable in the longer periods (>2- to 16-year band), and are relatively unstable in the shorter periods (mainly in <2-year band), demonstrating that from the viewpoint of longer periods, the annual maximum streamflow of the Yangtze River is controlled by the slowly changing climate, e.g. the large-scale ocean–atmosphere circulation of moisture. During shorter periods, however, the annual maximum streamflow of the Yangtze River is not only impacted by ENSO, but also by other factors, like urbanization (Jain and Lall, 2001). 3. The upper Yangtze River is mainly influenced by the Indian summer monsoon, and the lower Yangtze River is controlled by East Asian summer monsoon (Ding and Chan, 2005). A causal relationship was already found by studies based on observations (Hu et al., 2000) and by a coupled model study (Wei, 2005). The modeling results from the ECHAM4 general circulation model (GCM) (Cheng et al., 2005) indicate that the increases in SST will strengthen the convective precipitation in the lower Yangtze River Basin. Floods in East China including the lower Yangtze River is more likely caused by the strengthened convective precipitation associated with the increases in SST. The different phase relationships between ENSO and annual maximum streamflow in the lower and the upper Yangtze River show the different influences of different variables of the atmospheric system. Hydrological teleconnections linking climate indicators and the associated atmospheric fluxes of moisture with a considerable spatial and temporal structure are already understood through statistical analysis and numerical modeling results (Jain and Lall, 2001). The precipitation of the Yangtze River Basin is also influenced by snow conditions of the Tibet Plateau, which in turn, impacts the annual maximum streamflow. This can be one of the reasons for the more complex changes of phase angles outside the regions exceeding 95% confidence level. 4. The Asian monsoon system can be divided into two subsystems, the South Asian (or Indian) and the East Asian monsoon systems, which are independent of each other and, at the same time, interact with each other (Zhu, 1934; Yeh et al., 1959; Tao and Chen, 1987; Ding and Chan, 2005). Zhu et al. (1986) pointed out that the interaction between South Asian monsoon and East Asian monsoon might be accomplished by energy exchange, the propagation of low-frequency oscillation, and moisture transport. Nino3 is a numerical measure of sea surface Q. Zhang et al. temperatures in the tropical Pacific which may be used to identify and categorize ENSO events (Trenberth, 1997). It should be mentioned that ENSO affects monsoon circulation and vise versa (Roy, 1998). There are still mechanisms that are not well understood concerning relationships between ENSO and various climatic/hydrological hazards in China (Cheng et al., 1998). Many scientific uncertainties exist in the understanding and forecasting of ENSO and its impacts due to the lack of proper observation networks (UNU report, 2000). This can also be seen from the research results of this paper. The phase relation between ENSO and annual maximum runoff of the Yangtze River basin is changing in shorter periods. However, the discussion of mechanism and possible teleconnections between ENSO and East Asian summer monsoon or Indian monsoon is out of the scope of the current research. The research results of this paper present apparent opportunities for improving forecasting of streamflow especially annual maximum streamflow along the mainstream of the Yangtze River basin, which, in turn, will improve water resources management and human mitigation to hydrological hazards. Therefore, it will be greatly helpful for planning human adaptation to extreme hydrological events in the Yangtze River basin based on ENSO events. The runoff changes of the tributaries of the Yangtze River basin may be influenced by water reservoirs. The runoff of the mainstream of the Yangtze River is less impacted by the water reservoirs along the mainstream of the Yangtze River basin. Further study will be carried out on the influences of the Three Gorges Dam on runoff changes in the mainstream after its completion in 2009. Further investigation will also be performed on possible teleconnections between monthly maximum streamflow and ENSO events in the basin. Acknowledgements This research was financially supported by Alexander von Humboldt Foundation, Germany and Outstanding Oversea Chinese Scholars Fund from CAS (The Chinese Academy of Sciences) and Foundation from Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences (Grant No. S260018). Great thanks should be extended to Changjiang Water Resources Commission (Ministry of Water Resources, China) for providing the hydrological data and to two anonymous reviewers and Dr. Andreja Pisnik for their invaluable and constructive suggestions which greatly improved the quality of this paper. Wavelet software was provided by C. Torrence and G. Compo, and is available at: http:// paos.colorado.edu/research/wavelets/. Software by Grinsted, A., et al., is available at http://www.pol.ac.uk/ home/research/waveletcoherence/. References A UNEP/NCAR/UNU/WMO/ISDR Assessment, 2000. <http://www.unu.edu/env/govern/ElNino/CountryReports/pdf/china.pdf/>. Aceituno, P., 1988. On the functioning of the Southern Oscillation in the South American sector. Part I: Surface climate. Monthly Weather Review 116, 505–524. Possible influence of ENSO on annual maximum streamflow of the Yangtze River, China Allen, M.R., Smith, L.A., 1996. Monte Carlo SSA: detecting irregular oscillations in the presence of coloured noise. Journal of Climatology 9, 3373–3404. Amarasekera, K.N., Lee, R.F., Willianms, E.R., Eltahir, E.A.B., 1997. ENSO and the natural variability in the flow of tropical rivers. Journal of Hydrology 200, 24–39. Barlow, M., Nigam, S., Berbery, E.H., 2001. ENSO, Pacific decadal variability, and US summertime precipitation, drought, and streamflow. Journal of Climate 14, 2105–2128. Camilloni, A.I., Barros, R.V., 2003. Extreme discharge events in the Paraná River and their climate forcing. Journal of Hydrology 278, 94–160. Cane, M.A., 1992. Tropical Pacific ENSO models: ENSO as a mode of the coupled system. In: Trenberth, K.E. (Ed.), Climate System Modeling. Cambridge University Press, New York, pp. 583–616. Cardoso, A.O., Silva Dias, P.L., 2006. The relationship between ENSO and Paraná River flow. Advances in Geosciences 6, 189– 193. Chang, W.Y.B., King, G., 1999. Centennial climate changes and their global associations in the Yangtze River (Chang Jiang) Delta, China and subtropical Asia. Climate Research 2, 95– 103. Changjiang Water Resources Commission (Ministry of Water Resources, China) (CWRC), 2000. Hydrological Records of the Yangtze River. Cyclopaedia Press of China, Beijing (in Chinese). Cheng, Z.H., Kang, D., Chen, L.S., Xu, X.D., 1998. Interaction between tropical cyclone and Mei-yu front. Acta Meteorological Sinica 13 (1), 35–46. Cheng, Y.J., Lohmann, U., Zhang, J.H., Luo, Y.F., Liu, Z.T., Lesins, G., 2005. Contribution of changes in sea surface temperature and aerosol loading to the decreasing precipitation trend in Southern China. Journal of Climate 18, 1381–1390. Dilley, M., Heyman, B.N., 1999. ENSO and disaster: droughts, floods and El Niño/Southern Oscillation warm events. Disasters 19 (3), 181–193. Ding, Y.H., Chan, J.C.L., 2005. The East Asian summer monsoon: an overview. Meteorology and Atmospheric Physics 89, 117–142. Farge, M., 1992. Wavelet transform and their application to turbulence. Annual Review of Fluid Mechanics 24, 395–457. Gong, D.Y., Wang, S.W., 1999. Impacts of ENSO on global precipitation changes and precipitation in China. Chinese Science Bulletin 44 (3), 315–320, in Chinese. Grinsted, A., Moore, J.C., Jevrejeva, S., 2004. Application of the cross wavelet transform and wavelet coherence to geophysical time series. Nonlinear Processes in Geophysics 11, 561–566. Gu, D., Philander, S.G.H., 1995. Secular changes of annual and interannual variability in the Tropics during the past century. Journal of Climate 8, 864–876. Hu, Z., Latif, M., Roeckner, E., Bengtsson, L., 2000. Intensified Asian summer monsoon and its variability in a coupled model forced by increasing greenhouse gas concentrations. Geophysical Research Letter 27, 2681–2684. Intergovernmental Panel on Climate Change (IPCC), 2001. In: Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J., Xiaosu, D. (Eds.), Climate Change 2001: The Scientific Basis, Contribution of Working Group I to the Third Assessment Report of IPCC 2001. Cambridge University Press, Cambridge. Jain, S., Lall, U., 2000. The magnitude and timing of annual maximum floods: Trends and large-scale climatic associations for the Black-smith Fork River, Utah. Water Resources Research 36 (12), 3641–3652. Jain, S., Lall, U., 2001. Floods in a changing climate: does the past represent the future? Water Resources Research 37 (12), 3193– 3250. Jevrejeva, S., Moore, J.C., Grinsted, A., 2003. Influence of the Arctic Oscillation and El Niño-Southern Oscillation (ENSO) on ice conditions in the Baltic Sea: the wavelet approach. Journal of 273 Geophysical Research 108 (21), 4677. doi:10.1029/ 2003JD003417. Jiang, T., Zhang, Q., Zhu, D.M., Wu, Y.J., 2006. Yangtze floods and droughts (China) and teleconnections with ENSO activities (1470–2003). Quaternary International 144 (1), 29– 37. Lan, Y.C., Ma, Q.J., Kang, E., Zhang, J.S., Zhang, Z.H., 2002. Relationship between ENSO cycle and abundant or low runoff in the upper Yellow River (China). Journal of Desert Research 22 (3), 262–266, in Chinese. Menzel, L., Bürger, G., 2002. Climate change scenarios and runoff response in the Mulde catchment (Southern Elbe, Germany). Journal of Hydrology 267, 53–64. Olsen, J.R., Stedinger, J.R., Matalas, N.C., Stakhiv, E.Z., 1999. Climate variability and flood frequency estimation for the upper Mississippi and lower Missouri rivers. Journal of American Water Resource Association 35 (6), 1509–1523. Reynolds, R.W., Smith, T.M., 1994. Improved global sea surface temperature analyses. Journal of Climate 7, 929–948. Robson, A.J., Jones, T.K., Reed, D.W., Bayliss, A.C., 1998. A study of national trend and variation in UK floods. International Journal of Climatology 18, 168–182. Roy, N.S., 1998. ENSO and the Asian monsoon. The ENSO signal 9. e.g. <http://www.ogp.noaa.gov/library/ensosig/ensosig9.htm# Monsoon/>. Tao, S., Chen, L., 1987. A review of recent research on the East Asian summer monsoon. In: Chang, C.-P., Krishnamurti, T.N. (Eds.), China, Monsoon Meteorology. Oxford University Press, Oxford, pp. 60–92. Torrence, C., Compo, G.P., 1998. A practical guide to wavelet analysis. Bulletin of American Meteorological Society 79, 61–78. Torrence, C., Webster, P.J., 1998. The annual cycle of persistence in the El Niño-Southern Oscillation. Quarterly Journal of the Royal Meteorological Society 124, 1985–2004. Torrence, C., Webster, P.J., 1999. Interdecadal changes in the ENSO-monsoon system. Journal of Climatology 12, 2679–2690. Trenberth, K.E., 1997. The definition of El Niño. Bulletin of the American Meteorological Society 78, 2771–2777. Trenberth, K.E., Branstator, G.W., Karoly, D., Kumar, A., Lau, N.-C., Ropelewski, C., 1998. Progress during TOGA in understanding and modeling global teleconnections associated with tropical sea surface temperatures. Journal of Geophysical Research 103 (7), 14291–14324. Wang, B., Wang, Y., 1996. Temporal structure of the Southern Oscillation as revealed by wave-form and wavelet analysis. Journal of Climate 9, 1586–1598. Webster, P.J., Magaña, V., Palmer, T.N., Shukla, J., Tomas, R.A., Yanai, M., Yasunari, T., 1998. Monsoons: Processes, predictability and the prospects for prediction. Journal of Geophysical Research 103, 14451–14510. Wei, M., 2005. A coupled model study on the intensification of the Asian summer monsoon in IPCC SRES scenarios. Advance in Atmospheric Sciences 22 (6), 798–806. Xu, C.-Y., 2001. Statistical analysis of a conceptual water balance model, methodology and case study. Water Resources Management 15, 75–92. Yeh, T.C., Tao, S.Y., Li, M.C., 1959. The abrupt change of circulation over the Northern Hemisphere during June and October. In: Bolin, B. (Ed.), The Atmosphere and the Sea in Motion. Rockefeller Inst. Press, New York, pp. 249–267. Yin, H.F., Li, C.A., 2001. Human impact on floods and flood disasters on the Yangtze River. Geomorphology 41, 105–109. Zar, J.H., 1999. Biostatistical Analysis. Prentice-Hall, Old Tappan, NJ. Zhang, Q., Jiang, T., Gemmer, M., Becker, S., 2005. Precipitation, temperature and discharge analysis from 1951 to 2002 in the 274 Yangtze Catchment, China. Hydrological Sciences Journal 50 (1), 65–80. Zhang, Q., Liu, C.-L., Xu, C.-Y., Xu, Y.-P., Jiang, T., 2006. Observed trends of annual maximum water level and streamflow during past 130 years in the Yangtze River basin, China. Journal of Hydrology 324, 255–265. Q. Zhang et al. Zhu, K.Z., 1934. Monsoons in Southeast Asia and rainfall amount in China. Acta Geogr Sinica 1, 1–27. Zhu, Q., He, J., Wang, P., 1986. A study of the circulation differences between East Asian and India summer monsoon with their interaction. Advances in Atmospheric Sciences 3, 446–477.