Meteorol Atmos Phys (2015) 127:273–288 DOI 10.1007/s00703-014-0360-2 ORIGINAL PAPER Variations of annual and seasonal runoff in Guangdong Province, south China: spatiotemporal patterns and possible causes Qiang Zhang • Mingzhong Xiao • Vijay P. Singh Chong-Yu Xu • Jianfeng Li • Received: 2 March 2014 / Accepted: 25 November 2014 / Published online: 4 December 2014 Springer-Verlag Wien 2014 Abstract In this study, we thoroughly analyzed spatial and temporal distributions of runoff and their relation with precipitation changes based on monthly runoff dataset at 25 hydrological stations and monthly precipitation at 127 stations in Guangdong Province, south China. Trends of the runoff and precipitation are detected using Mann–Kendall trend test technique. Correlations between runoff and precipitation are tested using Spearman’s and Pearson’s correlation coefficients. The results indicate that: (1) annual maximum monthly runoff is mainly in decreasing tendency Responsible Editor: C. Simmer. Q. Zhang (&) M. Xiao Department of Water Resources and Environment, Sun Yat-sen University, Guangzhou 510275, China e-mail: zhangqnj@gmail.com; zhangq68@mail.sysu.edu.cn Q. Zhang M. Xiao Key Laboratory of Water Cycle and Water Security in Southern China of Guangdong High Education Institute, Sun Yat-sen University, Guangzhou 510275, China Q. Zhang School of Earth Sciences and Engineering, Suzhou University, Anhui 234000, China V. P. Singh Department of Biological and Agricultural Engineering and Zachry Department of Civil Engineering, Texas A&M University, College Station, Texas, USA C.-Y. Xu Department of Geosciences, University of Oslo, P O Box 1047, Blindern, 0316 Oslo, Norway J. Li Department of Geography and Resource Management, and Institute of Environment, Energy and Sustainability, The Chinese University of Hong Kong, Hong Kong, China and significant increasing annual minimum monthly runoff is observed in the northern and eastern Guangdong Province. In addition, annual mean runoff is observed to be increasing at the stations located in the West and North Rivers and the coastal region; (2) analysis of seasonal runoff variations indicates increasing runoff in spring, autumn and winter. Wherein, significant increase of runoff is found at 8 stations and only 3 stations are dominated by decreasing runoff in winter; (3) runoff changes of the Guangdong Province are mainly the results of precipitation changes. The Guangdong Province is wetter in winter, spring and autumn. Summer is coming to be drier as reflected by decreasing runoff in the season; (4) both precipitation change and water reservoirs also play important roles in the increasing of annual minimum monthly streamflow. Seasonal shifts of runoff variations may pose new challenges for the water resources management under the influences of climate changes and intensifying human activities. 1 Introduction Climate variability and change characterized by global warming exert tremendous influences on our planet, and it is also the case for the hydrological cycle (IPCC 2007). Arnell (1999) indicates that under the influences of global warming, the hydrological cycle will be intensified with more evaporation and more precipitation and the extra precipitation will be unequally distributed around the globe, resulting in more frequent floods and droughts. Some researches tend to advocate that the present warming tendency led to changes of the global hydrological cycle and to the amplitude of the increase of global and continental 123 274 runoff (e.g., Labat et al. 2004; Semenov and Bengtsson 2002). Projected global climate changes will probably further accelerate the global hydrological cycle, and which in turn will alter the spatiotemporal patterns of the flood/ drought hazards in both the global and the regional scales. Previous studies have shown that under the influence of the well-evidenced global warming situation, precipitation patterns are expected to change and extreme weather events (e.g., floods, droughts, and rainstorms) are likely to occur more frequently (WMO 2003). This situation is worsening by the sharp increase in water consumption owing to the population explosion, unprecedented rise in standard of living, and enormous economic development (Xu and Singh 2004). The hydrologic regime of a stream under specific geomorphic conditions represents the integrated basin response to various climatic inputs, with precipitation and temperature being very important ones (Zhang et al. 2001). Investigations of statistical properties of runoff variations are critical for evaluation of regional availability and variability of water resources and have been attached considerable importance (Zhang et al. 2001; Burn and Elnur 2002; Birsan et al. 2005; George 2007; Brabets and Walvoord 2009). Yang and Tian (2009) indicate that runoff in Haihe River basin of China is steadily declining due to climate change and human activity. Annual runoff and annual sediment load of the Yangtze River basin were systematically studied using Mann–Kendall trend method. Increasing annual runoff, especially in the lower Yangtze River basin was identified (Zhang et al. 2006). Zhang et al. (2009a) indicate that the runoff of the Yellow River basin is decreasing and water resource deficit tends to be more serious from the upper to the lower Yellow River basin. Zhang et al. (2008) analyzed annual water discharge and sediment load series (from the 1950 to 2004) at nine stations in the main channels and main tributaries of the Zhujiang (Pearl River), demonstrating that water discharges at all stations in the Zhujiang Basin show no significant trend or abrupt shift. Annual water discharges in the Pearl River basin are mainly influenced by precipitation variability. However, Lu (2004) indicates that water discharge in most Chinese rivers has experienced great changes due to climate change and anthropogenic impacts in the drainage basins. Guangdong Province involves the Pearl River Delta and the lower Pearl River basin, being the highly developed region in terms of social economy in China. Thorough evaluation of the availability and changes of water resources, in particular runoff, across the Guangdong Province, is of great scientific and practical merits, and the importance and significance of this current study are evident. However, to the best of our knowledge, changing properties of runoff in both time and space in the region 123 Q. Zhang et al. have not been thoroughly investigated. This is the major motivation of this study. Therefore, the major objective of this study is to analyze trends in various discharge parameters observed at 25 hydrological stations over the past 31–45 years across the Guangdong Province. Possible underlying causes behind the runoff variations are also discussed. 2 Data Seven data series, i.e., annual maximum and minimum monthly runoff, annual and seasonal mean runoff at 25 hydrological stations are selected for analysis. The detailed information of the dataset is displayed in Table 1. The locations of the discharge stations can be shown in Fig. 1. In the hydrological data, there is no missing data. To investigate possible influences of climate changes and human activities, precipitation from 127 rain stations and the information of the large- and middle-size water reservoirs covering 1956–2000 are collected. The locations of the precipitation stations are also illustrated in Fig. 1. The precipitation and discharge data before 1989 are extracted from the Hydrologic Year Book (published by the Hydrologic Bureau of the Ministry of Water Resources of China) and those after 1989 are provided by the Hydrological Bureau of Guangdong Province. Missing data in the precipitation dataset, accounting for \0.01 % of the total data, were found in March during 2001–2008 and were filled in with average values of neighboring months (Zhang et al. 2011). Xinfengjiang Reservoir and Fengshuba Reservoir are also taken into account to tentatively study the influences of the construction and operation of water reservoirs on the streamflow. Both reservoirs locate in the East River basin, a highly developed water basin in the south China. Table 2 displays the details about these two reservoirs. The information of reservoirs is extracted from the previous study (Zhou et al. 2012). 3 Methodology There are many statistical techniques available to detect trends within the time series such as moving average, linear regression, Mann–Kendall trend test (M–K), and filtering technology. Each method has its own strength and weakness in terms of detection of trends. However, non-parametric trend detection methods are less sensitive to outliers than parametric statistics. In addition, the rank-based non-parametric M–K test (Mann 1945; Kendall 1975) can test trends without requiring normality or linearity (Wang and Zhou 2005). Therefore, this method was highly recommended for general use by the World Meteorological Variations of annual and seasonal runoff in Guangdong Province Table 1 Detailed information of hydrological data analyzed in this study and major statistical properties of streamflow series in the study region No. 275 Drainage area (km2) Length of streamflow series Annual streamflow 23.03 351,535 1956–2000 2,217.05 0.19 23.34 8,273 1956–2000 81.80 0.26 Hydrological stations Long. 1 Gaoyao 112.28 2 Gulan 111.41 3 Guanliang 111.40 22.50 3,164 1956–2000 26.30 0.28 4 Yaogu 113.41 24.52 1,776 1959–2000 16.76 0.28 5 Lishi 113.16 23.51 6,976 1956–2000 60.46 0.30 6 Changba 112.57 23.34 6,794 1956–2000 60.60 0.28 7 8 Gaodao Hengshi 113.32 113.10 24.53 24.09 9,007 34,013 1956–2000 1956–2000 106.54 344.80 0.26 0.26 9 Shijiao 115.15 24.07 38,363 1956–2000 421.20 0.25 10 Longchuan 114.42 23.44 7,699 1956–2000 64.38 0.30 11 Heyuan 114.18 23.10 15,750 1956–2000 148.31 0.27 12 Boluo 113.51 23.21 25,325 1956–2000 237.42 0.24 13 Hengshan 112.48 23.07 12,624 1956–2000 100.70 0.30 14 Shuikou 112.50 23.10 6,480 1956–2000 51.01 0.29 15 Xikou 115.54 23.59 9,228 1956–2000 87.62 0.25 16 Chao’an 116.21 24.28 29,077 1956–2000 252.12 0.26 17 Ciyao 116.39 23.40 820 1956–2000 11.03 0.33 18 Jiaokeng 116.39 24.32 1,104 1956–2000 19.14 0.27 19 Dongqiaoyuan 116.08 23.29 2,016 1956–2000 27.80 0.27 20 Shuangjie 116.01 23.05 4,345 1956–2000 58.95 0.28 21 Huazhou 115.38 23.02 6,151 1956–2000 49.20 0.32 22 23 Gangwayao Qilinzui 111.48 110.38 21.57 21.39 3,086 2,866 1970–2000 1956–2000 17.55 38.31 0.46 0.29 24 Sanshui 110.04 21.30 – 1959–2000 437.78 0.37 25 Makou 112.17 22.52 – 1959–2000 2,321.45 0.17 Organization (Mitchell et al. 1966). This paper also uses the M–K test method to analyze trends within the runoff series over the Guangdong Province. It should be noted here that the result of the M–K test is heavily affected by serial correlation of the time series. To eliminate the effect of serial correlation on M–K results, von Storch and Navarra (1995) suggested the ‘‘pre-whitened’’ technique in the removal of effects of serial correlation before M–K analysis. Following Zhang et al. (2001), the statistically significant trend of the runoff series (x1, x2, x3, …, xn) is detected by following the procedure: (1) Compute the lag-1 serial correlation q1; (2) if q1 \ 0.1, the M–K test is applied to the time series directly; otherwise, (3) the M–K test is applied to the ‘‘pre-whitened’’ time series, i.e., x2 -q1x1, x3 -q1x2, …, xn -q1xn-1. The confidence level of 95 % is used to evaluate the significance of the trends, and the trend analysis was done on the entire period. To quantitatively evaluate the relations between runoff and precipitation variables, Spearman’s rank correlation analysis and Pearson’s correlation analysis are performed on the areal average precipitation and runoff variations. The Spearman rank correlation coefficient can be used to Lat. Mean (108 m3) Cv measure monotone association of two variables when the distribution of the data makes Pearson’s correlation coefficient undesirable or misleading. Mann–Whitney U test, also called Wilcoxon rank-sum test, is a non-parametric statistical test (Mann and Whitney 1947). Let f and g be the continuous cumulative distribution functions of two random variables x and y, respectively. The Mann–Whitney U test is to test the null hypothesis f = g against the alternative that x is stochastically smaller than y. The Mann–Whitney U test is used to assess whether the distributions of streamflow series before and after water reservoir construction are different, which indicates whether the construction of water reservoir alters the statistical behaviors of streamflow. 4 Results 4.1 Seasonal patterns of precipitation and runoff The mean annual precipitation for the study period (1956–2000) is about 1,500–2,000 mm in the region. 123 276 Q. Zhang et al. Fig. 1 Location of the study region, the hydrological stations (triangles) and the rain stations (dots) over the Guangdong Province Table 2 Detailed information of reservoirs Reservoir Construction time Drainage area (km2) Storage capacity (billion m3) Xinfengjiang 1958–1962 5,740 13.98 Fengshuba 1970–1974 5,150 1.94 Precipitation mainly occurs between April and September. Figure 2 illustrates the distribution of streamflow at Gaoyao station (No. 1 hydrological stations in Fig. 1). Figure 2a shows that higher runoff is found between April and September, particularly during June and August. Occurrence of high runoff is in good line with rainy season (Fig. 2b). More detailed examination of the relation between precipitation and discharge is provided in the discussion section. 4.2 Trends of annual maximum/minimum runoff and annual mean runoff In this study, the trends of annual maximum/minimum runoff and annual mean runoff have been analyzed. The spatial distribution of the temporal trend of annual minimum monthly runoff, i.e., the smallest monthly runoff of 1 year is demonstrated in Fig. 3. It can be observed from figure that the Guangdong Province is dominated by increasing annual minimum monthly runoff. Decreasing annual minimum monthly runoff, but non-significant at [95 % confidence level, can be found at two stations, i.e., Makou and Ciyao stations. Increasing annual minimum 123 monthly runoff is observed at 23 stations, accounting for 92 % of the total stations considered in this study. Significant increasing trends can be detected at 9 stations, accounting for 36 %. Stations with significant increasing annual minimum monthly runoff are found mainly in the eastern and northern Guangdong Province (Fig. 3). Specifically, significant increasing trend of annual minimum monthly runoff can be found in the upper East River, upper North River, and eastern Guangdong Province. Annual maximum monthly runoff changes over the Guangdong Province (Fig. 4) are not statistically significant. Insignificant decreasing annual maximum monthly runoff can be found along the coastal regions of the Guangdong Province and in the East River basin. In total 18 stations are characterized by decreasing annual maximum monthly runoff, accounting for 72 %. Figure 3 (increasing annual minimum monthly runoff) and Fig. 4 (decreasing annual maximum monthly runoff) show that the runoff variations over Guangdong Province have a decreasing degree of dispersion, except at the stations located in the northeastern Guangdong Province. Annual mean runoff variations are illustrated in Fig. 5. It can be seen from Fig. 5 that the eastern and northern Guangdong Province is dominated by insignificant increasing annual runoff, while opposition is true in the western Guangdong Province. Specifically, annual mean runoff in the East River basin and the North River basin is characterized by increasing tendency. However, these increasing or decreasing trends are non-significant at [95 % confidence level. Variations of annual and seasonal runoff in Guangdong Province 8 A Streamflow (106 m3) 7 6 5 4 3 2 1 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month 600 Precipitation (mm) 500 B 400 300 200 100 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month Fig. 2 Monthly allocation of streamflow at the Gaoyao station (a), and of areal mean precipitation in Guangdong Province (b) 4.3 Trends of seasonal runoff Figure 6 demonstrates spatial distributions of temporal trends of runoff in spring (Fig. 6a) and autumn (Fig. 6b). Generally, the Guangdong Province is dominated by increasing runoff in these two seasons. Runoff series at one station only, i.e., Sanshui station, is in significant increasing trend. Only 4 stations are characterized by insignificant decreasing trend in spring, accounting for 16 % of the total stations. The hydrological stations dominated by increasing runoff account for 84 %, though most of them are insignificant. Decreasing runoff in spring is found at the stations located along the coastal regions of the Guangdong Province. Runoff variation patterns in spring are similar with those of the annual mean runoff changes (Figs. 5, 6a). Increasing runoff in spring is observed mainly in the eastern, northern and northwestern Guangdong Province. Specifically, increasing runoff in spring can be found in the upper East River, upper North River, and eastern Guangdong Province. In autumn, there are 18 stations dominated by 277 increasing runoff, accounting for 72 % of the total hydrological stations studied in this study, and 3 of them are dominated by significant increasing runoff trends (Fig. 6b). In autumn, increasing runoff can be identified mainly in the northern and eastern Guangdong Province, particularly in the East and North River basins. Decreasing runoff can be found mainly in the western Guangdong Province. Spatial distribution of runoff changes in summer is illustrated in Fig. 7a. The majority of Guangdong Province is featured by insignificant decreasing summer runoff. There are 17 hydrological stations characterized by insignificant decreasing summer runoff, accounting for 68 % of the total stations considered in this study. Specifically, the East River basin, the North River basin and the coastal region are characterized by decreasing summer runoff. Increasing summer runoff can be identified mainly in the tributaries of the North River and in the northeastern corner of the Guangdong Province. Runoff variations in winter present distinctly different patterns in comparison with those in summer (Fig. 7a, b). Decreasing winter runoff, but not statistically significant, can be detected at only three stations, i.e., Ciyao, Shuangjie and Makou (Fig. 7b). The remaining 22 stations are dominated by increasing winter runoff, accounting 88 % of the total hydrological stations. In addition, eight stations are dominated by significant increasing winter runoff, accounting for 32 %. Stations characterized by significant increasing winter runoff are found mainly in the West River basin, the East River basin and in the eastern Guangdong Province. The monthly runoff variations are also anatomized and the results are shown in Fig. 8. It can be observed from Fig. 8a that percentage of stations characterized by increasing monthly runoff is the smallest in May and June. The larger percentages of stations dominated by increasing monthly runoff are found in December, January, February, March and April, i.e., winter and spring in this study region. Percentage of the stations with increasing monthly runoff is moderate in August, September, October and November. Percentage of stations featured by decreasing runoff is in adverse changing patterns when compared to that of the stations featured by increasing runoff. As for percentage of stations dominated by significant runoff changing trends, Fig. 8b shows that larger percentage of stations can be found mainly in December, January, February, March, and April. Smaller percentage of hydrological stations having significant changing trend is observed in summer, particularly in May to August. Figure 8c shows that the percentages of stations characterized by the increasing trends of monthly precipitation distribute very similar with Fig. 8a. The large percentages are also found in December, January, February, March, and April, and smaller percentages are observed in May and June. In September, October, and November, more increasing trends of monthly runoff are 123 278 Q. Zhang et al. Fig. 3 Spatial distributions of trends of minimum monthly streamflow changes across the Guangdong Province. Filled up triangles significant increasing trend; open up triangles nonsignificant increasing trend; filled inverted triangles significant decreasing trend; open inverted triangles nonsignificant decreasing trend Fig. 4 Spatial distributions of trends of maximum monthly streamflow changes across the Guangdong Province. Filled up triangles significant increasing trend; open up triangles nonsignificant increasing trend; filled inverted triangles significant decreasing trend; open inverted triangles nonsignificant decreasing trend found (Fig. 8a). However, as for monthly precipitation, more decreasing trends are found (Fig. 8c). A few stations have significant trends of monthly precipitation (Fig. 8d). The percentages of significant increasing trends peak in April, and those of significant decreasing trends peak in June. The percentages of significant decreasing trends of monthly runoff and precipitation both peak in June, 123 indicating that the decreasing of precipitation is important factor of the decreasing of runoff. As for the disparity in autumn (September, October, and November), it may be explained by the operation of water reservoirs. The water reservoirs usually discharge water into river in autumn, which increases the runoff in autumn even though the precipitation decreases. Variations of annual and seasonal runoff in Guangdong Province 279 Fig. 5 Spatial distributions of trends of annual streamflow changes across the Guangdong Province. Filled up triangles significant increasing trend; open up triangles nonsignificant increasing trend; filled inverted triangles significant decreasing trend; open inverted triangles nonsignificant decreasing trend 4.4 Case study of impacts of large water reservoirs on streamflow in the East River To analyze the possible impacts of water reservoirs on runoff, how Xinfengjiang Reservoir and Fengshuba Reservoir influenced the runoff in Longchuan, Heyuan, and Boluo stations is analyzed. It should be noted that Longchuan station is in the downstream of Fengshuba Reservoir, but the upstream of Xinfengjiang Reservoir, and the other two stations are in the downstream of both reservoirs. Zhou et al. (2012) split the whole streamflow series into three segments (1954–1957, 1963–1969, 1975–2009) to study the impacts of Fengshuba Reservoir and Xinfengjiang Reservoir according to their construction periods. In this study, as Longchun station is only regulated by the Fengshuba Reservoir, its streamflow series is divided into two segments (1956–1969, 1975–2000); as the other two stations are influenced by both Fengshuba Reservoir and Xinfengjiang Reservoir, their series are divided into three segments (1956–1957, 1963–1969, 1975–2005). Mann–Whitney U test is applied to assess whether or not the distributions of the continuous two segments are different. The results of the Mann–Whitney U test are listed in Table 3. In Table 3, the change point ‘‘1975’’ denotes the distributions of 1963–1969 and 1975–2005 are significantly different, which was the result of the construction of Fengshuba Reservoir. Change point of annual streamflow is detected only in Heyuan, while there are no change points for the other two stations. The constructions of two reservoirs did not cause significant alterations in the maximum monthly streamflow, as no change point is detected in all three stations. However, the minimum monthly streamflow in these three stations is remarkably influenced by the constructions of the Fengshuba Reservoir. No change point in 1963, caused by the Xinfengjiang Reservoir, is detected. The absence of change point in ‘‘1963’’ should be paid attention, because the streamflow data before the construction of Xinfengjiang Reservoir are very limited, which makes the results less reliable. The results indicate that the large reservoir in the East River had no impacts on the maximum monthly streamflow, but had significant impacts on the minimum monthly streamflow. Only annual streamflow in Heyuan changed due to water reservoirs. The case study in the East River sheds light on how the reservoirs influenced the streamflow in south China. 5 Discussions Runoff variations are the integrated results of geomorphological conditions of the river basin, precipitation changes, human activities such as withdrawal of freshwater and building of water reservoirs, land use changes and so on. To confirm the important role played by precipitation variations, we first analyze spatial distribution of precipitation changes (Fig. 9). Annual maximum monthly precipitation is decreasing across the study region (Fig. 9a). Most of the decreasing trends are insignificant, but four stations in the middle of Guangdong have significant decreasing trends. Few stations with increasing annual maximum monthly precipitation distribute sporadically over the Guangdong Province. Similar changing properties are identified in the 123 280 Q. Zhang et al. Fig. 6 Spatial distributions of trends of seasonal streamflow changes across the Guangdong Province in spring (a) and autumn (b), respectively. Filled up triangles significant increasing trend; open up triangles non-significant increasing trend; filled inverted triangles significant decreasing trend; open inverted triangles non-significant decreasing trend spatial distribution of summer precipitation changes (Fig. 9d). Precipitation changes in winter (dry) season (Fig. 9f) show different changing characteristics when compared to those in summer. East Guangdong Province is dominated by insignificant increasing precipitation. Regions covered by stations characterized by increasing annual monthly minimum precipitation extend to 112E in comparison with those in winter (Fig. 9b, f). More stations are characterized by increasing, but non-significant precipitation 123 in spring (Fig. 9c) than in other seasons (Fig. 9d–f). Autumn is featured by insignificant decreasing precipitation (Fig. 9e). Although most of the trends are insignificant, the spatial distribution of runoff is largely in good agreement with that of precipitation changes, and it is particularly the case for the runoff changes in winter. In addition, we present the temporal variations of areal average runoff and precipitation series (Fig. 10). The runoff and precipitation series are standardized with 0 mean Variations of annual and seasonal runoff in Guangdong Province 281 Fig. 7 Spatial distributions of trends of seasonal streamflow changes across the Guangdong Province in summer (a) and winter (b), respectively. Filled up triangles significant increasing trend; open up triangles non-significant increasing trend; filled inverted triangles significant decreasing trend; open inverted triangles non-significant decreasing trend and unit standard deviation. Figure 10 clearly displays generally consistent changing properties of precipitation and runoff, showing considerable impacts of precipitation on runoff variations. It is particularly true for the relations between annual precipitation and annual runoff. To evaluate quantitatively the relations between runoff and precipitation, we analyze the Spearman’s rank correlations coefficients and Pearson’s correlation coefficients of these two variables, i.e., precipitation and runoff (Table 4). It can be observed from Table 4 that the Spearman’s correlation coefficients are all significant at [95 % confidence level except between annual minimum monthly runoff and precipitation, while the Pearson’s correlation coefficient is significant in all the cases. But the Pearson’s correlation coefficient between annual minimum monthly runoff and precipitation is also the smallest. 123 282 Q. Zhang et al. Fig. 8 Percentage of stations showing increasing and decreasing trends of monthly streamflow (a), those showing significant (at the 5 % level) increasing and decreasing trends of monthly streamflow (b), those showing increasing and decreasing trends of monthly precipitation (c), and those showing significant (at the 5 % level) increasing and decreasing trends of monthly precipitation (d) Table 3 Change points between continuous segments in Longchuan, Heyuan and Boluo Annual streamflow Maximum monthly streamflow Minimum monthly streamflow Longchuan N/A N/A 1975 Heyuan 1975 N/A 1975 Boluo N/A N/A 1975 N/A no change point is detected 123 The linear relations between areal average precipitation and runoff after being standardized are studied (Fig. 11). R2 of maximum values, spring and summer are above 0.8, indicating that precipitation and runoff are closely correlated. Whereas most correlations are positive, the relationship of minimum values is negative, whose R2 is only 0.009. Furthermore, Fig. 12 is double mass curves of areal average precipitation and runoff before being standardized. The relations between precipitation and runoff of the whole series of annual maximum monthly series, spring, and Variations of annual and seasonal runoff in Guangdong Province 283 Fig. 9 Spatial patterns of precipitation variables over the Guangdong Province. Filled up triangles significant increasing trend; open up triangles non-significant increasing trend; filled inverted triangles significant decreasing trend; open inverted triangles non-significant decreasing trend 123 284 Q. Zhang et al. Fig. 10 Temporal variations of areal average runoff and precipitation variables across the Guangdong Province 2 3 Annual maximum monthly series Annual minimum monthly series 2 1 0 0 −2 −1 1960 1970 1980 4 1990 2000 1960 1970 1980 2 Spring 1990 2000 Summer 2 0 0 −2 1960 1970 1980 3 2 1 0 −1 1990 −2 2000 1960 1970 1980 4 Autumn 1990 2000 Winter 2 0 1960 1970 1980 1990 2000 1960 1970 1980 1990 2000 2 Annual Runoff Precipitation 0 −2 1960 1970 1980 1990 2000 Table 4 Spearman’s rank correlation coefficients and Pearson’s correlation coefficients between areal average runoff and precipitation variables Correlation AM Am Spring Summer Autumn Winter Annual Spearman 0.822 0.077* 0.890 0.747 0.697 0.622 0.887 Pearson 0.829 0.336 0.849 0.828 0.710 0.785 0.932 AM annual maximum monthly series (runoff or precipitation series), Am annual minimum monthly series (runoff or precipitation series) * Correlation coefficient is non-significant at [95 % confidence level Fig. 11 Streamflow vs. precipitation relations over the Guangdong Province 5 10 Maximum values Streamflow (m3/s) −5 −2 10 0 1 2 3 Spring −4 −2 0 2 4 −2 Winter −1 0 1 2 y=0.44x−0.68 R2=0.35 0 1 2 3 4 5 Annual y=0.98x−0.98 R =0.67 −2 −1 0 Precipitation (mm) 1 2 −10 −6 0 2 4 0 2 y=1.12x−2.39 R2=0.59 0 2 −5 −3 −2 y=0.81x+0.33 2 R =0.92 −5 −4 10 Autumn 0 −3 0 y=1.18x−0.74 R2=0.88 0 −5 −1 5 123 −10 −4 5 Summer 0 −10 −6 5 0 y=1.1x+0.51 R2=0.83 −1 y=−0.18x−4.69 R2=0.009 Minimum values 0 −4 −2 Variations of annual and seasonal runoff in Guangdong Province 285 Fig. 12 Double mass curves of precipitation and runoff variables across the Guangdong Province summer are in good agreement. Nevertheless, in the annual minimum monthly series, the relationship of cumulative precipitation and runoff is not so stable, and there is an abrupt disturbance in the middle. This result is correspondent with Fig. 11. Results of this study indicate that annual minimum monthly precipitation and precipitation in winter increase in the east Guangdong Province, though this increase is not statistically significant. Precipitation in summer and annual maximum monthly values are prevailingly decreasing. Similar patterns can be identified in runoff variation in both space and time. Zhang et al. (2009b) indicate that the number of wet months in rainy season (April–September) is decreasing; the number of wet months in winter (JFD) is increasing, implying that the Pearl River basin tends to be drier in the rainy season and comes to be wetter in winter. 123 286 The analysis results of precipitation and runoff variations in this current study further corroborate the results by the previous study (Zhang et al. 2009b). Besides, different wetting and drying properties can be identified across the basin: West parts of the basin tend to be drier; and southeast parts tend to be wetter (Zhang et al. 2009b). In the Guangdong Province, the wetting tendency is mainly reflected by increasing annual mean runoff, though the increasing trends are non-significant statistically. However, significant increasing annual minimum runoff and also increasing runoff in December, January, February, March, and November further corroborate the wetting tendency of the winter, spring and autumn. Moisture content analysis based on NCAR/NCEP reanalysis dataset indicated that increasing moisture content gives rise to an increasing number of wet months in winter (Zhang et al. 2009b). In addition, we also investigated moisture budget, moisture flux in the Pearl River basin and their influences on the runoff variations and precipitation changes (Zhang et al. 2009b). The results indicated close relations between water vapor flux, precipitation and runoff, and which may imply that the altered hydrological cycle in the Pearl River basin is mainly manifested by seasonal shifts of water vapor flux after early 1960s, and hence the seasonal transition of wet and dry conditions across the Pearl River basin. The foregoing research results can greatly help to clarify the circulation background of the runoff variations in the Guangdong Province. Seasonal shifts of runoff may be a new challenge for the water resource management in the Guangdong Province. Generally, precipitation changes are the major causes behind the runoff changes in both time and space. Analysis results of this study indicate that minimum monthly runoff is increasing. Runoff in winter is also in increasing tendency and some stations are dominated by significant increasing runoff. However, the relations between precipitation and runoff variables are not consistently even in both space and time. Weaker correlation coefficients between annual minimum monthly runoff and precipitation imply other factors besides precipitation exerting influences on the runoff variations. Most rivers in the Guangdong Province are within the territory of the study region and only few stations control flows from the outside of the Guangdong Province (Fig. 1), which are excluded in the calculation of correlation coefficient between precipitation and runoff. Thus, it is not reasonable to attribute runoff changes for influencing factors outside of the Guangdong Province. Up to the end of 2000, about 175 mid- to large-size water reservoirs had been built in the Guangdong Province with the regulating storage capacity of more than 28.95 billion cubic meters. These reservoirs have the potential to change annual and seasonal location of streamflow variations. Water reservoirs may have altered the hydrological 123 Q. Zhang et al. regimes. The case study in the East River shows that water reservoirs make tremendous influences on the minimum monthly streamflow. Thus, the increasing of minimum monthly streamflow should be affected by both the precipitation changes and the water reservoirs. On the other hand, water reservoirs have no influence on the maximum monthly streamflow and little influence on the annual streamflow. The comparisons between the influences of precipitation changes and reservoir constructions on the runoff indicate that the precipitation changes are the primary causes for the variability of the annual streamflow and the maximum monthly streamflow, while both precipitation changes and reservoir constructions altered the minimum monthly streamflow. This study shows that precipitation changes are still the main causes for the temporal and spatial variability of runoff in the region. 6 Conclusions Spatial and temporal distributions of runoff variations are thoroughly analyzed using statistical techniques, Mann– Kendall trend test in this study, based on monthly runoff dataset at 25 hydrological stations in the Guangdong Province. Possible impacts of precipitation changes on streamflow changes across the Guangdong Province are also investigated and compared with previous research results. Some interesting and important conclusions are drawn as follows: 1. 2. Increasing annual minimum monthly runoff is observed at a majority of the stations. Decreasing annual maximum runoff is found at the hydrological stations located along the coast regions of the Guangdong Province. Besides, increasing annual mean runoff is observed at most of the hydrological stations, particularly in the northern and eastern Guangdong Province, specifically in the East River basins. In this sense, we can conclude that the Guangdong Province is getting wetter, but less extreme. Investigation of trends of seasonal runoff indicates increasing runoff in spring, autumn and winter. Summer is dominated by decreasing runoff though the trends are not statistically significant. Decreasing runoff is observed mainly at the stations located along the coastal regions, particularly in the southwestern corner of the Guangdong Province. More stations are characterized by significant increasing runoff in January, February, March, November and December than in other months. Therefore, the wetting tendency is detected mainly in winter, and also in spring and autumn; however, the summer is coming to be drier. Variations of annual and seasonal runoff in Guangdong Province 3. 4. The dominating seasonal pattern in this region is that the runoff is usually higher in summer than that in winter. However, the results of this study indicate seasonal shifts of the runoff variations. The runoff changes are subject to large-scale circulation of moisture flux and also moisture budget over the Pearl River basin. Our previous analysis (Zhang et al. 2009b) indicates a general wetting tendency in the lower Pearl River basin, including the Guangdong Province considered in this study, which was well reflected by increasing annual mean runoff, though annual maximum monthly runoff is decreasing. Significant increasing annual minimum monthly runoff may contribute much to the wetting tendency of the Guangdong Province and also the operations of water reservoirs which usually discharge water into their lower rivers in the dry season when annual minimum monthly runoff most likely occurs. Increasing moisture content in winter and decreasing moisture in summer are the circulation background causing the seasonal shifts of precipitation changes, dryness/wetness conditions and also the runoff variations. Seasonal shifts of runoff may pose new challenges for the water resources management in the Guangdong Province, one of the economically prosperous regions in China. Precipitation changes are the major factors exerting tremendous influences on runoff variations in both space and time, which is reflected mainly by significant correlation coefficients. Weaker correlation coefficient between annual minimum monthly runoff and precipitation may indicate other factors played a role. The water reservoir is another important factor, as in the East River, significant change points in annual minimum monthly runoff are detected after the construction of the Fengshuba Reservoir. Nevertheless, more comprehensive and systematical studies about the influences of water reservoirs on runoff in the whole South China should be carried out in the future. Hence, we can tentatively conclude that the runoff changes in the Guangdong Province are mainly controlled by precipitation changes. Acknowledgments This work is financially supported by the Key International Collaboration Project (Grant No.: 51320105010), the Leading Expert Program of Anhui Province, China, and is fully supported by a grant from the Research Grants Council of the Hong Kong Special Administrative Region, China (Project No. CUHK441313). Our cordial gratitude is also extended to the editor, Prof. Dr. Clemens Simmer, and also three anonymous reviewers for their professional and pertinent comments and suggestions which are greatly helpful for further improvement of the quality of this manuscript. Besides, we again owe our special thanks to the editor, Prof. Dr. Clemens Simmer, for his hard work and his great efforts in processing this manuscript. 287 References Arnell NW (1999) Climate change and global water resources. Global Environ Change 9:S31–S49 Birsan VM, Molnar P, Burlando P, Pfaundler M (2005) Streamflow trends in Switzerland. J Hydrol 314:312–329 Brabets PT, Walvoord AM (2009) Trends in streamflow in the Yukon River Basin from 1944-2005 and the influence of Pacific Decadal Oscillation. J Hydrol 371:108–119 Burn HD, Elnur HAM (2002) Detection of hydrologic trends and variability. J Hydrol 255:107–122 George SS (2007) Streamflow in the Winnipeg River basin, Canada: trends, extremes and climate linkages. J Hydrol 332:396–411 IPCC (2007) In Climate Change 2007: the physical science basis. In: Solomon S, Qin D, Manning M, Chen Z, Marquis M, Averyt KB, Tignor M, Miller HL (eds) Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, p 996 Kendall MG (1975) Rank correlation methods. Griffin, London Labat D, Goddéris Y, Probst JL, Guyot JL (2004) Evidence for global runoff increase related to climate warming. Adv Water Resour 27:631–642 Lu XX (2004) Vulnerability of water discharge of large Chinese rivers to environmental changes: an overview. Reg Environ Change 4:182–191 Mann HB (1945) Nonparametric tests against trend. Econometrica 13:245–259 Mann HB, Whitney DR (1947) On a test for whether one of two random variables is stochastically larger than the other. Ann Math Stat 18(1):50–60 Mitchell JM, Dzerdzeevskii B, Flohn H, Hofmeyr WL, Lamb HH, Rao KN, Wallén CC (1966) Climate change, WMO Technical Note No. 79. World Meteorological Organization Semenov V, Bengtsson L (2002) Secular trends in daily precipitation characteristics: greenhouse gas simulation with a coupled AOGCM. Clim Dyn 19:123–140 von Storch H, Navarra A (eds) (1995) Analysis of climate variability—applications of statistical techniques. Springer, New York Wang YQ, Zhou L (2005) Observed trends in extreme precipitation events in China during 1961-2001 and the associated changes in large-scale circulation. Geophys Res Lett 32:L09707. doi:10. 1029/2005GL022574 World Meteorological Organization (WMO) (2003) Statement on the status of global climate in 2003. WMO Publ. no. 966. WMO, Geneva Xu C-Y, Singh VP (2004) Review on regional water resources assessment under stationary and changing climate. Water Resour Manage 18(6):591–612 Yang Y, Tian F (2009) Abrupt change of runoff and its major driving factors in Haihe River Catchment, China. J Hydrol 374:373–383 Zhang X, Harvey K, Hogg WD, Yuzyk TR (2001) Trends in Canadian streamflow. Water Resour Res 37(4):987–998 Zhang Q, Xu C-Y, Becker S, Jiang T (2006) Sediment and streamflow changes in the Yangtze River basin during past 50 years. J Hydrol 331:511–523 Zhang SR, Lu XX, Higgitt DL, Chen CT, Han J, Sun H (2008) Recent changes of water discharge and sediment load in the Zhujiang (Pearl River) Basin, China. Global Planet Change 60:365–380 Zhang Q, Xu C-Y, Yang T (2009a) Variability of water resource of the Yellow River basin of past 50 years, China. Water Resour Manage 23:1157–1170 Zhang Q, Xu C-Y, Zhang Z (2009b) Observed changes of drought/ wetness episodes in the Pearl River basin, China, using the 123 288 standardized precipitation index and aridity index. Theor Appl Climatol 98:89–99 Zhang Q, Xu C-Y, Chen X, Zhang Z (2011) Statistical behaviors of precipitation regimes in China and their links with atmospheric circulation 1960-2005. Int J Climatol 31(11):1665–1678 123 Q. Zhang et al. Zhou Y, Zhang Q, Li K, Chen X (2012) Hydrological effects of water reservoirs on hydrological processes: complexity evaluations based on the multi-scale entropy analysis. Hydrol Process 26(21):3253–3262