- Email: [email protected]
Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China
Quaternary International xxx (2016) 1e12
Contents lists available at ScienceDirect
Quaternary International journal homepage: www.elsevier.com/locate/quaint
Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China Chao Tan a, b, c, Bensheng Huang a, b, c, Kunsong Liu a, c, Hui Chen a, c, Feng Liu a, c, d, *, Jing Qiu b, Jingxue Yang b a
School of Marine Sciences, Sun Yat-sen University, Guangzhou 510275, China Guangdong Research Institute of Water Resources and Hydropower, Guangdong Provincial Key Laboratory of Hydrodynamics, Guangzhou 510630, China State-province Joint Engineering Laboratory of Estuarine Hydraulic Technology, Guangzhou 510275, China d State Key Laboratory of Estuarine and Coastal Research, East China Normal University, Shanghai 200062, China b c
a r t i c l e i n f o
a b s t r a c t
Article history: Available online xxx
In this study, we updated the hydrological data from the 1950s to 2012 and analysed the temporal variations in the hydrological series of the Pearl River using the wavelet transform method. Furthermore, we quantified the climatic and anthropogenic effects on the changes in the hydrological processes in the Pearl River. The results of the combined methods of the wavelet transform method and the MannKendall (MK) trend test (i.e., wavelet trend test) reveal that the water discharge series exhibited a statistically insignificant changing trend since the 1950s; however, the sediment load series displayed a statistically significant decreasing trend. Comparisons of the results of the MK trend test and the wavelet trend test indicate that annual and inter-annual periodic oscillations affect the trend changes in the hydrological series. Statistical analysis and the double mass curve indicate that climatic change domi~ o Southern Oscillation (ENSO), and human activities nates changes in water discharge, such as the El Nin were mainly responsible for the phase changes in the sediment load. From 1973 to 1986, deforestation in the basin dominated the sediment variability and caused 83.2% of the increase in the sediment load compared to the reference period of 1957e1972. Even though water and soil conservation projects have been carried out since the early 1990s, the vegetation cover in the catchment area decreased by 6.5 104 km2 from the late 1980s to the late 2000s, as detected by Landsat TM images, and water and soil conservation projects had little effect on the sediment reduction. Since the 1990s, dam construction has dominated sediment variability. Compared to the reference period, the sediment load due to dam construction decreased by 83.4 kg/s and 993.3 kg/s in the periods of 1987e1998 and 1998e2006, respectively, and the reduction in the sediment load increased to 1329.5 kg/s in the period of 2007e2012, indicating the intensifying impact of dam construction on the sediment reduction. The sediment sources in the Pearl River basin have changed, and scouring of the river channel has become a new sediment source in response to dam construction. The alteration of hydrological processes in the Pearl River will continue to occur in the future in the context of global climatic change, which is becoming increasingly important for river management in the Pearl River basin. © 2016 Elsevier Ltd and INQUA. All rights reserved.
Keywords: Hydrological variations Pearl River Climate change Human activities Wavelet trend test
1. Introduction As an important link between the land and ocean, river discharge terrestrial materials, such as freshwater, sediment and nutrients, into the ocean greatly affect the coastal environment.
* Corresponding author. School of Marine Sciences, Sun Yat-sen University, Guangzhou 510275, China. E-mail address: [email protected] (F. Liu).
The hydrological regime of rivers has been affected by climate change and human activities in the context of global climate change (Milliman et al., 2008). In recent decades, human activities (such as dam construction, land use change, and soil and water conservation practices) have become a major driving force resulting in variations in the hydrological regime (Syvitski et al., 2005). There have been multiple reports of reduced sediment load from rivers due to human activities and climate change, which has resulted in catastrophic morphological changes in
http://dx.doi.org/10.1016/j.quaint.2016.02.043 1040-6182/© 2016 Elsevier Ltd and INQUA. All rights reserved.
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
2
C. Tan et al. / Quaternary International xxx (2016) 1e12
river deltas, such as the Nile (Fanos, 1995), Colorado (Carriquiry and Sanchez, 1999), Mississippi (Blum and Roberts, 2009), Ebro (Mikhailova, 2003), Yellow (Wang et al., 2007, 2010; Liu et al., 2012a,b) and Yangtze rivers (Yang et al., 2006; Liu et al., 2014b). Consequently, the changes in the hydrological processes in global rivers have attracted attention worldwide (e.g., Walling and Fang, 2003; Siakeu et al., 2004; Walling, 2006; Panda et al., 2011). The Pearl River, which flows into the South China Sea (SCS), is China's second-largest river in terms of water discharge and the nation's third-largest river in terms of sediment load. The history of human activities in the Pearl River basin, which is recorded in Chinese documents, can be traced back 2000 years (Qin Ministry) (Luo et al., 2002). Since the implementation of China's open-door and reform policies in the late 1970s, human activities in the Pearl River basin have intensified (Liu et al., 2014a). These human activities have disturbed hydrological processes in the Pearl River, and the Pearl River has become one of the world's most regulated rivers (Nilsson et al., 2005). In recent years, several studies have identified temporal variations in the hydrological series of the Pearl River and have discussed the anthropogenic and climatic impacts on these changes (e.g., Dai et al., 2008; Zhang et al., 2008; Wu et al., 2012; Liu et al., 2014a). Many of these studies concentrated on periods before the mid-2000s and focused on the lower reaches of the Pearl River. For example, Zhang et al. (2008) reported that the sediment load series in the main Pearl River, the West River, and the North River showed insignificant decreasing trends between the 1950s and 2004. In the Pearl River basin, water and soil conservation projects have been implemented to stop soil erosion since the 1990s, and many large dams have been constructed in the Pearl River basin since the 2000s, such as the Longtan and Baise dams. However, relatively less quantitative data on the magnitude of dam construction and land use changes in the Pearl River basin have been available since the 2000s. Detection of hydrological data in longterm series is an extremely long-term exercise with continuously updated data and is of scientific and practical importance in water resource management (Kundzewicz, 2004; Zhang and Lu, 2009). Furthermore, although previous studies have investigated long-term trend changes in hydrological series (eg., Zhang et al., 2008; Wu et al., 2012), these studies have all neglected the effects of high-frequency components in the time series on the detection of trend changes, which might mask the true trend of the time series (Liu et al., 2014a). Wavelet transform is a powerful tool for multi-scale identification and can decompose and reconstruct a time series at different time scales (e.g., Brechet et al., 2007; Liu et al., 2011; Nourani and Andalib, 2015). In this study, we updated the hydrological data from the 1950s to 2012 and used the wavelet trend test to analyse temporal variations in the hydrological series of the Pearl River. Furthermore, we investigated the magnitude of dam construction and land use changes in the Pearl River basin to examine the anthropogenic impact on the sediment variability, especially since the 2000s. Therefore, our main objectives are: (1) to explore the long-term trend of hydrological series using the wavelet trend test method; (2) to examine anthropogenic and climatic effects on the hydrological processes in the Pearl River basin, especially for the period after the mid-2000s; and (3) to quantify the anthropogenic and climatic effects on the hydrological processes in different periods. Our study not only improves our understanding of the relationships between hydrological processes, climate change, and human activities, which provides scientific guidelines for water resource management, but also provides useful methods to examine temporal variations in hydrological series for future studies.
2. Study area As China's second-largest river in terms of water discharge, the Pearl River originates on the Yunnan Plateau and flows eastward through hill country and mountains to the SCS. The Pearl River drains the Yunnan, Guizhou, Guangxi, Guangdong, Hunan, and Jiangxi Provinces of China and the northern part of Vietnam, with a mainstream length of 2400 km and a catchment of 450,000 km2 (Fig. 1). The Pearl River is a compound river system and has three main tributaries, including the West River, the North River, and the East River. The West River is the largest tributary and stretches over a length of 2214 km, with a basin area of 351,500 km2. The average annual water discharge and sediment load at the Gaoyao station were 2174.3 108 m3/a and 64.5 106 t/a, representing 77% and 89% of the Pearl River's total water discharge and sediment load, respectively. The North River is 468 km long and drains an area of 38,400 km2. The North River's water discharge and sediment load at the Shijiao station were 415.7 108 m3/a and 5.5 106 t/a, respectively. The East River extends over 562 km and has a basin area of 25,300 km2. The East River's lowest water discharge and sediment load at the Buoluo station were 233.4 108 m3/a and 2.4 106 t/a, respectively. The Pearl River basin is a subtropical to tropical monsoon climate region that straddles the Tropic of Cancer. The annual mean precipitation ranges from 1200 mm/a to 2200 mm/a and gradually decreases from the eastern to the western side of the river basin. 3. Data and method 3.1. Data In this study, the hydrological data consist of the consecutive annual water discharge and sediment load at gauging stations on the Pearl River from the 1950s to 2012. These gauging stations include the Xiaolongtan, Tian'e, Qianjiang, Dahuajiangkou, Liuzhou, Nanning, Wuzhou, and Gaoyao stations in the West River; the Shijiao station on the North River; and the Boluo station on the East River. The hydrological data were collected from the Sediment Bulletins of China Rivers compiled by the Ministry of Water Resources of China (MWRC) and provided by the Water Bureau of Guangdong Province. In addition, annual precipitation data for the Pearl River basin, covering 1957 to 2012, were collected from the Information Centre of the China Meteorological Administration. The quality of the hydrological and meteorological data was strictly controlled by the authorities before their release. In addition, Landsat satellite images were used to detect land use changes in the Pearl River basin, including Thematic Mapper (TM) data from 1987 to 1989, 1997 to 1999, and 2007 to 2009, which were provided by GSCloud at the Computer Network Information Centre of the Chinese Academy of Sciences (http://www.gscloud.cn/). 3.2. Methods 3.2.1. Wavelet trend test method There are many methods used to detect long-term trends in time series, including non-parametric statistic tests (e.g., Zhang and Lu, 2009), bootstrap methods (e.g., Burn et al., 2010), and regression models (e.g., Liu et al., 2014a). However, high frequency components (i.e., annual and inter-annual time scales) of a time series can affect the identification of trends in the series (Partal and Küçük, 2006). In this study, the wavelet transform method was used to eliminate the impacts of high frequency components on the detection of trends. The wavelet transform method, which is a powerful tool for multi-scale identification, can decompose and reconstruct a time series at different scales. The low frequency
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
C. Tan et al. / Quaternary International xxx (2016) 1e12
components are obtained via a reconstruction of the low frequency coefficients, which are obtained by low-pass filtering the original data. The change trend of the low frequency components can represent the trend of the time series at relevant time scales. In this study, the Daubechies 3 (Db3) wavelet, which has a similar shape change to the hydrological series, was used to decompose and reconstruct the annual hydrological series following Mallat's algorithm. The Mann-Kendall (MK) trend test, which is widely used in the detection of trend changes in climatic and hydrological series, was applied to detect the trend changes in the hydrological series based on the low frequency components. The MK trend test is a nonparametric method for finding monotonic trends. The null hypothesis, H0, is that a sample of data, Xi (i ¼ 1, 2 … n), is independent and identically distributed with no trend. The null hypothesis, H0, is accepted if Z1a=2 Z Z1a=2 , where Z and P are the significance levels of the tested and computed standardised statistics, respectively (Zhang and Lu, 2009). In this study, a significance level of P ¼ 0.05 was applied. According to Liu et al. (2014a), periodic variations in the hydrological series in the Pearl River have occurred at significant annual (0.25 a, 0.5 a, and 1 a) and inter-annual (2 a to 8 a) time scales. In this study, the wavelet trend test was applied to detect the
3
coordinate system and using the vector diagram of the basin boundary for masking. Then, the vegetation cover data for the subbasins in the Pearl River basin, including the West River basin, North River basin, East River basin, and Pearl River delta, were obtained.
4. Results and discussion 4.1. Temporal variations in the hydrological processes 4.1.1. Wavelet trend test of the hydrological series Fig. 2 shows the temporal changes of the reconstructed water discharge series in the Pearl River and its tributaries since the 1950s. Although the reconstructed water discharge series in the Pearl River and its tributaries show a fluctuation change, regression analysis indicates that the reconstructed water discharge series display a statistically insignificant changing trend, as indicated by the results of the wavelet trend test (Table 1). Since the 1950s, the reconstructed water discharge series in the West River displayed an insignificant decreasing trend, which was similar to that of the Pearl River, and the reconstructed water discharge series in the North and East Rivers displayed a statistically insignificant changing trend (Table 1).
Table 1 The results of wavelet trend test and MK trend test on the hydrological series in the Pearl River. River
Period
Wavelet trend test Water discharge
Pearl River West River North River East River
1957e2012 1957e2012 1954e2012 1954e2012
MK trend test Sediment load
Water discharge
Sediment load
Z
p
Z
p
Z
p
Z
p
0.32 0.93 0.00 0.04
0.75 0.35 1.00 0.97
¡4.16 ¡4.02 ¡2.13 ¡7.93
0.00 0.00 0.03 0.00
0.87 0.91 0.13 0.04
0.38 0.36 0.90 0.97
¡3.74 ¡3.74 0.82 ¡4.85
0.00 0.00 0.41 0.00
Note: Hydrological series with significant trends at the 0.05 significance level are shown in bold.
long-term trend of the hydrological series in the Pearl River as follows. First, the hydrological series were decomposed and reconstructed using the Db3 wavelet transform at the scale of the third level, which eliminates the effects of periodic fluctuations less than 8 a on the temporal changes of the time series (Liu et al., 2011). Then, the MK trend test was employed to detect the trend changes of the hydrological processes based on the reconstructed hydrological series at the third level. 3.2.2. Detection of land use changes Landsat TM images with a spatial resolution of 30 m 30 m were applied to detect changes in the vegetation cover in the Pearl River basin. Due to the large scale of the Pearl River basin and the impact of clouds on the identification of the vegetation cover, Landsat TM images from just one year cannot cover the entire Pearl River basin. Therefore, Landsat TM data from 1987 to 1989, 1997 to 1999, and 2007 to 2009 were used to analyse the vegetation cover in the late 1980s, late 1990s, and late 2000s in the Pearl River basin. The procedures for land use detection were as follows: (1) one scene of a Landsat TM image was classified based on the support vector machine method by selecting the vegetative and nonvegetative sample areas; (2) noise, such as single pixels, was reduced in the classification images and pixels were clustered to obtain satisfactory classification results; and (3) image mosaic and image mask procedures were performed. The entire Pearl River basin was covered by 34 Landsat TM image scenes. After the images were classified, a classification image of the entire Pearl River basin was obtained using the mosaic method based on the geographic
The reconstructed sediment load series in the Pearl River displayed a significant decreasing trend since the 1950s (Fig. 2a and Table 1), with an increasing trend from 1957 to 1984 but a decreasing trend from 1985 to 2012 (Fig. 2a). The reconstructed sediment load series in the three tributaries to the Pearl River all displayed significant decreasing trends (Table 1). Of the three tributaries, the decreasing trend of the sediment load series in the East River was the most significant, followed by that in the West River (Fig. 2 and Table 1). In addition, the sediment load in the West and North Rivers exhibited an increasing trend from the 1950s to the early 1980s, but a decreasing trend after the early 1980s, while the East River displayed a gradual decreasing trend in the sediment load since the 1950s (Fig. 2bed). 4.1.2. MK trend test of the hydrological series The trend changes of the annual observed water discharge and sediment load series in the 1950s to 2012 were also investigated using the MK trend test (results shown in Table 1). The observed water discharge series in the Pearl River and its tributaries show an insignificant change trend since the 1950s, which is similar to the results indicated by the wavelet trend test. For the sediment load, the observed series shows a significant decreasing trend change in the main Pearl River, West River, and East River, which is the similar to the findings detected by the wavelet trend test. However, there are some differences between the results of the reconstructed and observed sediment load series in the Pearl River and its tributaries (Table 1). The observed sediment load series in the North River shows an insignificant decreasing trend, which is different from the
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
4
C. Tan et al. / Quaternary International xxx (2016) 1e12
Fig. 1. The sketch map of the Pearl River basin. The definition abbreviations of dams in the Pearl River basin can be found in Table 2.
Fig. 2. Temporal changes in the reconstructed water discharge and sediment load in the Pearl River and its tributaries.
results based on the reconstructed series. In addition, the negative Z values of the MK test of the observed sediment load series were greater than those of the reconstructed series (Table 1), indicating that the reconstructed series displays a more significant decreasing trend than the observed series. These results reveal that the highfrequency components of the hydrological series can influence trend changes and that eliminating their impacts will be helpful to detect the actual trend changes, which will contribute to a better identification of anthropogenic impacts on hydrological processes. 4.2. Climatic impacts on hydrological processes Climate change is primarily characterised by changing precipitation and temperature. Precipitation is one of the most important factors in the hydrological process and is the main driver of variability in the water balance over space and time. Fig. 3 shows the temporal changes in the reconstructed mean precipitation, water discharge, and sediment load in the Pearl River basin at the third level timescales using the Db3 wavelet since the 1950s. The results of the MK trend test on the reconstructed precipitation in the Pearl River basin display an insignificant decreasing trend since the
1950s, which is similar to that of the water discharge in the Pearl River. In addition, the variations in the reconstructed water discharge correspond to the reconstructed precipitation in the Pearl River since the 1950s (Fig. 3a), and the statistical analysis results show that there is a significant correlation between the reconstructed precipitation and the water discharge (Fig. 3b). Fig. 3a also shows that the variations in the reconstructed sediment load coincide with the reconstructed precipitation in the Pearl River basin in the period of 1957e1972 and that there is a significant correlation between precipitation and sediment load during that period (Fig. 3c). The mean precipitation in the Pearl River decreased from 1973 to 1986 and then increased from 1987 to 1998, the sediment load shows increasing and decreasing trends in the two respective periods (Fig. 3a). Furthermore, Fig. 3c reveals negative relationships between the mean annual precipitation and the sediment load in the two periods from 1973 to 1986 and 1987 to 1998, demonstrating the impacts of human activities on the sediment load. In the 2000s, both the mean precipitation and the sediment load in the Pearl River basin decreased, and there is a significant relationship between the precipitation and sediment load, which indicates that the decrease in precipitation contributed
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
C. Tan et al. / Quaternary International xxx (2016) 1e12
5
Fig. 3. Temporal changes in the reconstructed water discharge, sediment load and precipitation at the third level (a); the correlation relationship between the reconstructed water discharge and precipitation (b) and between the reconstructed sediment load and precipitation (c) in the Pearl River.
to the reduction in the sediment load (Fig. 3c). These results reveal that precipitation in the Pearl River was a primary factor affecting changes in the water discharge but was not the main cause of the significant decreasing trend in the sediment load in the Pearl River. ~ o Southern As the strongest inter-annual signal, the El Nin Oscillation (ENSO), a coupled oceanic-atmospheric phenomenon occurring in the tropical Pacific Ocean, is the largest source of climatic and hydrological variability on the global scale (Shrestha and Kostaschuk, 2005; Ward et al., 2010). In China, there is evidence for the impact of ENSO events on hydrological variables. For example, Wang et al. (2006) reported that drought might easily occur during ~ o in the Yellow River basin, which would cause a decrease in El Nin riverine discharge to the sea. Zhang et al. (2007) found that the inphase relationship between the annual maximum stream flow and ENSO occurred in the lower Yangtze River basin. However, the impacts of ENSO on the hydrological processes in the Pearl River are relatively poorly reported. Fig. 4 shows the relationship between ENSO events and hydrological processes in the Pearl River. ENSO events identified from monthly Sea Surface Temperature (SST) anomaly data (available at ftp://www.coaps.fsu.edu/pub/JMA_SST_ Index/) coincide with the relatively low annual precipitation and water discharge during 1957e2012. For example, the ENSO event in 1963 coincided with the lowest precipitation and water discharge in the last 56 years in the Pearl River basin. In general, the mean annual water discharge in the ENSO event years were 7.4% lower than in the non-ENSO event years from 1957 to 2012, and differences in the mean water discharge between ENSO event years and non-ENSO event years was even more evident before the 1990s, with a value of 11.6%. Even though there is no clear relationship between the sediment load and ENSO events due to strong anthropogenic disturbances, the reduced precipitation and water discharge resulting from ENSO events can lead to sediment reduction in the Pearl River.
4.3. Anthropogenic impacts on hydrological processes 4.3.1. Dam construction Of the human activities influencing the hydrological regime, dam construction is the most direct way for humans to manipulate water resources and has become the dominate factor causing sediment reduction in rivers worldwide (Syvitski et al., 2005). For example, the Nile River sediment discharge reduced from 100 Mt/a to almost zero after the completion of the Aswan Dam (Walling and Fang, 2003), and the sediment load in the Colorado River decreased from approximately 125 Mt/a to 3 Mt/a due to the construction of the Hoover dam (Meade and Parker, 1985). Since the 1950s, multiple dams were constructed in the Pearl River basin for different primary purposes, such as irrigation, power generation, and flood control (Table 2). The cumulative storage capacity of the reservoirs in the Pearl River basin was approximately 208.2 108 m3 in the later 1960s. In the Pearl River basin, many large reservoirs were constructed in the 1990s, and the cumulative storage capacity significantly increased to 498.6 108 m3 in 1999 (Fig. 5a), which represented 18.7% of the annual water discharge of the Pearl River. After the construction of additional dams, including the Baise and Longtan dams, the total storage capacity reached 893.1 108 m3 in 2006 (Fig. 5a), which amounted to 44.8% of the mean annual water discharge of the Pearl River. In addition, the cumulative storage capacities of the West River, North River, and East River reached 674.5 108 m3, 48.1 108 m3, and 170.6 108 m3 in the later 2000s (Fig. 5b), which were 32.8%, 12.2%, and 74.6% of the mean annual water discharge values of those three tributaries, respectively. However, the annual water discharge series in the Pearl River and its three tributaries show an insignificant changing trend, indicating that dam construction exerted little influence on the changes in water discharge in the Pearl River.
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
6
C. Tan et al. / Quaternary International xxx (2016) 1e12
Table 2 Large dams and reservoirs in the Pearl River basin. River basin
Reservoir
Height (m)
Total storage capacity (108 m3)
Time of completion
West River
Xijing (XD) Chengbihe (CD) Dahua (DD) Lubuge (LBD) Yantan (YD) Bailongtan (BLD) Tianshengqiao (TD) Baise (BD) Longtan (LD) Nanshui (ND) Changhu (CHD) Jinjiang (JD) Feilaixia (FD) Xinfengjiang (XFD) Fengshuba (FSD) Baipenzhu (BPD)
51 70 75 104 110 28 180
30 11.3 9.64 1.11 33.8 2.4 102.6
1964 1966 1982 1988 1992 1996 1997
130 192 80 66 63 52 124 92 66
56 273 12.43 1.49 1.9 19.5 138.9 19.3 12.2
2006 2006 1971 1973 1990 1999 1960 1973 1985
North River
East River
Even though the suspended sediment concentrations in the Pearl River are lower than those in the Yellow River and in the Yangtze River, the sediment trapped by reservoirs in the basin is substantial. The West River contributed more than 89% of the sediment load in the Pearl River, and therefore, the dams constructed in the West River basin have had a large impact on changes
Fig. 5. Temporal changes in the cumulative storage capacities and the ration of water storage capacities and water discharge (a) of the large reservoirs in the Pearl River and its three tributaries (b).
Fig. 4. Temporal changes in the ENSO indicator (i.e., monthly Sea Surface Temperature anomaly data, available at ftp://www.coaps.fsu.edu/pub/JMA_SST_Index/) (a) and annual precipitation, water discharge and sediment load (b) in the Pear River during the 1957e2012.
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
C. Tan et al. / Quaternary International xxx (2016) 1e12
in the sediment load in the Pearl River. Fig. 6 shows the temporal changes in the sediment load in the main stream of the West River and its tributaries since the 1950s. Although there were multiple dams constructed in the West River basin prior to the 1980s (Fig. 5b and Table 2), an increase in sediment load can be detected in the West River basin. Significant decreases in the sediment load in the West River basin began occurring in 1986 in response to the reduction in precipitation and the construction of the Lubuge dam, with an depositional rate of 2.26 Mt/a (Fig. 6ae6c), which started to trap water in 1985 and to generate electricity in 1988 in the Nanpanjiang River basin, which is an important sediment source for the upstream regions of the West River. In 1992, the Yantan dam was constructed along the Hongshuihe River, and the annual deposition amount reached 35 Mt/a. The annual sediment load at the Qianjiang station, located just below the Yantan dam, decreased by 27.6 Mt/a (58.7%) in the period of 1992e2002 compared to 1989e1991 (Fig. 6b). After the construction of the Tiaoshengqiao dam along the main stream in 1997, the annual sediment at the Tian'e and Qianjiang stations decreased by 28.0 Mt/a (50.1%) and 24.0 Mt/a (72.9%) in the period of 1998e2006 compared to 1989e1997 (Fig. 6a and b), respectively. In 2006, the Baise and Longtan dams were constructed along the Youjiang River, a tributary of the West River, and the main stream of the West River, respectively. Correspondingly, the annual sediment load at the Nanning station, below the Baise dam on the Youjiang River, decreased to 31.9% of 1950se2006 values in the period of 2007e2011 (Fig. 6d); the annual sediment load at the Tian'e and Qianjiang stations decreased by 94.8% and 88.4% after 2006 compared to the values in 1998e2006 (Fig. 6a and b), respectively. For the entire Pearl River basin, according to Dai et al. (2008), the ratios of sediment deposition in sample reservoirs to the reservoirs' storage capacities in the West River, North River, and East River were 1.3%, 0.087%, and 0.046%, respectively. Therefore, the annual deposition amounts in the three river basins were estimated to be 876 106 m3, 4.2 106 m3, and 7.8 106 m3, respectively. Assuming the sediment density to be 1.3 t/m3 (Chu et al., 2009), the annual deposition amounts behind the reservoirs in the three river basins were estimated to be 11.4 108 t/a, 5.5 106 t/a, and 10.1 106 t/a, respectively. Therefore, the total annual deposition in the reservoirs in the Pearl River basin was approximately 11.6 108 t/a during the 2000s, which is greater than the annual sediment load from the Pearl River (36.7 Mt/a during the 2000s). Fig. 7 shows temporal changes in sediment load and sediment budget along the mainstream of the West River. The annual sediment load increased gradually downwards towards the lower reaches of the West River before 1992, reflecting the sediment input from the tributaries to the West River (Fig. 7a). After the construction of the Yantan dam in 1992, the sediment load decreased by 58.7% from the Tian'e station to the Qianjiang station on the Hongshuihe River from 1993 to 1998. Moreover, from the Dahuangjiangkou station to the Wuzhou station, the sediment load decreased by 8.5% due to the construction of the Jingnan dam on a
7
tributary of the Guijiang River. However, the sediment load increased from the Xiaolongtan dam to the Tian'e station, from the Qianjiang station to the Dahuangjiangkou station, and from the Wuzhou station to the Gaoyao station (Fig. 7a). From 1999 to 2006, the sediment load downstream of the Tian'e station decreased compared to previous periods after the completion of the Tianshengqiao dam. The sediment load decreased by 67.5% from the Tian'e station to the Qianjiang station; however, the sediment load only gradually decreased downstream of the Qianjiang station. After the construction of the Longtan dam in 2006, the sediment load gradually decreased from the Xiaolongtan station to the Qianjiang station and then increased downstream of the Qiangjiang station from 2007 to 2012; the sediment load along the mainstream decreased compared to the previous time periods. Furthermore, the sediment budget between the Wuzhou station and the Gaoyao station increased from 0.89 Mt/a before 1992 to 11.7 Mt/a after 1992 (Fig. 7b), which may be due to sediment supplied by erosion from the main channel. Therefore, the sediment sources of the West River basin have changed in response to dam construction. 4.3.2. Land use changes With the explosive growth of the Pearl River basin's population, large areas have been deforested since the founding of the People's Republic of China. Previous studies indicated that the area of eroded soil in the Pearl River basin increased from 41,100 km2 in 1954 to 57,073 km2 in 1988, an increase of 38.9% (Liu et al., 2014a). The influence of deforestation exceeded that of dam construction during the 1973e1986, which increased the sediment load in the Pearl River (Fig. 3a). Based on Landsat TM images, land use changes were detected in the late 1980s, the 1990s, and the 2010s (Fig. 8). The vegetation-covered area in the West River, North River, and East River basins, which are the major sources of sediment yield in the Pearl River, was 34.0 104 km2 in the late 1980s and accounted for 77.8% of the total area of the three tributaries (Table 3). Since the early 1990s, water and soil conservation projects have been conducted in the Pearl River basin, and the vegetation-covered area in the North River basin increased by 5.7% from the late 1980s to the late 1990s; however, the vegetation-covered area in the West and East River basins decreased by 6.2% and 13.3% from the late 1980s to the late 1990s, respectively. In addition, the vegetation-covered area in the three sub-basins decreased by 14.0%, 8.1%, and 7.7% in the West River basin, North River basin, and East River basin from the late 1990s to the late 2000s, respectively (Table 3). Therefore, the water and soil conservation projects have had little impact on the decrease of sediment load; in contrast, there is evidence that the increase in deforestation from the late 1990s to the late 2000s resulted in an increase in the sediment load. For example, the reconstructed sediment load in the East River increased from 1996 to 2005; and even though multiple dams were constructed in the Liujiang River basin, a tributary of the West River, the sediment load increased from 4.36 Mt/a in the 1950se1980s to 5.93 Mt/a in the 1990se2000s (Fig. 6d).
Table 3 Change of vegetation cover area in the Pearl River basin and its tributaries. Drainage area
Total area (104 km2)
Vegetation cover Late 1980s
West River North River East River Pearl River delta Pearl River basin
35.3 4.8 3.6 3.2 46.8
Late 1990s
Late 2000s
Area (104 km2)
Percentage* (%)
Area (104 km2)
Percentage* (%)
Area (104 km2)
Percentage* (%)
27.5 3.5 3.0 1.3 35.3
77.9 73.6 83.6 41.1 75.4
25.8 3.7 2.6 1.0 33.1
73.1 77.7 72.8 31.6 70.7
22.2 3.4 2.4 0.8 28.8
62.9 71.4 67.2 25.3 61.5
Note: *Indicates the percentage of vegetation cover area accounting for the total area in the sub-basin.
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
8
C. Tan et al. / Quaternary International xxx (2016) 1e12
Fig. 6. Temporal changes in the sediment load in the West River and its tributaries since the 1950s (a, b, c, d).
4.4. Quantifying climatic and anthropogenic impacts on the sediment load To identify the effects of human activities on the sediment load, the double mass curve of the cumulative precipitation and the cumulative sediment load is plotted in Fig. 9a. If there is no effect of human activities on the sediment load, then the double mass curve is expected to be a straight line, which represents the best-fit relationship between the sediment load and the precipitation. If there are breaks in the double mass curve, then the change in the sediment load is influenced by not
only precipitation but also by human activities. From Fig. 9a, turning points occurred at approximately 1973, 1987, 1999 and 2007. From 1957 to 1972, the double mass curve was a straight line. And the double mass curve rises slightly from 1973 to 1986, indicating that the effects of deforestation during this period exceeded the effects of dam construction, which resulted in an increase in the sediment load. The double mass curve starts to bend from 1987 to 1998, and further from 1999 to 2012, revealing that the effects of dam construction begin to outweigh the effects of deforestation during these periods.
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
C. Tan et al. / Quaternary International xxx (2016) 1e12
9
Fig. 7. Temporal changes in mean annual sediment load and sediment budget at major gauging stations along the mainstream of the West River. Sediment budget between the two neighboring stations here is calculated by subtracting sediment load at the upstream station from that at the downstream station.
Based on the double mass curve of the cumulative precipitation and the cumulative sediment load, linear regression equations were applied to assess the effects of human activities and climate change on the sediment load. From Fig. 8a, the period from 1957 to 1972 is chosen as a reference period, as it was relatively less affected by human activities compared to the period afterward. An equation relating the annual precipitation and the annual sediment load in the Pearl River was established as follows (Fig. 9b):
y ¼ 3:477x 2730:69 R2 ¼ 0:61; p < 0:001 : Table 4 lists the decadal mean annual measured and predicted sediment loads for the different periods. The predicted sediment loads are close to the measured values in the reference period, giving confidence to the estimated values in the subsequent
periods. The differences between the measured and predicted values of the sediment load based on the precipitation reflect the effect of human activities. The differences between the measured sediment loads in different periods and the reference period reflect the total changes caused by both precipitation changes and human activities. The contributions of climate change and of human activities to changes in the sediment load during different periods are summarised in Table 4. From 1973 to 1986, the sediment load in the Pearl River increased by 466.5 kg/s, and human activities and precipitation changes contributed 83.2% and 16.8%, respectively, to the change in the sediment load. While human activities resulted in a decrease of 83.4 kg/s from 1987 to 1998, precipitation changes caused the sediment load to increase by 98.8 kg/s; therefore, the sediment load increased by 15.4 kg/s from 1987 to 1998. Since 1999, both human activities and precipitation changes have caused the
Table 4 Changes of precipitation, measured and estimated sediment and the impacts of human activities and precipitation on the sediment load in different period. Period
Reference period 1973e1986 1987e1998 1999e2006 2007e2012
Precipitation (mm)
1374.5 1397.0 1402.9 1338.7 1270.3
Sediment load (kg/s)
Change* (kg/s)
Measured value
Estimated value
By human activities
By precipitation variation
By both factors
2408.5 2875.0 2424.0 1290.6 716.5
2408.5 2486.8 2507.3 2283.8 2046.0
e
e
e
388.2 83.4 993.3 1329.5
(83.2%) (-540%) (88.8%) (78.6%)
78.3 98.8 124.7 362.5
(16.8%) (640%) (11.2%) (21.4%)
466.5 15.4 1118.0 1692.0
Note: *Positive or negative value indicates increase or decrease of sediment load.
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
10
C. Tan et al. / Quaternary International xxx (2016) 1e12
Fig. 9. Plot of cumulative sediment load versus cumulative precipitation (a) and the regression correlation between the sediment load and precipitation (b) in the Pearl River.
sediment load to decrease. From 1999 to 2006, human activities (i.e., dam construction) and precipitation reduction caused the sediment load to decrease, with values of 993.3 kg/s and 124.7 kg/s, which contributed 88.8% and 11.2%, respectively, to the total decrease in the sediment load. After the construction of the Longtan and Baise dams, human activities caused an even larger decrease in the sediment load with a value of 1329.5 kg/s from 2007 to 2012, which contributed to 78.6% of the total sediment reduction. In addition, reduced precipitation caused a decrease in the sediment load of 362.5 kg/s, which accounted for 21.4% of the total sediment reduction. 4.5. Comparisons with previous studies and implications
Fig. 8. The changes in the vegetation cover in the Pearl River basin in the late 1980s (a), late 1990s (b) and late 2000s (c).
Several studies have examined the changes in water discharge and sediment load in the Pearl River and their causes (e.g., Dai et al., 2008; Zhang et al., 2008; Wu et al., 2012; Liu et al., 2014a). Some of these studies revealed that the water discharge series displayed an insignificant trend, but the sediment load series displayed a significant decreasing trend from the 1950s to the mid-2000s using the MK trend test method (e.g., Zhang et al., 2008; Wu et al., 2012). In our study, we applied the wavelet trend test and the MK trend test methods to detected trend changes in the hydrological series of the Pearl River from the 1950s to 2012. Our study showed a more significant decreasing trend in the sediment load series than had previous studies using the wavelet trend test. Our results demonstrate the impacts of climatic fluctuations on short time scales on the detection of the actual trend in the sediment load series and indicate an intensifying impact of human activities on the sediment variability after the construction of large dams, such as the Longtan
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
C. Tan et al. / Quaternary International xxx (2016) 1e12
and Baise dams in 2006. In addition, our study showed a more significant decreasing trend in the water discharge and precipitation series than had previous studies, and furthermore, the reduced sediment loads resulting from the precipitation variations increased from 1999 to 2006 to 2007 to 2012. In addition, our study revealed that ENSO events may reduce precipitation and water discharge in the Pearl River. Therefore, even though human activities have been a dominant factor affecting the sediment discharge in recent decades, we need to pay attention to the impacts of climate change on future hydrological processes in the context of global climate changes, including the ENSO event. In addition, the sediment sources for the Pearl River have changed in response to dam construction in the catchment, and the erosion of the river channel has become a new sediment source due to sediment loss. Changes in sediment sources will affect sediment characteristics and influence the evolution of river channels and river deltas. For example, due to sediment reduction, the erosion of the lower river channel of the Yellow River increased sediment flux and caused relatively coarser suspended sediments to be delivered into the estuary, which has affected the evolution mode of the Yellow River delta. Therefore, it is necessary to explore the impacts of changes in the sediment sources on the evolution of the river channel and river deltas of the Pearl River in future studies, as it is of scientific and practical importance for water resource management and coastal protection. 5. Conclusions In this study, trend changes in the hydrological regime in the Pearl River from the 1950s to 2012 were detected using the wavelet trend test method. Since the 1950s, the water discharge series displayed an insignificant change trend; however, the sediment load series displayed a statistically significant decreasing trend. Variations in the water discharge were affected by precipitation changes, while human activities exerted little influence. The phase and trend changes of the sediment load were caused by human activities. From 1973 to 1986, the impacts of deforestation exceeded those of dam construction and caused 83.2% of the increase in the sediment load compared to the reference period of 1957e1972. Since the 1990s, dam construction has dominated sediment variability, and the water and soil conservation projects, which have been carried out since the early 1990s, have exerted little influence on the sediment reduction. Compared to the reference period, the annual sediment load due to human activities decreased by 83.4 kg/ s and 993.3 kg/s in the periods of 1987e1998 and 1999e2006, respectively. After the construction of the Longtan and Baise Dams, the reduced annual sediment load resulting from human activities increased to 1329.5 kg/s in the period between 2007 and 2012, indicating an intensifying impact of dam construction on sediment reduction. In addition, dam construction has led to a change in the sediment sources in the Pearl River basin, and scouring of the river channel has become a new sediment source. Our study has identified temporal variations in the hydrological series in the Pearl River up to 2012 and has examined the impacts of human activities on the hydrological processes in the entire Pearl River, which will become increasingly important for the management of water resources in the Pearl River basin in the context of global climate change. Acknowledgements This research was financially supported by the Open Research Fund of State Key Laboratory of Estuarine and Coastal Research (No: SKLEC-KF201409), China Postdoctoral Science Foundation funded project (No: 2013M531890), National Natural Science Foundation
11
of China (No: 41476073), Guangdong Provincial Hydraulic Technology Innovation Project (No:2011-08; 2012-03) and Guangdong Provincial Hydraulic Technology Key Innovation Project (No: 201406). We thank the anonymous reviewers whose invaluable and constructive suggestions greatly improve the scientific quality of the original manuscript. References Blum, M.D., Roberts, H.H., 2009. Drowning of the Mississippi Delta due to insufficient sediment supply and global sea-level rise. Nature Geoscience 2, 488e491. Brechet, L., Lucas, M.F., Doncarli, C., Farina, D., 2007. Compression of biomedical signals with mother wavelet optimization and best-basis wavelet packet selection. Biomedical Engineering. IEEE Transactions On 54 (12), 2186e2192. Burn, D.H., Sharif, M., Zhang, K., 2010. Detection of trends in hydrological extremes for Canadian watersheds. Hydrological Processes 24 (13), 1781e1790. Carriquiry, J.D., Sanchez, A., 1999. Sedimentation in the Colorado River delta and upper Gulf of California after nearly a century of discharge loss. Marine Geology 158, 125e145. Chu, Z.X., Zhai, S.K., Lu, X.X., Liu, J.P., Xu, J.X., Xu, K.H., 2009. A quantitative assessment of human impacts on decrease in sediment flux from major Chinese rivers entering the western Pacific Ocean. Geophysical Research Letters 36. http://dx.doi.org/10.1029/2009GL039513. Dai, S.B., Yang, S.L., Cai, A.M., 2008. Impacts of dams on the sediment flux of the Pearl River, southern China. Catena 76, 36e43. Fanos, A.M., 1995. The impact of human activities on the erosion and accretion of the Nile delta coast. Journal of Coastal Research 11, 821e833. Kundzewicz, Z.W., 2004. Searching for change in hydrological data. Hydrological Sciences Journal 49, 3e6. Liu, F., Chen, S.L., Dong, P., Peng, J., 2012a. Spatial and temporal variability of water discharge in the Yellow River Basin over the past 60 years. Journal of Geographical Sciences 22 (6), 1013e1033. Liu, F., Chen, S.L., Peng, J., Chen, G.Q., 2012b. Temporal variations of water discharge and sediment load of Huanghe River, China. Chinese Geographical Science 22 (5), 507e521. Liu, F., Chen, S.L., Peng, J., Chen, G.Q., 2011. Temporal variability of water discharge and sediment load of the Yellow River into the sea during 19502008. Journal of Geographical Sciences 21 (6), 1047e1061. Liu, F., Yuan, L.R., Yang, Q.S., Ou, S.Y., Xie, L.L., Cui, X., 2014a. Hydrological responses to the combined influence of diverse human activities in the Pearl River delta, China. Catena 113, 41e55. Liu, F., Yang, Q.S., Chen, S.L., Luo, Z.F., Yuan, F., Wang, R.T., 2014b. Temporal and spatial variability of sediment flux into the sea from the three largest rivers in China. Journal of Asian Earth Sciences 87, 102e115. Luo, X.L., Yang, Q.S., Jia, L.W., et al., 2002. River-bed Evolution of the Pearl River Delta Network. Zhongshan (Sun Yat-sen) University Press, Guangzhou, China (in Chinese). Meade, R.H., Parker, R.S., 1985. Sediment in rivers of the United States. National water summary, 1984. US Geological Survey Water-Supply Paper 2275, 49e60. Mikhailova, M.V., 2003. Transformation of the Ebro River Delta under the impact of intense human-induced reduction of sediment and runoff. Water Resources 30 (4), 370e378. Milliman, J.D., Farnsworth, K.L., Jones, P.D., Xu, K.H., Smith, L.C., 2008. Climatic and anthropogenic factors affecting river discharge to the Global Ocean, 1951e2000. Global and Planetary Change 62, 187e194. Nilsson, C., Reidy, C.A., Dynesius, M., Revenga, C., 2005. Fragmentation and flow regulation of the world's large river systems. Science 308, 405e408. Nourani, Vahid, Andalib, G., 2015. Daily and monthly suspended sediment load predictions using wavelet based artificial intelligence approaches. Journal of Mountain Science 12 (1), 85e100. Panda, D.K., Kumar, A., Mohanty, S., 2011. Recent trends in sediment load of the tropical (Peninsular) river basins of India. Global and Planetary Change 75, 108e118. Partal, T., Küçük, M., 2006. Long-term trend analysis using discrete wavelet components of annual precipitations measurements in Marmara region (Turkey). Physics and Chemistry of the Earth Parts A/B/C 31 (18), 1189e1200. ~ o/Southern Oscillation (ENSO)-related Shrestha, A., Kostaschuk, R., 2005. El Nin variability in mean-monthly streamflow in Nepal. Journal of Hydrology 308, 33e49. Siakeu, J., Oguchi, T., Aoki, T., 2004. Change in riverine suspended sediment concentration in central Japan in response to late 20th century human activities. Catena 55 (2), 231e254. €ro €smarty, C.J., Kettner, A.J., Green, P., 2005. Impact of humans on Syvitski, J.P.M., Vo the flux of terrestrial sediment to the global coastal ocean. Science 308, 376e380. http://dx.doi.org/10.1126/science.1109454. Walling, D.E., 2006. Human impact on landeocean sediment transfer by the world's rivers. Geomorphology 79 (3e4), 192e216. Walling, D.E., Fang, D., 2003. Recent trends in the suspended sediment loads of the world's rivers. Global and Planetary Change 39, 111e126. http://dx.doi.org/ 10.1016/S0921-8181(03)00020-1. Wang, H.J., Bi, N., Saito, Y., Wang, Y., Sun, X., Zhang, J., 2010. Recent changes in sediment delivery by the Huanghe (Yellow River) to the Sea: causes and
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
12
C. Tan et al. / Quaternary International xxx (2016) 1e12
environmental implications in its Estuary. Journal of Hydrology 39 (3e4), 302e313. Wang, H.J., Yang, Z.S., Saito, Y., Liu, J.P., Sun, X., Wang, Y., 2007. Stepwise decreases of the Huanghe (Yellow River) sediment load (1950e2005): impacts of climate changes and human activities. Global and Planetary Change 57 (3e4), 331e354. Wang, H.J., Yang, Z.S., Saito, Y., Liu, J.P., Sun, X., 2006. Interannual and seasonal variation of the Huanghe (Yellow River) water discharge over the past 50 years: connections to impacts from ENSO events and dams. Global and Planetary Change 50, 212e225. Ward, P.J., Beets, W., Bouwer, L.M., Aerts, J.C.J.H., Renssen, H., 2010. Sensitivity of river discharge to ENSO. Geophysical Research Letters 37, L12402. http:// dx.doi.org/10.1029/2010GL043215. Wu, C.S., Yang, S.L., Lei, Y.P., 2012. Quantifying the anthropogenic and climatic impacts on water discharge and sediment load in the Pearl River (Zhujiang), China (1954e2009). Journal of Hydrology 190e204.
Yang, Z.S., Wang, H., Saito, Y., Milliman, J.D., Xu, K., Qiao, S., Shi, S., 2006. Dam impacts on the Changjiang (Yangtze) River sediment discharge to the sea: the past 55 years and after the three Gorges Dam. Water Resources Research 42, W04407. http://dx.doi.org/10.1029/2005WR003970. Zhang, Q., Xu, C.Y., Jiang, T., Wu, Y.J., 2007. Possible influence of ENSO on annual maximum streamflow of the Yangtze River, China. Journal of Hydrology 333 (2e4), 265e274. Zhang, S.R., Lu, X.X., 2009. Hydrological responses to precipitation variation and diverse human activities in a mountainous tributary of the lower Xijiang, China. Catena 77, 130e142. Zhang, S.R., Lu, X.X., Higgitt, D.L., Chen, C.T.A., Han, J.T., Sun, H.G., 2008. Recent changes of water discharge and sediment load in the Zhujiang (Pearl River) Basin, China. Global and Planetary Change 60, 365e380.
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
Contents lists available at ScienceDirect
Quaternary International journal homepage: www.elsevier.com/locate/quaint
Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China Chao Tan a, b, c, Bensheng Huang a, b, c, Kunsong Liu a, c, Hui Chen a, c, Feng Liu a, c, d, *, Jing Qiu b, Jingxue Yang b a
School of Marine Sciences, Sun Yat-sen University, Guangzhou 510275, China Guangdong Research Institute of Water Resources and Hydropower, Guangdong Provincial Key Laboratory of Hydrodynamics, Guangzhou 510630, China State-province Joint Engineering Laboratory of Estuarine Hydraulic Technology, Guangzhou 510275, China d State Key Laboratory of Estuarine and Coastal Research, East China Normal University, Shanghai 200062, China b c
a r t i c l e i n f o
a b s t r a c t
Article history: Available online xxx
In this study, we updated the hydrological data from the 1950s to 2012 and analysed the temporal variations in the hydrological series of the Pearl River using the wavelet transform method. Furthermore, we quantified the climatic and anthropogenic effects on the changes in the hydrological processes in the Pearl River. The results of the combined methods of the wavelet transform method and the MannKendall (MK) trend test (i.e., wavelet trend test) reveal that the water discharge series exhibited a statistically insignificant changing trend since the 1950s; however, the sediment load series displayed a statistically significant decreasing trend. Comparisons of the results of the MK trend test and the wavelet trend test indicate that annual and inter-annual periodic oscillations affect the trend changes in the hydrological series. Statistical analysis and the double mass curve indicate that climatic change domi~ o Southern Oscillation (ENSO), and human activities nates changes in water discharge, such as the El Nin were mainly responsible for the phase changes in the sediment load. From 1973 to 1986, deforestation in the basin dominated the sediment variability and caused 83.2% of the increase in the sediment load compared to the reference period of 1957e1972. Even though water and soil conservation projects have been carried out since the early 1990s, the vegetation cover in the catchment area decreased by 6.5 104 km2 from the late 1980s to the late 2000s, as detected by Landsat TM images, and water and soil conservation projects had little effect on the sediment reduction. Since the 1990s, dam construction has dominated sediment variability. Compared to the reference period, the sediment load due to dam construction decreased by 83.4 kg/s and 993.3 kg/s in the periods of 1987e1998 and 1998e2006, respectively, and the reduction in the sediment load increased to 1329.5 kg/s in the period of 2007e2012, indicating the intensifying impact of dam construction on the sediment reduction. The sediment sources in the Pearl River basin have changed, and scouring of the river channel has become a new sediment source in response to dam construction. The alteration of hydrological processes in the Pearl River will continue to occur in the future in the context of global climatic change, which is becoming increasingly important for river management in the Pearl River basin. © 2016 Elsevier Ltd and INQUA. All rights reserved.
Keywords: Hydrological variations Pearl River Climate change Human activities Wavelet trend test
1. Introduction As an important link between the land and ocean, river discharge terrestrial materials, such as freshwater, sediment and nutrients, into the ocean greatly affect the coastal environment.
* Corresponding author. School of Marine Sciences, Sun Yat-sen University, Guangzhou 510275, China. E-mail address: [email protected] (F. Liu).
The hydrological regime of rivers has been affected by climate change and human activities in the context of global climate change (Milliman et al., 2008). In recent decades, human activities (such as dam construction, land use change, and soil and water conservation practices) have become a major driving force resulting in variations in the hydrological regime (Syvitski et al., 2005). There have been multiple reports of reduced sediment load from rivers due to human activities and climate change, which has resulted in catastrophic morphological changes in
http://dx.doi.org/10.1016/j.quaint.2016.02.043 1040-6182/© 2016 Elsevier Ltd and INQUA. All rights reserved.
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
2
C. Tan et al. / Quaternary International xxx (2016) 1e12
river deltas, such as the Nile (Fanos, 1995), Colorado (Carriquiry and Sanchez, 1999), Mississippi (Blum and Roberts, 2009), Ebro (Mikhailova, 2003), Yellow (Wang et al., 2007, 2010; Liu et al., 2012a,b) and Yangtze rivers (Yang et al., 2006; Liu et al., 2014b). Consequently, the changes in the hydrological processes in global rivers have attracted attention worldwide (e.g., Walling and Fang, 2003; Siakeu et al., 2004; Walling, 2006; Panda et al., 2011). The Pearl River, which flows into the South China Sea (SCS), is China's second-largest river in terms of water discharge and the nation's third-largest river in terms of sediment load. The history of human activities in the Pearl River basin, which is recorded in Chinese documents, can be traced back 2000 years (Qin Ministry) (Luo et al., 2002). Since the implementation of China's open-door and reform policies in the late 1970s, human activities in the Pearl River basin have intensified (Liu et al., 2014a). These human activities have disturbed hydrological processes in the Pearl River, and the Pearl River has become one of the world's most regulated rivers (Nilsson et al., 2005). In recent years, several studies have identified temporal variations in the hydrological series of the Pearl River and have discussed the anthropogenic and climatic impacts on these changes (e.g., Dai et al., 2008; Zhang et al., 2008; Wu et al., 2012; Liu et al., 2014a). Many of these studies concentrated on periods before the mid-2000s and focused on the lower reaches of the Pearl River. For example, Zhang et al. (2008) reported that the sediment load series in the main Pearl River, the West River, and the North River showed insignificant decreasing trends between the 1950s and 2004. In the Pearl River basin, water and soil conservation projects have been implemented to stop soil erosion since the 1990s, and many large dams have been constructed in the Pearl River basin since the 2000s, such as the Longtan and Baise dams. However, relatively less quantitative data on the magnitude of dam construction and land use changes in the Pearl River basin have been available since the 2000s. Detection of hydrological data in longterm series is an extremely long-term exercise with continuously updated data and is of scientific and practical importance in water resource management (Kundzewicz, 2004; Zhang and Lu, 2009). Furthermore, although previous studies have investigated long-term trend changes in hydrological series (eg., Zhang et al., 2008; Wu et al., 2012), these studies have all neglected the effects of high-frequency components in the time series on the detection of trend changes, which might mask the true trend of the time series (Liu et al., 2014a). Wavelet transform is a powerful tool for multi-scale identification and can decompose and reconstruct a time series at different time scales (e.g., Brechet et al., 2007; Liu et al., 2011; Nourani and Andalib, 2015). In this study, we updated the hydrological data from the 1950s to 2012 and used the wavelet trend test to analyse temporal variations in the hydrological series of the Pearl River. Furthermore, we investigated the magnitude of dam construction and land use changes in the Pearl River basin to examine the anthropogenic impact on the sediment variability, especially since the 2000s. Therefore, our main objectives are: (1) to explore the long-term trend of hydrological series using the wavelet trend test method; (2) to examine anthropogenic and climatic effects on the hydrological processes in the Pearl River basin, especially for the period after the mid-2000s; and (3) to quantify the anthropogenic and climatic effects on the hydrological processes in different periods. Our study not only improves our understanding of the relationships between hydrological processes, climate change, and human activities, which provides scientific guidelines for water resource management, but also provides useful methods to examine temporal variations in hydrological series for future studies.
2. Study area As China's second-largest river in terms of water discharge, the Pearl River originates on the Yunnan Plateau and flows eastward through hill country and mountains to the SCS. The Pearl River drains the Yunnan, Guizhou, Guangxi, Guangdong, Hunan, and Jiangxi Provinces of China and the northern part of Vietnam, with a mainstream length of 2400 km and a catchment of 450,000 km2 (Fig. 1). The Pearl River is a compound river system and has three main tributaries, including the West River, the North River, and the East River. The West River is the largest tributary and stretches over a length of 2214 km, with a basin area of 351,500 km2. The average annual water discharge and sediment load at the Gaoyao station were 2174.3 108 m3/a and 64.5 106 t/a, representing 77% and 89% of the Pearl River's total water discharge and sediment load, respectively. The North River is 468 km long and drains an area of 38,400 km2. The North River's water discharge and sediment load at the Shijiao station were 415.7 108 m3/a and 5.5 106 t/a, respectively. The East River extends over 562 km and has a basin area of 25,300 km2. The East River's lowest water discharge and sediment load at the Buoluo station were 233.4 108 m3/a and 2.4 106 t/a, respectively. The Pearl River basin is a subtropical to tropical monsoon climate region that straddles the Tropic of Cancer. The annual mean precipitation ranges from 1200 mm/a to 2200 mm/a and gradually decreases from the eastern to the western side of the river basin. 3. Data and method 3.1. Data In this study, the hydrological data consist of the consecutive annual water discharge and sediment load at gauging stations on the Pearl River from the 1950s to 2012. These gauging stations include the Xiaolongtan, Tian'e, Qianjiang, Dahuajiangkou, Liuzhou, Nanning, Wuzhou, and Gaoyao stations in the West River; the Shijiao station on the North River; and the Boluo station on the East River. The hydrological data were collected from the Sediment Bulletins of China Rivers compiled by the Ministry of Water Resources of China (MWRC) and provided by the Water Bureau of Guangdong Province. In addition, annual precipitation data for the Pearl River basin, covering 1957 to 2012, were collected from the Information Centre of the China Meteorological Administration. The quality of the hydrological and meteorological data was strictly controlled by the authorities before their release. In addition, Landsat satellite images were used to detect land use changes in the Pearl River basin, including Thematic Mapper (TM) data from 1987 to 1989, 1997 to 1999, and 2007 to 2009, which were provided by GSCloud at the Computer Network Information Centre of the Chinese Academy of Sciences (http://www.gscloud.cn/). 3.2. Methods 3.2.1. Wavelet trend test method There are many methods used to detect long-term trends in time series, including non-parametric statistic tests (e.g., Zhang and Lu, 2009), bootstrap methods (e.g., Burn et al., 2010), and regression models (e.g., Liu et al., 2014a). However, high frequency components (i.e., annual and inter-annual time scales) of a time series can affect the identification of trends in the series (Partal and Küçük, 2006). In this study, the wavelet transform method was used to eliminate the impacts of high frequency components on the detection of trends. The wavelet transform method, which is a powerful tool for multi-scale identification, can decompose and reconstruct a time series at different scales. The low frequency
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
C. Tan et al. / Quaternary International xxx (2016) 1e12
components are obtained via a reconstruction of the low frequency coefficients, which are obtained by low-pass filtering the original data. The change trend of the low frequency components can represent the trend of the time series at relevant time scales. In this study, the Daubechies 3 (Db3) wavelet, which has a similar shape change to the hydrological series, was used to decompose and reconstruct the annual hydrological series following Mallat's algorithm. The Mann-Kendall (MK) trend test, which is widely used in the detection of trend changes in climatic and hydrological series, was applied to detect the trend changes in the hydrological series based on the low frequency components. The MK trend test is a nonparametric method for finding monotonic trends. The null hypothesis, H0, is that a sample of data, Xi (i ¼ 1, 2 … n), is independent and identically distributed with no trend. The null hypothesis, H0, is accepted if Z1a=2 Z Z1a=2 , where Z and P are the significance levels of the tested and computed standardised statistics, respectively (Zhang and Lu, 2009). In this study, a significance level of P ¼ 0.05 was applied. According to Liu et al. (2014a), periodic variations in the hydrological series in the Pearl River have occurred at significant annual (0.25 a, 0.5 a, and 1 a) and inter-annual (2 a to 8 a) time scales. In this study, the wavelet trend test was applied to detect the
3
coordinate system and using the vector diagram of the basin boundary for masking. Then, the vegetation cover data for the subbasins in the Pearl River basin, including the West River basin, North River basin, East River basin, and Pearl River delta, were obtained.
4. Results and discussion 4.1. Temporal variations in the hydrological processes 4.1.1. Wavelet trend test of the hydrological series Fig. 2 shows the temporal changes of the reconstructed water discharge series in the Pearl River and its tributaries since the 1950s. Although the reconstructed water discharge series in the Pearl River and its tributaries show a fluctuation change, regression analysis indicates that the reconstructed water discharge series display a statistically insignificant changing trend, as indicated by the results of the wavelet trend test (Table 1). Since the 1950s, the reconstructed water discharge series in the West River displayed an insignificant decreasing trend, which was similar to that of the Pearl River, and the reconstructed water discharge series in the North and East Rivers displayed a statistically insignificant changing trend (Table 1).
Table 1 The results of wavelet trend test and MK trend test on the hydrological series in the Pearl River. River
Period
Wavelet trend test Water discharge
Pearl River West River North River East River
1957e2012 1957e2012 1954e2012 1954e2012
MK trend test Sediment load
Water discharge
Sediment load
Z
p
Z
p
Z
p
Z
p
0.32 0.93 0.00 0.04
0.75 0.35 1.00 0.97
¡4.16 ¡4.02 ¡2.13 ¡7.93
0.00 0.00 0.03 0.00
0.87 0.91 0.13 0.04
0.38 0.36 0.90 0.97
¡3.74 ¡3.74 0.82 ¡4.85
0.00 0.00 0.41 0.00
Note: Hydrological series with significant trends at the 0.05 significance level are shown in bold.
long-term trend of the hydrological series in the Pearl River as follows. First, the hydrological series were decomposed and reconstructed using the Db3 wavelet transform at the scale of the third level, which eliminates the effects of periodic fluctuations less than 8 a on the temporal changes of the time series (Liu et al., 2011). Then, the MK trend test was employed to detect the trend changes of the hydrological processes based on the reconstructed hydrological series at the third level. 3.2.2. Detection of land use changes Landsat TM images with a spatial resolution of 30 m 30 m were applied to detect changes in the vegetation cover in the Pearl River basin. Due to the large scale of the Pearl River basin and the impact of clouds on the identification of the vegetation cover, Landsat TM images from just one year cannot cover the entire Pearl River basin. Therefore, Landsat TM data from 1987 to 1989, 1997 to 1999, and 2007 to 2009 were used to analyse the vegetation cover in the late 1980s, late 1990s, and late 2000s in the Pearl River basin. The procedures for land use detection were as follows: (1) one scene of a Landsat TM image was classified based on the support vector machine method by selecting the vegetative and nonvegetative sample areas; (2) noise, such as single pixels, was reduced in the classification images and pixels were clustered to obtain satisfactory classification results; and (3) image mosaic and image mask procedures were performed. The entire Pearl River basin was covered by 34 Landsat TM image scenes. After the images were classified, a classification image of the entire Pearl River basin was obtained using the mosaic method based on the geographic
The reconstructed sediment load series in the Pearl River displayed a significant decreasing trend since the 1950s (Fig. 2a and Table 1), with an increasing trend from 1957 to 1984 but a decreasing trend from 1985 to 2012 (Fig. 2a). The reconstructed sediment load series in the three tributaries to the Pearl River all displayed significant decreasing trends (Table 1). Of the three tributaries, the decreasing trend of the sediment load series in the East River was the most significant, followed by that in the West River (Fig. 2 and Table 1). In addition, the sediment load in the West and North Rivers exhibited an increasing trend from the 1950s to the early 1980s, but a decreasing trend after the early 1980s, while the East River displayed a gradual decreasing trend in the sediment load since the 1950s (Fig. 2bed). 4.1.2. MK trend test of the hydrological series The trend changes of the annual observed water discharge and sediment load series in the 1950s to 2012 were also investigated using the MK trend test (results shown in Table 1). The observed water discharge series in the Pearl River and its tributaries show an insignificant change trend since the 1950s, which is similar to the results indicated by the wavelet trend test. For the sediment load, the observed series shows a significant decreasing trend change in the main Pearl River, West River, and East River, which is the similar to the findings detected by the wavelet trend test. However, there are some differences between the results of the reconstructed and observed sediment load series in the Pearl River and its tributaries (Table 1). The observed sediment load series in the North River shows an insignificant decreasing trend, which is different from the
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
4
C. Tan et al. / Quaternary International xxx (2016) 1e12
Fig. 1. The sketch map of the Pearl River basin. The definition abbreviations of dams in the Pearl River basin can be found in Table 2.
Fig. 2. Temporal changes in the reconstructed water discharge and sediment load in the Pearl River and its tributaries.
results based on the reconstructed series. In addition, the negative Z values of the MK test of the observed sediment load series were greater than those of the reconstructed series (Table 1), indicating that the reconstructed series displays a more significant decreasing trend than the observed series. These results reveal that the highfrequency components of the hydrological series can influence trend changes and that eliminating their impacts will be helpful to detect the actual trend changes, which will contribute to a better identification of anthropogenic impacts on hydrological processes. 4.2. Climatic impacts on hydrological processes Climate change is primarily characterised by changing precipitation and temperature. Precipitation is one of the most important factors in the hydrological process and is the main driver of variability in the water balance over space and time. Fig. 3 shows the temporal changes in the reconstructed mean precipitation, water discharge, and sediment load in the Pearl River basin at the third level timescales using the Db3 wavelet since the 1950s. The results of the MK trend test on the reconstructed precipitation in the Pearl River basin display an insignificant decreasing trend since the
1950s, which is similar to that of the water discharge in the Pearl River. In addition, the variations in the reconstructed water discharge correspond to the reconstructed precipitation in the Pearl River since the 1950s (Fig. 3a), and the statistical analysis results show that there is a significant correlation between the reconstructed precipitation and the water discharge (Fig. 3b). Fig. 3a also shows that the variations in the reconstructed sediment load coincide with the reconstructed precipitation in the Pearl River basin in the period of 1957e1972 and that there is a significant correlation between precipitation and sediment load during that period (Fig. 3c). The mean precipitation in the Pearl River decreased from 1973 to 1986 and then increased from 1987 to 1998, the sediment load shows increasing and decreasing trends in the two respective periods (Fig. 3a). Furthermore, Fig. 3c reveals negative relationships between the mean annual precipitation and the sediment load in the two periods from 1973 to 1986 and 1987 to 1998, demonstrating the impacts of human activities on the sediment load. In the 2000s, both the mean precipitation and the sediment load in the Pearl River basin decreased, and there is a significant relationship between the precipitation and sediment load, which indicates that the decrease in precipitation contributed
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
C. Tan et al. / Quaternary International xxx (2016) 1e12
5
Fig. 3. Temporal changes in the reconstructed water discharge, sediment load and precipitation at the third level (a); the correlation relationship between the reconstructed water discharge and precipitation (b) and between the reconstructed sediment load and precipitation (c) in the Pearl River.
to the reduction in the sediment load (Fig. 3c). These results reveal that precipitation in the Pearl River was a primary factor affecting changes in the water discharge but was not the main cause of the significant decreasing trend in the sediment load in the Pearl River. ~ o Southern As the strongest inter-annual signal, the El Nin Oscillation (ENSO), a coupled oceanic-atmospheric phenomenon occurring in the tropical Pacific Ocean, is the largest source of climatic and hydrological variability on the global scale (Shrestha and Kostaschuk, 2005; Ward et al., 2010). In China, there is evidence for the impact of ENSO events on hydrological variables. For example, Wang et al. (2006) reported that drought might easily occur during ~ o in the Yellow River basin, which would cause a decrease in El Nin riverine discharge to the sea. Zhang et al. (2007) found that the inphase relationship between the annual maximum stream flow and ENSO occurred in the lower Yangtze River basin. However, the impacts of ENSO on the hydrological processes in the Pearl River are relatively poorly reported. Fig. 4 shows the relationship between ENSO events and hydrological processes in the Pearl River. ENSO events identified from monthly Sea Surface Temperature (SST) anomaly data (available at ftp://www.coaps.fsu.edu/pub/JMA_SST_ Index/) coincide with the relatively low annual precipitation and water discharge during 1957e2012. For example, the ENSO event in 1963 coincided with the lowest precipitation and water discharge in the last 56 years in the Pearl River basin. In general, the mean annual water discharge in the ENSO event years were 7.4% lower than in the non-ENSO event years from 1957 to 2012, and differences in the mean water discharge between ENSO event years and non-ENSO event years was even more evident before the 1990s, with a value of 11.6%. Even though there is no clear relationship between the sediment load and ENSO events due to strong anthropogenic disturbances, the reduced precipitation and water discharge resulting from ENSO events can lead to sediment reduction in the Pearl River.
4.3. Anthropogenic impacts on hydrological processes 4.3.1. Dam construction Of the human activities influencing the hydrological regime, dam construction is the most direct way for humans to manipulate water resources and has become the dominate factor causing sediment reduction in rivers worldwide (Syvitski et al., 2005). For example, the Nile River sediment discharge reduced from 100 Mt/a to almost zero after the completion of the Aswan Dam (Walling and Fang, 2003), and the sediment load in the Colorado River decreased from approximately 125 Mt/a to 3 Mt/a due to the construction of the Hoover dam (Meade and Parker, 1985). Since the 1950s, multiple dams were constructed in the Pearl River basin for different primary purposes, such as irrigation, power generation, and flood control (Table 2). The cumulative storage capacity of the reservoirs in the Pearl River basin was approximately 208.2 108 m3 in the later 1960s. In the Pearl River basin, many large reservoirs were constructed in the 1990s, and the cumulative storage capacity significantly increased to 498.6 108 m3 in 1999 (Fig. 5a), which represented 18.7% of the annual water discharge of the Pearl River. After the construction of additional dams, including the Baise and Longtan dams, the total storage capacity reached 893.1 108 m3 in 2006 (Fig. 5a), which amounted to 44.8% of the mean annual water discharge of the Pearl River. In addition, the cumulative storage capacities of the West River, North River, and East River reached 674.5 108 m3, 48.1 108 m3, and 170.6 108 m3 in the later 2000s (Fig. 5b), which were 32.8%, 12.2%, and 74.6% of the mean annual water discharge values of those three tributaries, respectively. However, the annual water discharge series in the Pearl River and its three tributaries show an insignificant changing trend, indicating that dam construction exerted little influence on the changes in water discharge in the Pearl River.
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
6
C. Tan et al. / Quaternary International xxx (2016) 1e12
Table 2 Large dams and reservoirs in the Pearl River basin. River basin
Reservoir
Height (m)
Total storage capacity (108 m3)
Time of completion
West River
Xijing (XD) Chengbihe (CD) Dahua (DD) Lubuge (LBD) Yantan (YD) Bailongtan (BLD) Tianshengqiao (TD) Baise (BD) Longtan (LD) Nanshui (ND) Changhu (CHD) Jinjiang (JD) Feilaixia (FD) Xinfengjiang (XFD) Fengshuba (FSD) Baipenzhu (BPD)
51 70 75 104 110 28 180
30 11.3 9.64 1.11 33.8 2.4 102.6
1964 1966 1982 1988 1992 1996 1997
130 192 80 66 63 52 124 92 66
56 273 12.43 1.49 1.9 19.5 138.9 19.3 12.2
2006 2006 1971 1973 1990 1999 1960 1973 1985
North River
East River
Even though the suspended sediment concentrations in the Pearl River are lower than those in the Yellow River and in the Yangtze River, the sediment trapped by reservoirs in the basin is substantial. The West River contributed more than 89% of the sediment load in the Pearl River, and therefore, the dams constructed in the West River basin have had a large impact on changes
Fig. 5. Temporal changes in the cumulative storage capacities and the ration of water storage capacities and water discharge (a) of the large reservoirs in the Pearl River and its three tributaries (b).
Fig. 4. Temporal changes in the ENSO indicator (i.e., monthly Sea Surface Temperature anomaly data, available at ftp://www.coaps.fsu.edu/pub/JMA_SST_Index/) (a) and annual precipitation, water discharge and sediment load (b) in the Pear River during the 1957e2012.
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
C. Tan et al. / Quaternary International xxx (2016) 1e12
in the sediment load in the Pearl River. Fig. 6 shows the temporal changes in the sediment load in the main stream of the West River and its tributaries since the 1950s. Although there were multiple dams constructed in the West River basin prior to the 1980s (Fig. 5b and Table 2), an increase in sediment load can be detected in the West River basin. Significant decreases in the sediment load in the West River basin began occurring in 1986 in response to the reduction in precipitation and the construction of the Lubuge dam, with an depositional rate of 2.26 Mt/a (Fig. 6ae6c), which started to trap water in 1985 and to generate electricity in 1988 in the Nanpanjiang River basin, which is an important sediment source for the upstream regions of the West River. In 1992, the Yantan dam was constructed along the Hongshuihe River, and the annual deposition amount reached 35 Mt/a. The annual sediment load at the Qianjiang station, located just below the Yantan dam, decreased by 27.6 Mt/a (58.7%) in the period of 1992e2002 compared to 1989e1991 (Fig. 6b). After the construction of the Tiaoshengqiao dam along the main stream in 1997, the annual sediment at the Tian'e and Qianjiang stations decreased by 28.0 Mt/a (50.1%) and 24.0 Mt/a (72.9%) in the period of 1998e2006 compared to 1989e1997 (Fig. 6a and b), respectively. In 2006, the Baise and Longtan dams were constructed along the Youjiang River, a tributary of the West River, and the main stream of the West River, respectively. Correspondingly, the annual sediment load at the Nanning station, below the Baise dam on the Youjiang River, decreased to 31.9% of 1950se2006 values in the period of 2007e2011 (Fig. 6d); the annual sediment load at the Tian'e and Qianjiang stations decreased by 94.8% and 88.4% after 2006 compared to the values in 1998e2006 (Fig. 6a and b), respectively. For the entire Pearl River basin, according to Dai et al. (2008), the ratios of sediment deposition in sample reservoirs to the reservoirs' storage capacities in the West River, North River, and East River were 1.3%, 0.087%, and 0.046%, respectively. Therefore, the annual deposition amounts in the three river basins were estimated to be 876 106 m3, 4.2 106 m3, and 7.8 106 m3, respectively. Assuming the sediment density to be 1.3 t/m3 (Chu et al., 2009), the annual deposition amounts behind the reservoirs in the three river basins were estimated to be 11.4 108 t/a, 5.5 106 t/a, and 10.1 106 t/a, respectively. Therefore, the total annual deposition in the reservoirs in the Pearl River basin was approximately 11.6 108 t/a during the 2000s, which is greater than the annual sediment load from the Pearl River (36.7 Mt/a during the 2000s). Fig. 7 shows temporal changes in sediment load and sediment budget along the mainstream of the West River. The annual sediment load increased gradually downwards towards the lower reaches of the West River before 1992, reflecting the sediment input from the tributaries to the West River (Fig. 7a). After the construction of the Yantan dam in 1992, the sediment load decreased by 58.7% from the Tian'e station to the Qianjiang station on the Hongshuihe River from 1993 to 1998. Moreover, from the Dahuangjiangkou station to the Wuzhou station, the sediment load decreased by 8.5% due to the construction of the Jingnan dam on a
7
tributary of the Guijiang River. However, the sediment load increased from the Xiaolongtan dam to the Tian'e station, from the Qianjiang station to the Dahuangjiangkou station, and from the Wuzhou station to the Gaoyao station (Fig. 7a). From 1999 to 2006, the sediment load downstream of the Tian'e station decreased compared to previous periods after the completion of the Tianshengqiao dam. The sediment load decreased by 67.5% from the Tian'e station to the Qianjiang station; however, the sediment load only gradually decreased downstream of the Qianjiang station. After the construction of the Longtan dam in 2006, the sediment load gradually decreased from the Xiaolongtan station to the Qianjiang station and then increased downstream of the Qiangjiang station from 2007 to 2012; the sediment load along the mainstream decreased compared to the previous time periods. Furthermore, the sediment budget between the Wuzhou station and the Gaoyao station increased from 0.89 Mt/a before 1992 to 11.7 Mt/a after 1992 (Fig. 7b), which may be due to sediment supplied by erosion from the main channel. Therefore, the sediment sources of the West River basin have changed in response to dam construction. 4.3.2. Land use changes With the explosive growth of the Pearl River basin's population, large areas have been deforested since the founding of the People's Republic of China. Previous studies indicated that the area of eroded soil in the Pearl River basin increased from 41,100 km2 in 1954 to 57,073 km2 in 1988, an increase of 38.9% (Liu et al., 2014a). The influence of deforestation exceeded that of dam construction during the 1973e1986, which increased the sediment load in the Pearl River (Fig. 3a). Based on Landsat TM images, land use changes were detected in the late 1980s, the 1990s, and the 2010s (Fig. 8). The vegetation-covered area in the West River, North River, and East River basins, which are the major sources of sediment yield in the Pearl River, was 34.0 104 km2 in the late 1980s and accounted for 77.8% of the total area of the three tributaries (Table 3). Since the early 1990s, water and soil conservation projects have been conducted in the Pearl River basin, and the vegetation-covered area in the North River basin increased by 5.7% from the late 1980s to the late 1990s; however, the vegetation-covered area in the West and East River basins decreased by 6.2% and 13.3% from the late 1980s to the late 1990s, respectively. In addition, the vegetation-covered area in the three sub-basins decreased by 14.0%, 8.1%, and 7.7% in the West River basin, North River basin, and East River basin from the late 1990s to the late 2000s, respectively (Table 3). Therefore, the water and soil conservation projects have had little impact on the decrease of sediment load; in contrast, there is evidence that the increase in deforestation from the late 1990s to the late 2000s resulted in an increase in the sediment load. For example, the reconstructed sediment load in the East River increased from 1996 to 2005; and even though multiple dams were constructed in the Liujiang River basin, a tributary of the West River, the sediment load increased from 4.36 Mt/a in the 1950se1980s to 5.93 Mt/a in the 1990se2000s (Fig. 6d).
Table 3 Change of vegetation cover area in the Pearl River basin and its tributaries. Drainage area
Total area (104 km2)
Vegetation cover Late 1980s
West River North River East River Pearl River delta Pearl River basin
35.3 4.8 3.6 3.2 46.8
Late 1990s
Late 2000s
Area (104 km2)
Percentage* (%)
Area (104 km2)
Percentage* (%)
Area (104 km2)
Percentage* (%)
27.5 3.5 3.0 1.3 35.3
77.9 73.6 83.6 41.1 75.4
25.8 3.7 2.6 1.0 33.1
73.1 77.7 72.8 31.6 70.7
22.2 3.4 2.4 0.8 28.8
62.9 71.4 67.2 25.3 61.5
Note: *Indicates the percentage of vegetation cover area accounting for the total area in the sub-basin.
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
8
C. Tan et al. / Quaternary International xxx (2016) 1e12
Fig. 6. Temporal changes in the sediment load in the West River and its tributaries since the 1950s (a, b, c, d).
4.4. Quantifying climatic and anthropogenic impacts on the sediment load To identify the effects of human activities on the sediment load, the double mass curve of the cumulative precipitation and the cumulative sediment load is plotted in Fig. 9a. If there is no effect of human activities on the sediment load, then the double mass curve is expected to be a straight line, which represents the best-fit relationship between the sediment load and the precipitation. If there are breaks in the double mass curve, then the change in the sediment load is influenced by not
only precipitation but also by human activities. From Fig. 9a, turning points occurred at approximately 1973, 1987, 1999 and 2007. From 1957 to 1972, the double mass curve was a straight line. And the double mass curve rises slightly from 1973 to 1986, indicating that the effects of deforestation during this period exceeded the effects of dam construction, which resulted in an increase in the sediment load. The double mass curve starts to bend from 1987 to 1998, and further from 1999 to 2012, revealing that the effects of dam construction begin to outweigh the effects of deforestation during these periods.
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
C. Tan et al. / Quaternary International xxx (2016) 1e12
9
Fig. 7. Temporal changes in mean annual sediment load and sediment budget at major gauging stations along the mainstream of the West River. Sediment budget between the two neighboring stations here is calculated by subtracting sediment load at the upstream station from that at the downstream station.
Based on the double mass curve of the cumulative precipitation and the cumulative sediment load, linear regression equations were applied to assess the effects of human activities and climate change on the sediment load. From Fig. 8a, the period from 1957 to 1972 is chosen as a reference period, as it was relatively less affected by human activities compared to the period afterward. An equation relating the annual precipitation and the annual sediment load in the Pearl River was established as follows (Fig. 9b):
y ¼ 3:477x 2730:69 R2 ¼ 0:61; p < 0:001 : Table 4 lists the decadal mean annual measured and predicted sediment loads for the different periods. The predicted sediment loads are close to the measured values in the reference period, giving confidence to the estimated values in the subsequent
periods. The differences between the measured and predicted values of the sediment load based on the precipitation reflect the effect of human activities. The differences between the measured sediment loads in different periods and the reference period reflect the total changes caused by both precipitation changes and human activities. The contributions of climate change and of human activities to changes in the sediment load during different periods are summarised in Table 4. From 1973 to 1986, the sediment load in the Pearl River increased by 466.5 kg/s, and human activities and precipitation changes contributed 83.2% and 16.8%, respectively, to the change in the sediment load. While human activities resulted in a decrease of 83.4 kg/s from 1987 to 1998, precipitation changes caused the sediment load to increase by 98.8 kg/s; therefore, the sediment load increased by 15.4 kg/s from 1987 to 1998. Since 1999, both human activities and precipitation changes have caused the
Table 4 Changes of precipitation, measured and estimated sediment and the impacts of human activities and precipitation on the sediment load in different period. Period
Reference period 1973e1986 1987e1998 1999e2006 2007e2012
Precipitation (mm)
1374.5 1397.0 1402.9 1338.7 1270.3
Sediment load (kg/s)
Change* (kg/s)
Measured value
Estimated value
By human activities
By precipitation variation
By both factors
2408.5 2875.0 2424.0 1290.6 716.5
2408.5 2486.8 2507.3 2283.8 2046.0
e
e
e
388.2 83.4 993.3 1329.5
(83.2%) (-540%) (88.8%) (78.6%)
78.3 98.8 124.7 362.5
(16.8%) (640%) (11.2%) (21.4%)
466.5 15.4 1118.0 1692.0
Note: *Positive or negative value indicates increase or decrease of sediment load.
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
10
C. Tan et al. / Quaternary International xxx (2016) 1e12
Fig. 9. Plot of cumulative sediment load versus cumulative precipitation (a) and the regression correlation between the sediment load and precipitation (b) in the Pearl River.
sediment load to decrease. From 1999 to 2006, human activities (i.e., dam construction) and precipitation reduction caused the sediment load to decrease, with values of 993.3 kg/s and 124.7 kg/s, which contributed 88.8% and 11.2%, respectively, to the total decrease in the sediment load. After the construction of the Longtan and Baise dams, human activities caused an even larger decrease in the sediment load with a value of 1329.5 kg/s from 2007 to 2012, which contributed to 78.6% of the total sediment reduction. In addition, reduced precipitation caused a decrease in the sediment load of 362.5 kg/s, which accounted for 21.4% of the total sediment reduction. 4.5. Comparisons with previous studies and implications
Fig. 8. The changes in the vegetation cover in the Pearl River basin in the late 1980s (a), late 1990s (b) and late 2000s (c).
Several studies have examined the changes in water discharge and sediment load in the Pearl River and their causes (e.g., Dai et al., 2008; Zhang et al., 2008; Wu et al., 2012; Liu et al., 2014a). Some of these studies revealed that the water discharge series displayed an insignificant trend, but the sediment load series displayed a significant decreasing trend from the 1950s to the mid-2000s using the MK trend test method (e.g., Zhang et al., 2008; Wu et al., 2012). In our study, we applied the wavelet trend test and the MK trend test methods to detected trend changes in the hydrological series of the Pearl River from the 1950s to 2012. Our study showed a more significant decreasing trend in the sediment load series than had previous studies using the wavelet trend test. Our results demonstrate the impacts of climatic fluctuations on short time scales on the detection of the actual trend in the sediment load series and indicate an intensifying impact of human activities on the sediment variability after the construction of large dams, such as the Longtan
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
C. Tan et al. / Quaternary International xxx (2016) 1e12
and Baise dams in 2006. In addition, our study showed a more significant decreasing trend in the water discharge and precipitation series than had previous studies, and furthermore, the reduced sediment loads resulting from the precipitation variations increased from 1999 to 2006 to 2007 to 2012. In addition, our study revealed that ENSO events may reduce precipitation and water discharge in the Pearl River. Therefore, even though human activities have been a dominant factor affecting the sediment discharge in recent decades, we need to pay attention to the impacts of climate change on future hydrological processes in the context of global climate changes, including the ENSO event. In addition, the sediment sources for the Pearl River have changed in response to dam construction in the catchment, and the erosion of the river channel has become a new sediment source due to sediment loss. Changes in sediment sources will affect sediment characteristics and influence the evolution of river channels and river deltas. For example, due to sediment reduction, the erosion of the lower river channel of the Yellow River increased sediment flux and caused relatively coarser suspended sediments to be delivered into the estuary, which has affected the evolution mode of the Yellow River delta. Therefore, it is necessary to explore the impacts of changes in the sediment sources on the evolution of the river channel and river deltas of the Pearl River in future studies, as it is of scientific and practical importance for water resource management and coastal protection. 5. Conclusions In this study, trend changes in the hydrological regime in the Pearl River from the 1950s to 2012 were detected using the wavelet trend test method. Since the 1950s, the water discharge series displayed an insignificant change trend; however, the sediment load series displayed a statistically significant decreasing trend. Variations in the water discharge were affected by precipitation changes, while human activities exerted little influence. The phase and trend changes of the sediment load were caused by human activities. From 1973 to 1986, the impacts of deforestation exceeded those of dam construction and caused 83.2% of the increase in the sediment load compared to the reference period of 1957e1972. Since the 1990s, dam construction has dominated sediment variability, and the water and soil conservation projects, which have been carried out since the early 1990s, have exerted little influence on the sediment reduction. Compared to the reference period, the annual sediment load due to human activities decreased by 83.4 kg/ s and 993.3 kg/s in the periods of 1987e1998 and 1999e2006, respectively. After the construction of the Longtan and Baise Dams, the reduced annual sediment load resulting from human activities increased to 1329.5 kg/s in the period between 2007 and 2012, indicating an intensifying impact of dam construction on sediment reduction. In addition, dam construction has led to a change in the sediment sources in the Pearl River basin, and scouring of the river channel has become a new sediment source. Our study has identified temporal variations in the hydrological series in the Pearl River up to 2012 and has examined the impacts of human activities on the hydrological processes in the entire Pearl River, which will become increasingly important for the management of water resources in the Pearl River basin in the context of global climate change. Acknowledgements This research was financially supported by the Open Research Fund of State Key Laboratory of Estuarine and Coastal Research (No: SKLEC-KF201409), China Postdoctoral Science Foundation funded project (No: 2013M531890), National Natural Science Foundation
11
of China (No: 41476073), Guangdong Provincial Hydraulic Technology Innovation Project (No:2011-08; 2012-03) and Guangdong Provincial Hydraulic Technology Key Innovation Project (No: 201406). We thank the anonymous reviewers whose invaluable and constructive suggestions greatly improve the scientific quality of the original manuscript. References Blum, M.D., Roberts, H.H., 2009. Drowning of the Mississippi Delta due to insufficient sediment supply and global sea-level rise. Nature Geoscience 2, 488e491. Brechet, L., Lucas, M.F., Doncarli, C., Farina, D., 2007. Compression of biomedical signals with mother wavelet optimization and best-basis wavelet packet selection. Biomedical Engineering. IEEE Transactions On 54 (12), 2186e2192. Burn, D.H., Sharif, M., Zhang, K., 2010. Detection of trends in hydrological extremes for Canadian watersheds. Hydrological Processes 24 (13), 1781e1790. Carriquiry, J.D., Sanchez, A., 1999. Sedimentation in the Colorado River delta and upper Gulf of California after nearly a century of discharge loss. Marine Geology 158, 125e145. Chu, Z.X., Zhai, S.K., Lu, X.X., Liu, J.P., Xu, J.X., Xu, K.H., 2009. A quantitative assessment of human impacts on decrease in sediment flux from major Chinese rivers entering the western Pacific Ocean. Geophysical Research Letters 36. http://dx.doi.org/10.1029/2009GL039513. Dai, S.B., Yang, S.L., Cai, A.M., 2008. Impacts of dams on the sediment flux of the Pearl River, southern China. Catena 76, 36e43. Fanos, A.M., 1995. The impact of human activities on the erosion and accretion of the Nile delta coast. Journal of Coastal Research 11, 821e833. Kundzewicz, Z.W., 2004. Searching for change in hydrological data. Hydrological Sciences Journal 49, 3e6. Liu, F., Chen, S.L., Dong, P., Peng, J., 2012a. Spatial and temporal variability of water discharge in the Yellow River Basin over the past 60 years. Journal of Geographical Sciences 22 (6), 1013e1033. Liu, F., Chen, S.L., Peng, J., Chen, G.Q., 2012b. Temporal variations of water discharge and sediment load of Huanghe River, China. Chinese Geographical Science 22 (5), 507e521. Liu, F., Chen, S.L., Peng, J., Chen, G.Q., 2011. Temporal variability of water discharge and sediment load of the Yellow River into the sea during 19502008. Journal of Geographical Sciences 21 (6), 1047e1061. Liu, F., Yuan, L.R., Yang, Q.S., Ou, S.Y., Xie, L.L., Cui, X., 2014a. Hydrological responses to the combined influence of diverse human activities in the Pearl River delta, China. Catena 113, 41e55. Liu, F., Yang, Q.S., Chen, S.L., Luo, Z.F., Yuan, F., Wang, R.T., 2014b. Temporal and spatial variability of sediment flux into the sea from the three largest rivers in China. Journal of Asian Earth Sciences 87, 102e115. Luo, X.L., Yang, Q.S., Jia, L.W., et al., 2002. River-bed Evolution of the Pearl River Delta Network. Zhongshan (Sun Yat-sen) University Press, Guangzhou, China (in Chinese). Meade, R.H., Parker, R.S., 1985. Sediment in rivers of the United States. National water summary, 1984. US Geological Survey Water-Supply Paper 2275, 49e60. Mikhailova, M.V., 2003. Transformation of the Ebro River Delta under the impact of intense human-induced reduction of sediment and runoff. Water Resources 30 (4), 370e378. Milliman, J.D., Farnsworth, K.L., Jones, P.D., Xu, K.H., Smith, L.C., 2008. Climatic and anthropogenic factors affecting river discharge to the Global Ocean, 1951e2000. Global and Planetary Change 62, 187e194. Nilsson, C., Reidy, C.A., Dynesius, M., Revenga, C., 2005. Fragmentation and flow regulation of the world's large river systems. Science 308, 405e408. Nourani, Vahid, Andalib, G., 2015. Daily and monthly suspended sediment load predictions using wavelet based artificial intelligence approaches. Journal of Mountain Science 12 (1), 85e100. Panda, D.K., Kumar, A., Mohanty, S., 2011. Recent trends in sediment load of the tropical (Peninsular) river basins of India. Global and Planetary Change 75, 108e118. Partal, T., Küçük, M., 2006. Long-term trend analysis using discrete wavelet components of annual precipitations measurements in Marmara region (Turkey). Physics and Chemistry of the Earth Parts A/B/C 31 (18), 1189e1200. ~ o/Southern Oscillation (ENSO)-related Shrestha, A., Kostaschuk, R., 2005. El Nin variability in mean-monthly streamflow in Nepal. Journal of Hydrology 308, 33e49. Siakeu, J., Oguchi, T., Aoki, T., 2004. Change in riverine suspended sediment concentration in central Japan in response to late 20th century human activities. Catena 55 (2), 231e254. €ro €smarty, C.J., Kettner, A.J., Green, P., 2005. Impact of humans on Syvitski, J.P.M., Vo the flux of terrestrial sediment to the global coastal ocean. Science 308, 376e380. http://dx.doi.org/10.1126/science.1109454. Walling, D.E., 2006. Human impact on landeocean sediment transfer by the world's rivers. Geomorphology 79 (3e4), 192e216. Walling, D.E., Fang, D., 2003. Recent trends in the suspended sediment loads of the world's rivers. Global and Planetary Change 39, 111e126. http://dx.doi.org/ 10.1016/S0921-8181(03)00020-1. Wang, H.J., Bi, N., Saito, Y., Wang, Y., Sun, X., Zhang, J., 2010. Recent changes in sediment delivery by the Huanghe (Yellow River) to the Sea: causes and
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043
12
C. Tan et al. / Quaternary International xxx (2016) 1e12
environmental implications in its Estuary. Journal of Hydrology 39 (3e4), 302e313. Wang, H.J., Yang, Z.S., Saito, Y., Liu, J.P., Sun, X., Wang, Y., 2007. Stepwise decreases of the Huanghe (Yellow River) sediment load (1950e2005): impacts of climate changes and human activities. Global and Planetary Change 57 (3e4), 331e354. Wang, H.J., Yang, Z.S., Saito, Y., Liu, J.P., Sun, X., 2006. Interannual and seasonal variation of the Huanghe (Yellow River) water discharge over the past 50 years: connections to impacts from ENSO events and dams. Global and Planetary Change 50, 212e225. Ward, P.J., Beets, W., Bouwer, L.M., Aerts, J.C.J.H., Renssen, H., 2010. Sensitivity of river discharge to ENSO. Geophysical Research Letters 37, L12402. http:// dx.doi.org/10.1029/2010GL043215. Wu, C.S., Yang, S.L., Lei, Y.P., 2012. Quantifying the anthropogenic and climatic impacts on water discharge and sediment load in the Pearl River (Zhujiang), China (1954e2009). Journal of Hydrology 190e204.
Yang, Z.S., Wang, H., Saito, Y., Milliman, J.D., Xu, K., Qiao, S., Shi, S., 2006. Dam impacts on the Changjiang (Yangtze) River sediment discharge to the sea: the past 55 years and after the three Gorges Dam. Water Resources Research 42, W04407. http://dx.doi.org/10.1029/2005WR003970. Zhang, Q., Xu, C.Y., Jiang, T., Wu, Y.J., 2007. Possible influence of ENSO on annual maximum streamflow of the Yangtze River, China. Journal of Hydrology 333 (2e4), 265e274. Zhang, S.R., Lu, X.X., 2009. Hydrological responses to precipitation variation and diverse human activities in a mountainous tributary of the lower Xijiang, China. Catena 77, 130e142. Zhang, S.R., Lu, X.X., Higgitt, D.L., Chen, C.T.A., Han, J.T., Sun, H.G., 2008. Recent changes of water discharge and sediment load in the Zhujiang (Pearl River) Basin, China. Global and Planetary Change 60, 365e380.
Please cite this article in press as: Tan, C., et al., Using the wavelet transform to detect temporal variations in hydrological processes in the Pearl River, China, Quaternary International (2016), http://dx.doi.org/10.1016/j.quaint.2016.02.043