Spatial–temporal evolution of the distribution pattern of river systems in the plain river network region of the Taihu Basin, China

Quaternary International xxx (2015) 1e9

Contents lists available at ScienceDirect

Quaternary International journal homepage: www.elsevier.com/locate/quaint

Spatialetemporal evolution of the distribution pattern of river systems in the plain river network region of the Taihu Basin, China Xiaojun Deng a, b, Youpeng Xu a, *, Longfei Han a, Mingnan Yang a, Liu Yang a, Song Song a, Guang Li a, Yuefeng Wang a a b

School of Geographic and Oceanographic Science, Nanjing University, No. 163 Xianlin Avenue, Nanjing, Jiangsu 210023, China School of Environmental Science and Resources, Guangxi Normal University, No. 15 Yucai Road, Guilin, Guangxi 541004, China

a r t i c l e i n f o

a b s t r a c t

Article history: Available online xxx

Understanding the spatialetemporal distribution characteristic and evolution tendency of river systems is important in integrated river basin management. In this paper, six indicators were adopted to describe the physical characteristic of river systems. The spatialetemporal evolution of the distribution pattern of river systems in the plain river network region of the Taihu Basin (PRNRTB) during the 1960se2000s was analyzed using exploratory spatial data analysis (ESDA) and the gravity centre model. The impacts of urbanization were then investigated. Results indicated that the global distribution patterns of box dimension and river density were all statistically significant spatially clustered, and river sinuosity was partially spatially clustered, while water surface ratio, main river area length ratio and river development coefficient were not spatially clustered. Moreover, the water surface ratio was the most stable parameter, box dimension was also relatively stable parameter, but the others were unstable. Meanwhile, the mutual transformations of hot and cold spots of river systems especially water surface ratio, river development coefficient, and river sinuosity were frequent. The hot spot regions were mainly located in the northeast region. In addition, most gravity centres migrated gradually from southeast to northwest with low distance, except those of river development coefficient and river density. These aforementioned differences were caused by the rapid urbanization. Based on the integrated river basin management suggestions for PRNRTB, the river density, box dimension and river sinuosity should be maintained and restored across the whole watershed, and the river conservation and restoration should focus on countermeasures against the reduction of river development coefficient, and main attention should be paid on the river systems conserving and restoring in Haiyan and Haining in southeast PRNRTB. © 2015 Elsevier Ltd and INQUA. All rights reserved.

Keywords: River systems Distribution pattern Spatialetemporal evolution Impacts of urbanization Exploratory spatial data analysis Taihu Basin

1. Introduction River systems are important natural heritage sites. The formation and development of river systems is influenced by many natural factors, such as geology, topography, soil, hydrology, climate, and vegetation. Moreover, river systems are usually altered to meet the needs of people in water, energy, transportation, recreation, storage and discharge (Costanza et al., 1997; Nilsson et al., 2005; dec and Statzner, 2008; Jia and Chen, 2013). The influences Dole of urbanization on river systems have been widely recognized as the most significant among all human activities. Approximately 60% of river systems have been changed profoundly because of

* Corresponding author. E-mail address: [email protected] (Y. Xu).

urbanization (Sear and Newson, 2003). These changes have strongly threatened the ecological integrity and ecosystem functions of river systems (Dudgeon, 2006). Thus, an increasing need to investigate the impacts of urbanization on river systems for the sustainable planning, management and conservation of rapidly urbanized river basins has been realized (Karr, 1999; Norris and Thoms, 1999; James and Marcus, 2006; Junior et al., 2010; Pinto and Maheshwari, 2011). The impacts of urbanization on river systems have been introduced and investigated as a broad and specific question since the mid-20th century (Lane, 1955; Strahler, 1956). Considerable progress has been achieved in rapidly urbanized regions around the world over the past 60 years (Chin, 2006). These studies documented that urbanization could change the quantity, morphology and structure of river systems. River lengths and water surface areas have generally deteriorated in rapidly urbanized regions.

http://dx.doi.org/10.1016/j.quaint.2015.04.010 1040-6182/© 2015 Elsevier Ltd and INQUA. All rights reserved.

Please cite this article in press as: Deng, X., et al., Spatialetemporal evolution of the distribution pattern of river systems in the plain river network region of the Taihu Basin, China, Quaternary International (2015), http://dx.doi.org/10.1016/j.quaint.2015.04.010

2

X. Deng et al. / Quaternary International xxx (2015) 1e9

Main rivers have also been widened because of the increase in runoff volume caused by river channel dredging and impervious surface increase, while the tributaries have been narrowed gradually and even disappeared because of river channel sedimentation and urban occupation (Vanacker et al., 2005; Gregory, 2006). Moreover, spatial distribution of the urbanization impacts has been observed, and found that the impacts typically decrease from the city to suburbs (Yuan et al., 2006). The variations of river systems can be considered as spatialetemporal processes. These procedures are controlled in spatial and time scale by many natural and social factors, particularly urbanization. Moreover, the spatial distribution of urbanization and its effects on river systems has the characteristic of agglomeration and dispersion. Previous studies have generally focused on the numeric characteristic of river system variation. These studies assumed the existence of mutual independence in spatial distribution (Yang et al., 2004; Xu et al., 2013). However, studies on the spatial association and spatial heterogeneity of the variations of river systems are lacking. Understanding the spatialetemporal distribution characteristic and the evolution tendency of river systems is necessary for sustainable river basin management. Exploratory spatial data analysis (ESDA) has been used successfully in the variation studies of economic patterns, ecological environments and social issues (Le Gallo and Ertur, 2003; Buttafuoco et al.,  n et al., 2013). 2005; Anselin et al., 2007; Ye and Wu, 2011; Rinco However, the methodology of ESDA in the evolution of river systems research is still at the beginning. Moreover, the plain river network region of the Taihu Basin (PRNRTB) is one of the regions

the region is only 0.46% of the total area of China. Moreover, PRNRTB is one of the most rapidly urbanized regions in the world. There are 22 large, medium and small size cities around Taihu Lake, such as Hangzhou, Suzhou, Wuxi, Changzhou, Jiaxing, Huzhou, Kunshan and Jiangyin. As one of the famous water-towns in the world, this region is characterized by the presence of many rivers and lakes, such as the Jiangnan Canal (the southern section of the BeijingeHangzhou Grand Canal), Wangyu River, Taipu River, Yangcheng Lake, Cheng Lake and Dianshan Lake. Rapid urbanization in the past 50 years has caused dramatic changes on the underlying surface of this region, particularly the river systems. Consequently, the rivers and lakes in PRNRTB have decreased significantly. 2.2. Data The river systems data of the 1960s and 1980s is derived from digitalized paper topographic maps, the river systems maps and the data of 2000s is from a digital line graphic at a scale of 1:50,000. According to the classification method for streams (Strahler, 1952) and its natural feature and social attribute, rivers are divided into four stream orders in PRNRTB (Table 1). Rivers wider than 40 m are classified as primary rivers. Rivers with width between 20 and 40 m are considered as secondary rivers. Those sized between 10 and 20 m in width are classified as tertiary rivers. Rivers less than 10 m in width are the quaternary rivers. Primary and secondary rivers are viewed as main rivers, and tertiary and quaternary rivers are considered tributary rivers.

Table 1 Classification method of rivers of PRNRTB. Types

Orders

Widths

Graphical representation Paper topographic map

Digital line graphic

Main rivers

Primary rivers Secondary rivers Tertiary rivers Quaternary rivers

>40 m 20e40 m 10e20 m <10 m

Double line rivers (>0.8 mm) Double line rivers (0.4e0.8 mm) Wide single line rivers (0.3 mm) Narrow single line rivers (0.15 mm)

Planar canal, surface rivers and main channel Planar canal, surface rivers and main channel Linear surface rivers and main channel Linear branch channel

Tributary rivers

with the fastest rate of urbanization in China. Rapid urbanization has caused significant changes in river systems, such as water degradation, flood disaster and other ecological and environmental problems. Therefore, many major indicators are selected to describe the physical characteristic of river systems. ESDA and the gravity centre model are employed to explore the spatialetemporal evolution of the distribution pattern of river systems in PRNRTB. The aim of this study is to analyze the evolution characteristics and laws of the distribution pattern of river systems, to discuss the possible impacts of urbanization, and finally to provide scientific basis and decision-making references for integrated river basin management in PRNRTB. 2. Study area and data 2.1. Study area PRNRTB is located in the centre of the Yangtze River Delta in eastern China and covers an area of 15,757 km2, 2 me4 m above sea level (Fig. 1). PRNRTB is one of the most densely populated regions in China with approximately 3.47 million inhabitants and a population density of 834 individuals per km2. PRNRTB is also one of the most economically developed regions in China, i.e., PRNRTB had a local GDP of 3874.12 billion Yuan in 2013, approximately 7% of the GDP of China. However, the total area of

Main functions

Discharge Storage

3. Methodologies 3.1. Characteristic indicators of river systems River systems can be characterized by their physical properties, including original descriptive indicators and complex integrated indicators. The original descriptive indicators of river systems, including river length, river number, river bifurcation, stream order and water surface area, have rarely been used directly in previous studies. The river complex integrated indicators were in the other hand have been widely used in recent decades, such as river density, river frequency, water surface ratio, river development coefficient, river systems complexity, river systems stability, river bifurcation ratio, river length ratio, river sinuosity, main river area length ratio, box dimension and other complex integrated indicators of river systems (Tarboton et al., 1988; Roth et al., 1996; Schuller et al., 2001; Wang et al., 2011). In our research, these complex integrated indicators are classified into quantitative characteristics, morphological characteristics and structure characteristics indicators. In accordance with the characteristics of river systems in PRNRTB and the possibility of data acquisition, we select six indicators, including river density, water surface ratio, river development coefficient, river sinuosity, main river area length ratio and box dimension to describe the physical characteristic of river systems in PRNRTB (Table 2).

Please cite this article in press as: Deng, X., et al., Spatialetemporal evolution of the distribution pattern of river systems in the plain river network region of the Taihu Basin, China, Quaternary International (2015), http://dx.doi.org/10.1016/j.quaint.2015.04.010

X. Deng et al. / Quaternary International xxx (2015) 1e9

3

Table 2 Characteristic indicators and computational methods of river systems. Category

Indicator

Computational methods

Quantitative characteristics

River density Water surface ratio River development coefficient River sinuosity

Dr ¼ L/A (L is the total river length, and A is the river basin area) Wp ¼ Aw/A (Aw is the total area of rivers and lakes under the mean water level) Ku ¼ Lt/Lm (Lt is the total tributary length, and Lm is the total main river length) P Lai Lai Sr ¼ m i L  Lsi (Lai is the actual length of the ith river, and Lsi is the straight-line

Morphological characteristics

Structure characteristics

Main river area length ratio Box dimension

distance from the ith river starting point to its terminal point) Rm ¼ Am/Lm (Am is the main river total area) D0 ¼  lim

r/0

lgNðrÞ lgr

(N is the number of boxes with some rivers, and r is the side

length of the box)

3.2. Exploratory spatial data analysis (ESDA) ESDA is an extension of exploratory data analysis that focuses on the description and interpretation of the spatial relationships of regionalized variables, particularly spatial autocorrelation and spatial heterogeneity. ESDA is also used to describe and visualize spatial distributions, identify atypical locations or spatial outliers, discover patterns of spatial association and suggest all types of spatial heterogeneity (Anselin, 1995; Anselin et al., 2007). In this paper, we adopt the Global Moran Index and the Getis-Ord Gi* Index to explore the spatialetemporal evolution of the distribution pattern of river systems. 3.2.1. Global spatial autocorrelation The global evolution characteristic of the distribution pattern of river systems is detected by the Global Moran Index. It is a

measurement method of overall spatial autocorrelation for quantifying the degree of clustering or dispersion. This method is calculated as follows (Moran, 1950):

I ¼ Pn i¼1

n Pn j¼1

Pn

i¼1

wi; j

Pn

j¼1

Pn

wi; j zi zj

2 i¼1 zi

;

(1)

where Zi and Zj are the standardization of the actual values of the characteristic indicators of river systems (xi and xj) in cities i and j, wi,j is the spatial weight between cities i and j and n is equal to the total number of cities. According to the Z score and p-value, we can evaluate the significance of the Global Moran Index of river systems. The value range of the Global Moran Index of river systems is from 1 to þ1. The value of the Global Moran Index near þ1 indicates the clustering distribution of river systems, and the value

Fig. 1. Location of the plain river network region of the Taihu Basin (PRNRTB).

Please cite this article in press as: Deng, X., et al., Spatialetemporal evolution of the distribution pattern of river systems in the plain river network region of the Taihu Basin, China, Quaternary International (2015), http://dx.doi.org/10.1016/j.quaint.2015.04.010

4

X. Deng et al. / Quaternary International xxx (2015) 1e9

of the Global Moran Index near 1 indicates the dispersion distribution of river systems. The random distribution of river systems is indicated when the value of the Global Moran Index is zero. 3.2.2. Local spatial autocorrelation The Getis-Ord Gi* Index can be used to explore the existence of local spatial clusters with high values (hot spots) or low values (cold spots) and recognize inner spatial heterogeneity by analysing the correlation extent of the characteristic indicators of river systems in the close city. This index can be described as follows (Getis and Ord, 1992):

G*i ¼

n X

, wi;j ðdÞxj

j¼1

n X

xj ;

(2)

j¼1

 ffi  rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi   Var G*i ; Z G*i ¼ G*i  E G*i

(3)

4. Results and discussion 4.1. Global evolution of the distribution pattern of river systems The Global Moran Index of characteristic indicators of river systems during the 1960se2000s are calculated by the spatial statistics tools of the spatial autocorrelation (Moran's I) in ArcGIS 9.3. Moreover, the adjacency method is adopted to calculate the spatial weight matrix. Table 3 shows that the positive Moran's I values of characteristic indicators of river systems and indicates the clustered characteristics of the spatial distribution of river systems. Regions with higher (resp. lower) characteristic indicators are close to each other. Large differences are found between the Moran's I values of six characteristic indicators. These values are sorted in descending order as Dr, D0, Sr, Wp, Rm and Ku. Moreover, the Moran's I values of D0 and Dr in the three periods are all at the 95% confidence level. The Moran's I values of Sr in the 1980s and 2000s pass the test in the confidence intervals of 95% and 90%, respectively.

Table 3 Moran's I of characteristic indicators of river systems in PRNRTB during the 1960se2000s. Dr

Moran's I Z score Sig

Wp 1980s

2000s

1960s

1980s

2000s

1960s

1980s

2000s

0.41 2.77 0.01

0.48 3.17 0.01

0.43 2.93 0.01

0.15 1.31 e

0.15 1.33 e

0.17 1.52 e

0.18 1.39 e

0.06 0.66 e

0.03 0.10 e

1960s

1980s

2000s

1960s

1980s

2000s

1960s

1980s

2000s

0.13 1.08 e

0.37 2.50 0.05

0.26 1.92 0.10

0.19 1.55 e

0.11 0.41 e

0.17 1.27 e

0.38 2.57 0.05

0.40 2.74 0.01

0.34 2.46 0.05

Rm

Sr

Moran's I Z score Sig

Ku

1960s

D0

where xi and xj are the actual values of the characteristic indicators of river systems in cities i and j, wi,j is the spatial weight between the i city and j city, n is equal to the total number of cities, Z(Gi*) is the standardization of Gi*, E(Gi*) is the mathematical expectation value of Gi* and Var(Gi*) is the variance of Gi*. Larger Z(Gi*) values indicate a more intense clustering of high values (hot spots) of river systems distributions, whereas smaller Z(Gi*) values indicate a more intense clustering of low values (cold spots) of river systems distributions. 3.3. Gravity centre model The gravity centre model is a spatial statistics method that evaluates the geographic distribution and identifies the geographic centre (or the centre of concentration) of regionalized variable. The direction, velocity and distance of gravity centre movement can describe the degree of clustering and migration of a regionalized variable in time and space:

X¼

n X

, wi xi

i¼1

Y¼

n X i¼1

n X

wi ;

(4)

wi ;

(5)

i¼1

, wi yi

n X i¼1

where xi and yi are the coordinates for feature i, wi is the weight at feature i and n is equal to the total number of features.

In conclusion, the global distribution patterns of Dr and D0 are statistically significant and spatially clustered, but the variations of clustering are insignificant overall. The main reason is that the most dense and complex river network is located dramatically in northeast PRNRTB. This status is changed little because of the dramatic decrease of the number of rivers over the past 50 years. Meanwhile, the global distribution pattern of Sr is partially spatially clustered in three periods. Generally, the Sr of the natural river is larger than the artificial river, and the largest Sr is always located in the most dense and complex river network region. Rivers are modified dramatically to meet the demands of discharge and landscape in PRNRTB since 1960s, and the distributions of Sr change gradually into clustering from dispersion. The clustering of Wp, Rm and Ku is insignificant. Despite the amount of rivers and lakes dramatically decreased in PRNRTB over the past decades, the original dispersive water surfaces are nearly unchanged. The variations of Rm are mainly affected by river dredging and river diversion, and the distribution of Rm is always dispersive. Moreover, there is always the dispersive Ku in PRNRTB from 1960s to 2000s, due to the inconsistent decrease of the tributary in different regions. The analysis indicates that almost all characteristic indicators of river systems in PRNRTB decreased over the past 50 years. The decrease of Dr and D0 involved the whole watershed during the 1960se2000s, and the Sr is decreased gradually in part of the region in the 1st period and extended to the whole watershed later. The decrease of Wp, Rm and Ku all are partial in the three periods. Therefore, Dr, D0 and Sr must be maintained and restored at the scale of the whole watershed, and Wp, Rm and Ku must be maintained and restored at the regional scale.

Please cite this article in press as: Deng, X., et al., Spatialetemporal evolution of the distribution pattern of river systems in the plain river network region of the Taihu Basin, China, Quaternary International (2015), http://dx.doi.org/10.1016/j.quaint.2015.04.010

X. Deng et al. / Quaternary International xxx (2015) 1e9

4.2. Local evolution of the distribution pattern of river systems The Getis-Ord Gi* statistic is used to measure the local spatial autocorrelation of characteristic indicators of river systems to identify the evolution trend of the distribution pattern of river systems in PRNRTB. Furthermore, the Getis-Ord Gi* statistic of different periods are divided into hot spots, sub-hot spots, sub-cold spots and cold spots according to the classification method of Natural Breaks (Jenks). Here, hot spots and sub-hot spots are the low risk regions of the decrease of river systems due to their larger characteristic value of river systems, and cold spots and sub-cold spots regions are under high risk. Fig. 2 shows the significant spatial difference in the local evolution of the distribution patterns of river systems in PRNRTB.

5

cold spots of D0 change gradually into a spindle structure. The quantitative structure of hot and cold spots of Dr was also in a spindle structure in the 1960s, but this structure was destroyed because of the dramatic increase of the number of hot spots since the 1980s. Moreover, the quantitative spindle structures of hot and cold spots of Rm changed in the 1980s. However, the quantitative structures of hot and cold spots of Wp, Ku, and Sr are always out of the spindle structure because of the frequent changes of the hot and cold spots. Overall, the quantitative structures of hot and cold spots of river systems all are unstable. This indicates the frequent mutual transformations of hot and cold spots of river systems, especially Wp, Ku, and Sr. Therefore, river systems should be conserved and restored, and focus should be on protecting against the reduction of Wp, Ku, and Sr in PRNRTB.

Table 4 The number and proportion of hot and cold spots of six characteristic indicators of river systems in PRNRTB. Indicators

Dr

Wp

Ku

Sr

Rm

D0

Regions

hot spots sub-hot spots sub-cold spots cold spots hot spots sub-hot spots sub-cold spots cold spots hot spots sub-hot spots sub-cold spots cold spots hot spots sub-hot spots sub-cold spots cold spots hot spots sub-hot spots sub-cold spots cold spots hot spots sub-hot spots sub-cold spots cold spots

1960s

1980s

2000s

Quantity

Proportion

Quantity

Proportion

Quantity

Proportion

3 6 8 5 7 3 6 6 2 6 7 7 6 7 5 4 4 3 8 7 5 6 5 6

13.64% 27.27% 36.36% 22.73% 31.82% 13.64% 27.27% 27.27% 9.09% 27.27% 31.82% 31.82% 27.27% 31.82% 22.73% 18.18% 18.18% 13.64% 36.36% 31.82% 22.73% 27.27% 22.73% 27.27%

8 3 8 3 7 5 5 5 2 5 7 8 6 5 7 4 3 5 8 6 5 6 8 3

36.36% 13.64% 36.36% 13.64% 31.82% 22.73% 22.73% 22.73% 9.09% 22.73% 31.82% 36.36% 27.27% 22.73% 31.82% 18.18% 13.64% 22.73% 36.36% 27.27% 22.73% 27.27% 36.36% 13.64%

8 2 9 3 7 3 7 5 4 11 4 3 6 6 5 5 3 7 5 7 4 6 9 3

36.36% 9.09% 40.91% 13.64% 31.82% 13.64% 31.82% 22.73% 18.18% 50.00% 18.18% 13.64% 27.27% 27.27% 22.73% 22.73% 13.64% 31.82% 22.73% 31.82% 18.18% 27.27% 40.91% 13.64%

Overall, the distribution patterns of Wp is the most stable, the central PRNRTB are always the clustering regions of water surface, and cold spots are consistently located in northwest and southeast. Therefore, Wp should be protected against reduction in northwest and southeast. Moreover, the distribution patterns of D0 are also relatively stable. The hot spots of D0 are mainly located in northeast PRNRTB, and the cold spots are mainly located in the southwest. D0 protection is in higher emergency in the southwest. However, almost all hot spots of Rm are clustered in centre and southern, but cold spots are mainly located in the northeast and southeast. Therefore, Rm should be protected against reduction in northeast and southeast. Meanwhile, the hot spots of Dr and Ku migrate gradually from southeast to northeast. Therefore, Dr and Ku should be protected against reduction in the southwest. The hot spots of Sr migrate gradually from east to south, but the cold spots are mainly located in the southeast. Therefore, Sr should be protected against reduction in southeast. In conclusion, southwest and southeast should be the main regions for river conservation and restoration in PRNRTB, Dr, Ku and D0 should be protected against reduction in the southwest, and Wp, Sr and Rm should also be protected against reduction in the southeast. Moreover, wide differences of the quantitative structure of hot and cold spots are also detected among six characteristic indicators of river systems (Table 4). The quantitative structures of hot and

Fig. 3 shows the total number of hot and cold spots of different regions in PRNRTB in three periods. Compared with other regions, Qingpu, Songjiang, Jinshan, Taicang, Zhangjiagang, Changshu, and Kunshan in eastern PRNRTB are the main hot spots regions. A main reason is that the eastern region is always the most dense and complex river network regions in PRNRTB. Moreover, the quantity, morphology and structure of river systems in this region are also changed over the past 50 years, but these variations are smaller than other regions in PRNRTB. The main cold spots are located in Deqing, Yuhang and Hangzhou in southwest PRNRTB, and Changzhou, Jiangyin and Wuxi in northwest PRNRTB, and Haiyan and Haining in southeast PRNRTB. It is mainly due to the smaller quantity, and the simpler morphological structure of river systems in this region. Therefore, river conservation and restoration should be focused on the Deqing, Hangzhou, Yuhang, Changzhou, Haiyan and Haining in PRNRTB. 4.3. Migration of the gravity centres of river systems The above analysis revealed the global and local spatial differences of distribution pattern of river systems at different periods. The gravity centre model is used to calculate the gravity centre coordinates of the characteristic indicators of river systems at different periods to further explore the spatial evolution process of river

Please cite this article in press as: Deng, X., et al., Spatialetemporal evolution of the distribution pattern of river systems in the plain river network region of the Taihu Basin, China, Quaternary International (2015), http://dx.doi.org/10.1016/j.quaint.2015.04.010

6

X. Deng et al. / Quaternary International xxx (2015) 1e9

Fig. 2. Local evolution of the distribution patterns of river systems in PRNRTB during the 1960se2000s.

systems. The spatial migration maps of the gravity centre coordinates are plotted in accordance with the aforementioned analysis. Fig. 4 shows 18 gravity centres of six characteristic indicators of river systems in three periods. Among these centres, almost all gravity centres are located in Wujiang in central PRNRTB. The figure shows that the Wujiang river systems are important in PRNRTB. However, a large difference is also found in the directions and distances of the gravity centre migrations of the different characteristic indicators of river systems. Overall, the gravity centre of Dr, Ku, and D0 migrate towards northwest with larger distances during the 1980se2000s. It indicates the decreases of Dr, Ku and D0 in southeast PRNRTB are more intense, and more significant during the 1980se2000s. Meanwhile, despite the gravity centre of Wp and Sr also migrate towards northwest, but larger distances are migrated during the 1960se1980s. The gravity centre of Rm migrates a larger distance towards southwest during the

1960se1980s. It indicates the decrease of Rm is more intense spatially in northeast PRNRTB during the 1960se1980s in time series. Moreover, the total distances of the gravity centre migration of Sr, Rm, D0 and Wp are extremely small, and it is 0.35, 0.96, 1.06 and 1.91 km, respectively, while the total distance of the gravity centre migration of Dr and Ku is 4.14 and 10.89 km, respectively. It indicates the spatial variations of Sr, Rm, D0 and Wp are much smaller than Dr and Ku. Therefore, Ku should be conserved and restored preferentially, and the southeast region should be the key region for conserving and restoring river systems in PRNRTB. 4.4. Impacts of urbanization on the distribution pattern of river systems Urban development occupies water surfaces via extensive buildings, roads and bridges. The dramatic reduction of rivers and

Fig. 3. The total number of hot and cold spots of different regions in PRNRTB.

Please cite this article in press as: Deng, X., et al., Spatialetemporal evolution of the distribution pattern of river systems in the plain river network region of the Taihu Basin, China, Quaternary International (2015), http://dx.doi.org/10.1016/j.quaint.2015.04.010

X. Deng et al. / Quaternary International xxx (2015) 1e9

7

Fig. 4. Migration of the gravity centres of the characteristic indicators of river systems in PRNRTB during the 1960se2000s.

lakes, particularly the low-order river, is an important impact of urbanization on river systems (Gregory, 2006; Xu et al., 2013). This phenomenon is common in rapidly urbanized regions in PRNRTB, such as Haiyan. The total length of quaternary rivers is 2324.74 km in Haiyan in the 1960s, 1543.86 km in the 1980s and 915.10 km in the 2000s. It evidently decreases 60.64% over the past 50 years. Fig. 5 shows the decrement of the tributary in southeast is larger than in northeast in the recent 50 years in PRNRTB. However, rivers in northeast PRNRTB are denser than southeast in 1960s, especially in Zhangjiagang, Changshu, Taicang and Kunshan. Therefore, the hot spots of Dr and Ku migrate gradually from southeast to northeast. Moreover, the dramatic reduction of rivers and lakes also influence the distribution pattern of D0 to a certain degree. D0 in northern PRNRTB are all about 1.60 in three periods, but constantly less than 1.50 in southern part. Meanwhile, the average variation rate of D0 in northern PRNRTB is only 5.69% during the 1960se2000s. Additionally, the average decrease rate of D0 in southern PRNRTB is larger than in the northern part. Therefore, northern PRNRTB remains as the location in recent decades, despite the different ranges of hot spots of D0 in the three periods. In addition, Fig. 6 shows that the lakes are mainly located in the central PRNRTB. Meanwhile, the total decrease rate of Wp in PRNRTB is 18.95%. However, Wp in Qingpu, Wujiang, Suzhou and Kunshan are all above 16% during the 1960se2000s, but it is constantly less than 10% in other cities. Thus, hot spots of Wp consistently remain in central PRNRTB in recent decades. Rapid urban growth often threatens the survival and development of human beings by causing water resource shortage and water environment pollution. In this instance, human beings repeatedly change the river systems to achieve long distance water transfer and improvement of water environment, such as river dredging and river diversion. Meanwhile, several channel modifications, such as straightening and widening, are also made in rapidly urbanized regions to meet the needs of flood control and inland navigation. Moreover, some urban river channels are regularly changed to meet landscape and recreation needs, such as channelization and hardening (Chin, 2006; Gregory, 2006). Similarly, the various abovementioned changes are also extremely

common in PRNRTB, especially before the 1980s. These changes partially influence the distribution pattern of Sr and Rm. Overall, the values variation of Sr over the past 50 years in PRNRTB is extremely small. These values range from 1.03 to 1.21.

Fig. 5. Spatial difference of the decrement of the tributary in PRNRTB during the 1960se2000s.

Please cite this article in press as: Deng, X., et al., Spatialetemporal evolution of the distribution pattern of river systems in the plain river network region of the Taihu Basin, China, Quaternary International (2015), http://dx.doi.org/10.1016/j.quaint.2015.04.010

8

X. Deng et al. / Quaternary International xxx (2015) 1e9

Fig. 6. Spatial distribution of main lakes and drainage projects in PRNRTB.

Meanwhile, Sr in southeast PRNRTB, such as Songjiang, Jinshan, Haiyan, Haining, and Pinghu, is slightly larger than other regions in the 1960s. Moreover, the decrease of the low-order river, straightening of main rivers and channelization of urban river channels causes the decrease of Sr in most of PRNRTB in recent 50 years. Minimal change for Sr in Huzhou and Deqing is observed since 1960s. Slower urbanization compared with other cities in PRNRTB minimizes the changes. Therefore, hot spots of Sr migrate gradually from southeast to southwest. In addition, frequent and serious flood disasters are considered as constant main natural calamities in the Taihu Basin in recent decades. To mitigate flood disasters, several drainage projects were constructed to discharge floods into the Yangtze River, Huangpu River and Hangzhou Bay in PRNRTB (Fig. 6), such as Taipu River, Hongqi Tang, Shanghai Tang and Pinghu Tang. Moreover, Dongtiao River was dredged and widened in recent decades to control the flood of Hangzhou city. However, these drainage projects are mainly located in central and southern PRNRTB. These projects were densely located on eastern PRNRTB before 1980s, but developed in the central and southwest PRNRTB between 1980s and 2000s. Therefore, hot spots of Rm migrated gradually from the east to centre and southwest. 5. Conclusions In this paper, the spatialetemporal evolution of the distribution pattern of river systems in PRNRTB during the 1960s, 1980s, and 2000s was analysed. Moreover, the impacts of urbanization were further discussed. The following conclusions are gained:

(1) The global distribution pattern of river systems has partially clustered characteristics in PRNRTB over the past 50 years. The global distribution patterns of D0 and Dr were all statistically significant spatially clustered in three periods. In addition, the global distribution pattern of Sr was partially spatially clustered in the 1980s and 2000s. (2) Significant differences were also found in the local evolution of the distribution pattern of characteristic indicators of river systems in PRNRTB over the past 50 years. Overall, the distribution patterns of Wp was the most stable, D0 was also relatively stable, but the other indicators all were unstable. The mutual transformations of hot and cold spots of river systems were frequent, especially Wp, Ku, and Sr. Moreover, eastern PRNRTB was the main hot spots region, but the main cold spots are located in the southwest, southeast and northwest PRNRTB. (3) Most gravity centres of characteristic indicators of river systems are located in Wujiang in central PRNRTB, and they migrated gradually towards the northwest from southeast during the 1960s and 2000s. The total distance of the gravity centre migration of Ku was 10.89 km over the past 50 years, and the total migration distance of Dr was 4.14 km, while the total migration distances of other characteristic indicators all were about 1.00 km. (4) These aforementioned differences in the spatialetemporal evolution of the distribution patterns of river systems were caused by rapid urbanization. An important impact of urbanization on river systems was the dramatic decrease of the low-order river in PRNRTB. This effect caused the evolution of the distribution patterns of Dr, Ku, D0 and Wp. Meanwhile, urbanization also indirectly influenced the distribution patterns of Sr and Rm by straightening, widening, hardening, channelization, river dredging, and diversion. (5) Overall, ESDA and the gravity centre model could effectively combine the advantages of statistics analysis and spatial analysis, and they could describe and visualize the spatialetemporal distribution characteristic and evolution tendency of river systems. Moreover, our results provided suggestions for integrated river basin management in PRNRTB, as follows: Dr, D0 and Sr should be maintained and restored at the scale of the whole watershed, but Wp, Rm and Ku should be maintained and restored at the regional scale; river conservation and restoration should focus on protecting against the reduction of Wp, Ku, and Sr, while Ku should be conserved and restored preferentially; southwest and southeast PRNRTB should be the main areas of river conservation and restoration, and the main attention should be paid to river conserving and restoring in Haiyan and Haining in southeast PRNRTB. In conclusion, urbanization has an evident influence on the evolution of the distribution pattern of river systems. Nevertheless, the specific contribution of urbanization to the distribution pattern evolution of each characteristic indicator of river systems in PRNRTB should be further quantified. Moreover, further studies using longer time data and more characteristic indicators from other river basins are required to strengthen the conclusions on the impacts of urbanization on the distribution pattern evolution of river systems. In addition, more studies should be conducted on the impacts of the distribution pattern evolution of river systems and urbanization on hydrologic connectivity and water level in PRNRTB. The results of these studies would provide more scientific basis and decision-making reference for integrated river basin management in PRNRTB.

Please cite this article in press as: Deng, X., et al., Spatialetemporal evolution of the distribution pattern of river systems in the plain river network region of the Taihu Basin, China, Quaternary International (2015), http://dx.doi.org/10.1016/j.quaint.2015.04.010

X. Deng et al. / Quaternary International xxx (2015) 1e9

Acknowledgements This research work was supported by the National Natural Science Foundation of China under Grant 41371046, 41401035; and Commonweal and Specialized Program for Scientific Research, Ministry of Water Resources of China under Grant 201201072, 201301075; and the Natural Science Foundation of Jiangsu Province under Grant BK20131276. References Anselin, L., 1995. Local indicators of spatial association-LISA. Geographical Analysis 27 (2), 93e115. Anselin, L., Sridharan, S., Gholston, S., 2007. Using exploratory spatial data analysis to leverage social indicator databases: the discovery of interesting patterns. Social Indicators Research 82 (2), 287e309. Buttafuoco, G., Castrignano, A., Busoni, E., Dimase, A.C., 2005. Studying the spatial structure evolution of soil water content using multivariate geostatistics. Journal of Hydrology 311 (1), 202e218. Chin, A., 2006. Urban transformation of river landscapes in a global context. Geomorphology 79 (3), 460e487. Costanza, R., d'Arge, R., de Groot, R., Farber, S., Grasso, M., Hannon, B., Limburg, K., Naeem, S., Oneill, R.V., Paruelo, J., Raskin, R.G., Sutton, P., vanden Belt, M., 1997. The value of the world's ecosystem services and natural capital. Nature 387 (6630), 253e260. Doledec, S., Statzner, B., 2008. Invertebrate traits for the biomonitoring of large European rivers: an assessment of specific types of human impact. Freshwater Biology 53 (3), 617e634. Dudgeon, D., 2006. The impacts of human disturbance on stream benthic invertebrates and their drift in North Sulawesi, Indonesia. Freshwater Biology 51 (9), 1710e1729. Getis, A., Ord, J.K., 1992. The analysis of spatial association by use of distance statistics. Geographical Analysis 24 (3), 189e206. Gregory, K.J., 2006. The human role in changing river channels. Geomorphology 79 (3), 172e191. James, L.A., Marcus, W.A., 2006. The 2006 Binghamton geomorphology symposium on the human role in changing fluvial systems. Geomorphology 79 (3), 144e147. Jia, Y.T., Chen, Y.F., 2013. River health assessment in a large river: bioindicators of fish population. Ecological Indicators 26, 24e32. Junior, O.C., Guimaraes, R., Freitas, L., Gomes-Loebmann, D., Gomes, R.A., Martins, E., Montgomery, D.R., 2010. Urbanization impacts upon catchment hydrology and gully development using multi-temporal digital elevation data analysis. Earth Surface Processes and Landforms 35 (5), 611e617. Karr, J.R., 1999. Defining and measuring river health. Freshwater Biology 41 (2), 221e234.

9

Lane, E.W., 1955. The importance of fluvial morphology in hydraulic engineering. Proceedings of the American Society of Civil Engineers 81, 1e17. Le Gallo, J., Ertur, C., 2003. Exploratory spatial data analysis of the distribution of regional per capita GDP in Europe, 1980e1995. Papers in Regional Science 82 (2), 175e201. Moran, P.A.P., 1950. Notes on continuous stochastic phenomena. Biometrika 37 (1e2), 17e23. Nilsson, C., Reidy, C.A., Dynesius, M., Revenga, C., 2005. Fragmentation and flow regulation of the world's large river systems. Science 308 (5720), 405e408. Norris, R.H., Thoms, M.C., 1999. What is river health. Freshwater Biology 41 (2), 197e209. Pinto, U., Maheshwari, B.L., 2011. River health assessment in peri-urban landscapes: an application of multivariate analysis to identify the key variables. Water Research 45 (13), 3915e3924. n, R.A., Pascual, U., Romero, M., 2013. An exploratory spatial analysis of illegal Rinco coca cultivation in Colombia using local indicators of spatial association and socioecological variables. Ecological Indicators 34, 103e112. Roth, G., La Barbera, P., Greco, M., 1996. On the description of the basin effective drainage structure. Journal of Hydrology 187 (1), 119e135. Schuller, D.J., Rao, A.R., Jeong, G.D., 2001. Fractal characteristics of dense stream networks. Journal of Hydrology 243 (1), 1e16. Sear, D.A., Newson, M.D., 2003. Environmental change in river channels: a neglected element. Towards geomorphological typologies, standards and monitoring. Science of the Total Environment 310 (1), 17e23. Strahler, A.N., 1952. Hypsometric (area-altitude) analysis of erosional topography. Geological Society of America Bulletin 63, 1117e1142. Strahler, A.N., 1956. The nature of induced erosion and aggradation. In: Thomas, W.L. (Ed.), Man's Role in Changing the Face of the Earth. University of Chicago Press, Chicago, USA, pp. 621e638. Tarboton, D.G., Bras, R.L., Rodriguez-Iturbe, I., 1988. The fractal nature of river networks. Water Resource Research 24 (8), 1317e1322. Vanacker, V., Molina, A., Govers, G., Poesen, J., Dercon, G., Deckers, S., 2005. River channel response to short-term human-induced change in landscape connectivity in Andean ecosystems. Geomorphology 72 (1), 340e353. Wang, C., Wang, D., Wang, H., Dong, R., 2011. Impacts of urbanisation on river systems and their functions in Yanggong River watershed of Lijiang City. International Journal of Sustainable Development & World Ecology 18 (6), 498e502. Xu, G.L., Xu, Y.P., Wang, L.Y., 2013. Temporal and spatial changes of river systems in Hangzhou-Jiaxing-Huzhou Plain during 1960se2000s. Acta Geographica Sinica 68 (7), 966e974 (in Chinese). Yang, K., Yuan, W., Zhao, J., Xu, S.Y., 2004. Stream structure characteristics and its urbanization responses to tidal river system. Acta Geographica Sinica 59 (4), 557e564 (in Chinese). Ye, X., Wu, L., 2011. Analyzing the dynamics of homicide patterns in Chicago: ESDA and spatial panel approaches. Applied Geography 31 (2), 800e807. Yuan, W., Philip, J., Yang, K., 2006. Impact of urbanization on structure and function of river system. Chinese Geographical Science 16 (2), 102e108.

Please cite this article in press as: Deng, X., et al., Spatialetemporal evolution of the distribution pattern of river systems in the plain river network region of the Taihu Basin, China, Quaternary International (2015), http://dx.doi.org/10.1016/j.quaint.2015.04.010