Hydrological response to climate change and human activities:A case studyof Taihu Basin,China

Jun Wu ,Zhi-yong Wu *,He-jun Lin ,Hi-ping Ji ,Min Liu

a Bureau of Hydrology,Taihu Basin Authority of Ministry of Water Resources,Shanghai 200434,China

bCollege of Hydrology and Water Resources,Hohai University,Nanjing 210098,China

Abstract Climate change and human activities have changed a number of characteristics of river flow in the Taihu Basin.Based on long-term time series of hydrological data from 1986 to 2015,we analyzed variability in precipitation,water stage,water diversion from the Yangtze River,and net inflow into Taihu Lake with the Mann-Kendall test.The non-stationary relationship between precipitation and water stage was first analyzed for the Taihu Basin and the Wuchengxiyu(WCXY)sub-region.The optimized regional and urban regulation schemes were explored to tackle high water stage problems through the hydrodynamic model.The results showed the following:(1)The highest,lowest,and average Taihu Lake water stages of all months had increasing trends.The total net inflow into Taihu Lake from the Huxi(HX)sub-region and the Wangting Sluice increased significantly.(2)The Taihu Lake water stage decreased much more slowly after 2002;it was steadier and higher after 2002.After the construction of Wuxi urban flood control projects,the average water stage of the inner city was 0.16-0.40 m lower than that of suburbs in the flood season,leading to the transfer of flooding in inner cities to suburbs and increasing inflow from HX into Taihu Lake.(3)The regional optimized schemes were more satisfactory in not increasing the inner city flood control burden,thereby decreasing the average water stage by 0.04-0.13 m,and the highest water stage by 0.04-0.09 m for Taihu Lake and the sub-region in the flood season.Future flood control research should set the basin as the basic unit.Decreasing diversion and drainage lines along the Yangtze River can take an active role in flood control.© 2020 Hohai University.Production and hosting by Elsevier B.V.This is an open access article under the CC BY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords:Hydrological response;Climate change;Human activities;Flood control;Mann-Kendall test;Taihu Basin

1.Introduction

The hydrological cycle of a catchment is a complex process that is influenced by climate,physical characteristics of the catchment,and human activities(Liu and Xu,2015;Wang et al.,2011b;Zhu et al.,2015b;Xia et al.,2018;Hasan and Wyseure,2018).Climate determines the main inputs to the hydrological cycle,such as rainfall and radiation.Human activities have persistent influences on runoff distribution.In the past sixty years,both global climate change and intensive human activities have caused remarkable influences on hydrological process all over the world(Liu et al.,2016).A number of studies have shown that elements of the hydrological cycle,such as precipitation and water stage,have changed both in time and space,due to hydraulic engineering construction(Gu et al.,2017).Urbanization on river systems has been widely recognized as the most significant influence of all human activities(Shi et al.,2010),leading to hydrological regime variation and uncertainties in flood control.With impervious areas associated with buildings and transportation infrastructure increasing(Gregory,2006;Zhang et al.,2017),both surface runoff and flow velocity into rivers have increased significantly,leading to frequent regional precipitous flood occurrence (Yuan et al., 2006).

With the background of global climate change (Todorov et al., 2018), the distribution of water resources and the frequency of floods and droughts for the Taihu Basin in recent years has undergone a lot of change,elevating the discrepancy between water supply and demand (Zhou et al., 2013). Most rivers in the Taihu Basin have been interrupted by sluices,pump stations,and floodgates(Zhao et al.,2011).There are 68 sluices along the Yangtze River,186 sluices surrounding Taihu Lake, and nine sluices along Hangzhou Bay to defend against flooding and meet water resources demand (Yin et al., 2009).

According to Li et al. (2013), urbanization in China has been in a state of acceleration since 1978. Following a basinwide catastrophic flood in 1991, the first round of comprehensive regulation projects were begun. The three main basin flood control projects(the Wangyu River,the Taipu River,and Lake Levee) were completed by the end of 1999, while regional flood control projects were completed by the end of 2002. The Taihu Basin Authority (TBA) began to carry out water diversion from the Yangtze River to the Taihu Basin through the Wangyu River and to supply water downstream through the Taipu River in 2002. To ensure flood prevention security in the cities, including Suzhou, Huzhou, Jiaxing,Wuxi,and Changzhou,urban flood control projects have been built around Taihu Lake since 2003 with a flood protection standard of 100 years or 200 years.The flood control capacity has been further enhanced since 2002, when the complex system of flood control infrastructure was formed, including dikes, large sluice gates, and pumping stations. This study addressed the effects of human activities on the basin, and on regional and urban flood control projects across the basin.Therefore, 2002 can be considered an important demarcation point in this study or any study of hydrological regime variation in the Taihu Basin.

Research in the Taihu Basin has focused on hydrological simulation(Liu et al.,2013;Wang and Yang,2007),flood risk management (Gong and Lin, 2009), water quality (Ye et al.,2017), and ecological problems (Liu et al., 2018). Research on hydrological response to urbanization has been a topic of wide concern in recent years (Zhang et al., 2018). Zhu et al.(2015a) found that river connectivity in Shanghai degraded significantly and land use changed with more building areas and fewer agricultural areas. Yang et al. (2014) analyzed precipitation differences between cities and suburbs in the Taihu Basin and revealed that the maximum daily precipitation and the number of rainstorm days increased significantly.Deng et al.(2015)investigated temporal and spatial change of river systems in the Taihu Basin, and demonstrated that changes in river density, water surface ratio, and main river area-to-length ratio in the rapid urbanization period were much greater than those in the slow urbanization period.Wang et al. (2016) assessed the contributions of precipitation and human activity to water stage increase in the plain river network region of the Taihu Basin, demonstrating that human activities have played more and more important roles in extreme water stage changes since the late 1980s.

By the end of 2015, the population of Taihu Basin reached 59.97 million and the GDP was 6 688 billion Yuan, representing about 4.4% and 9.9% of the nation's totals, respectively. Due to its rapid social-economic development, the natural hydrological cycle in the Taihu Basin has been converted to an artificially dominated natural-social water cycle,and the Taihu Basin suffers high vulnerability to natural disasters (Zhao and Wen, 2012). Therefore, it is important to evaluate the impacts of climate change and human activities,summarize new characteristics of potential threats to storage and drainage capacity, and propose countermeasures to deal with the challenging situation of hydrological regime variation in the basin.

In this study, both climate change (precipitation) and changes in human activities (water diversion and drainage along the Yangtze River, inflow and outflow of Taihu Lake,urban flood control projects, and urbanization) were considered to identify hydrological response at different spatiotemporal scales. First, the Mann-Kendall test was used to detect the trends of sub-region precipitation of the Taihu Basin, the water stage of Taihu Lake, water diversion and drainage along the Yangtze River, and inflow and outflow of Taihu Lake. Second, possible reasons for Taihu Lake water stage change and increase of net inflow into Taihu Lake were examined after considering the operation of Wuxi urban flood control projects. Third, with the aim of increasing basin and regional drainage ability and tackling high water stage problems in the Jiangnan Canal, optimized regulation schemes of regional and urban projects were determined through hydrodynamic models. These can provide scientific suggestions for further research, such as impacts of urbanization on the hydrological cycle, water resources availability, and sustainable development of water safety in the Taihu Basin.

2. Study area and data

2.1. Study area

The Taihu Basin (30°28′N to 32°15′N, and 119°11′E to 121°53′E) is located in the Yangtze River Delta, with the Yangtze River at the north, the Qiantang River at the south,and the East Sea at the east. The western parts of the Taihu Basin are mountainous areas(Peng et al.,2016).The total area of the Taihu Basin is 36 895 km2, with portions in Jiangsu Province covering 19 399 km2, portions in Zhejiang Province covering 12 095 km2, portions in Shanghai City covering 5 176 km2,and portions in Anhui Province covering 225 km2.The hydrographic network of the Taihu Basin is complicated(Deng et al., 2016), with rivers and lakes accounting for 17%of the total area of the basin. The elevation of low land and polder around the Taihu Lake is no more than 5 m.About 80%of the Taihu Basin is plains, while the remaining 20% is occupied by low hills and mountains. As the third largest freshwater lake in China, with a water area of 2 338 km2,Taihu Lake is a flood storage and regulation center, whose average water depth is 1.89 m and water volume is 4.428 billion m3. The average lake bottom elevation is 1 m above mean sea level. The longest river in the Taihu Basin is the Jiangnan Canal,with a total length of 203 km,flowing through Changzhou, Wuxi, and Suzhou.

The average annual precipitation of the Taihu Basin is 1 218.1 mm, mainly concentrated in the flood season from May to September (Wang et al., 2011a). The normal Meiyu period with intensive precipitation from mid-June to early July often lasts for 25 d, accounting for about 19.8% of the annual precipitation, but large differences of precipitation amount exist in various years. For example, the Meiyu did not appear in 1958 and 1978, whereas the precipitation amounts of the Meiyu period in 1991 and 1999 were much greater than the average precipitation.

2.2. Data

Daily precipitation and evaporation of hydrometric stations,water stage stations of Taihu Lake and Wuxi, and the water diversion and drainage amounts along the Yangtze River were excerpted from theAnnual Hydrological Reports. The inflow and outflow data of Taihu Lake were obtained from the Bureau of Hydrology, Taihu Basin Authority of Ministry of Water Resources.

(1)Precipitation data were from 1986 to 2015.As shown in Fig. 1, based on the topography and water features, the whole basin can be divided into seven sub-regions with 106 hydrometric stations to calculate average precipitation,including the Zhexi (ZX) sub-region with 23 stations, the Huxi (HX) subregion with 19 stations, the Taihu (TH) sub-region with eight stations, the Wuchengxiyu (WCXY) sub-region with 12 stations, the Yangchengdianmao (YCDM) sub-region with 13 stations,the Hangjiahu(HJH)sub-region with 17 stations,and the Pudongpuxi (PDPX) sub-region with 14 stations. The arithmetic mean method was applied in calculating the average precipitation of each sub-region, while the areaweighted mean method was applied in calculating the average precipitation of the Taihu Basin based on sub-region area. In Fig. 1, the green shadow represents polder areas greater than 33.3 km2, and the blue line represents the Jiangnan Canal, flowing through Changzhou, Wuxi, and Suzhou.

Fig. 1. Taihu Basin and its sub-regions.

(2) Water stage data were from 1986 to 2015. The Taihu Lake water stage is the average value of five stations surrounding Taihu Lake: Dapukou, Wangtingtai, Xishan, Xiaomeikou, and Jiapu. The Wuxi suburb water stage station is located in the Jiangnan Canal close to urban flood control projects, while the Wuxi inner station is located in the urban area, which is considered to be protected by the urban flood control projects.Suffering from severe land subsidence caused by extensive groundwater exploitation since the 1980s, the maximum cumulative subsidence was over 1 m for some stations in the Taihu Basin (Wu et al., 2009). However, the water stage in theAnnual Hydrological Reportswas observed based on a stationary datum. The occurrence of land subsidence leads to height datum distortion. Thus, the water level ought to be modified. All the water stage data were based on the Zhenjiang Wusong elevation system and modified before trend analysis. Daily observed flood events with different magnitudes of precipitation and peak feature(single-or multipeak) of Taihu Lake and Wuxi water stage stations were selected to evaluate urbanization effects.

(3) The water diversion data, including inflow and outflow of Taihu Lake, were from 1986 to 2015. The water diversion project from the Yangtze River to the Taihu Basin through the Wangyu River and the supply downstream through the Taipu River were launched in 2002. Water diversion and drainage amount were from 14 sluices along the Yangtze River, including four sluices in HX, four sluices in WCXY, the Wangyu Sluice, and five sluices in YCDM.Inflow and outflow of Taihu Lake were from 23 tour gauging discharge stations, including the Taipu Sluice, the Wangting Sluice, three stations (Changxing, Yangjiabu, and Hangchangqiao) in ZX, six stations (Zhihugang, Wujingang,Yapugang,Wuxiqiao,Chengdonggangqiao,and Dagangqiao)in HX, five stations (Daxuanhe pump station, Dushan Sluice,Meilianghu pump station, Wulihu Sluice, and Yanhu Sluice)in WCXY, four stations (Tongkeng Sluice, Xujiangdaqiao,Guajingkou, and Lianhuqiao) in YCDM, and three stations(Chengbei Sluice, Huanlou Sluice, and Tuanjieqiao) in HJH.All the stations surrounding Taihu Lake are shown in Fig. 2.

(4) Land use data were acquired by digitizing the 1985,2000, and 2010 land cover maps (1:10 000), which were monitored by the interpreted thematic mapper(TM).The land cover maps were classified into four categories: cropland,forest, urban area, and water area. All scenes adopted from Landsat TM by manual interpretation included radiometric and geometric corrections. The land cover map of 2010 was adopted in the hydrodynamic model.

(5)Evaporation and discharge data from 1999,2000,2003,2008, 2009, and 2010 were used in the model. Daily evaporation and discharge data from the Gangkou hydrometric station, Deqing hydrometric station, Fushi Reservoir, and Qingshan Reservoir were used in the Xin'anjiang model.Daily evaporation data from the Hangchangqiao Station and the Taipu Sluice were used in the river network hydrodynamic model.

3. Methods

3.1.Nonparametric Mann-Kendall test for monotonic trend

The Mann-Kendall test is a simple but useful method to identify changes in long-term hydrological series (Mann,1945), especially for complicated basins where it is difficult for mathematical modeling to provide satisfactory results(Guo et al.,2018a).In the Mann-Kendall test,the null hypothesisH0states that there is no trend in the population data set(x1，x2，…，xn)(Guo et al.,2018b).The alternative hypothesisH1of a twosided test is that a monotonic trend exists in the data set. The statisticSis computed as follows:

For sample sizes larger than 10, the statisticsSare approximately normally distributed, with meanE（S） and varianceV（S） as follows:

The standardized statistics (Z) are formulated as follows:

Positive and negativeZvalues indicate upward and downward trends, respectively. A value ofZ1-α/2is determined using a standard normal distribution table.The existing trend is considered to be statistically significant if |Z|＞Z1-α/2.In this study,the significance level α was set to be 0.05 and 0.10 withZ1-α/2values equal to 1.96 and 1.65,respectively.

3.2.Runoff model and hydrodynamic model in Taihu Basin

In order to study the effects of water diversion projects and urban flood control projects on hydraulic structure regulation,a water quantity numerical model for the Taihu Basin was adopted.It was composed of rainfall-runoff simulation and the river network hydrodynamic model. Developed by Hohai University (Guo et al., 2016), the rainfall-runoff simulation provides the lateral inflow and discharge boundary for the river network hydrodynamic model. The water quantity numerical model is composed of six parts:(1)rainfall-runoff simulation,(2) unsteady flow simulation of river networks, (3) generalization of river networks and lakes, (4) generalization and simulation of hydraulic structures, (5) simulation of boundary conditions, and (6) calibration and verification.

Fig. 2. Locations of water stage and discharge stations surrounding Taihu Lake.

(1) Rainfall-runoff simulation: The runoff model includes runoff yield and concentration. Both hilly regions and plain river networks exist in the Taihu Basin, and the traditional Xin'anjiang model(Zhao,1992)was applied to hilly regions of the Zhexi sub-region, whose discharge was offered to plain areas as inputs. Each sub-region in plain areas was classified into four kinds of underlying surfaces: water surface, paddy fields, arid land, and urban construction land. Therefore,different runoff models and confluence processing patterns were adopted based on different characteristics of confluence inside and outside of polder areas. For the water surface,runoff could be obtained using rainfall minus evaporation.For paddy fields, irrigation programs of different growth phases were considered, as well as water demand coefficients, infiltration rate, and irrigation-drainage mode. For arid land, the Xin'anjiang model of three layers of evaporation was used.For urban construction land, runoff could be obtained using net rainfall multiplied by the runoff coefficient.Drainage modulus was considered in the polder area confluence, and a plain region confluence unit hydrograph was used outside of polder areas.

(2) Unsteady flow simulation of river networks: Basic equations of one-dimensional unsteady flow in the aqueduct included the continuous equation and momentum equation(Chen et al.,2016).At each river network node,conservation of mass and momentum was available. The river network was dispersed with the four-point implicit difference method,and basic equations were resolved with the iteration method.

(3)Generalization of river networks and lakes:As shown in Fig.3,1 482 rivers,4 275 cross-sections,and 1 164 nodes were generalized in the hydrodynamic model, including 138 storage nodes and 215 control buildings.The whole generalized area for the Taihu Basin was 36847.8 km2(which is very close to the real basin area), including the plain regions of 28534.4 km2,river networks and lakes of 1 083.3 km2, and hilly regions of 7 230.1 km2.

Fig. 3. Generalized rivers and nodes for hydrodynamic model in Taihu Basin.

(4) Generalization and simulation of hydraulic structures:The hydraulic structures include water gates, ship locks, culverts, and pump stations. The nodes were set at upstream and downstream of hydraulic structures, and discharge was calculated with the weir formula.

(5) Simulation of boundary conditions: The boundary conditions included precipitation, evaporation, and tidal level. For each sub-region, the initial water stage of each cross-section was assigned the same value, while the initial discharge was 0. The boundary condition along the Yangtze River and Hangzhou Bay was the tidal level. For tidal stations with observation data, the unit hydrograph was adopted to calculate integral tidal level based on daily semidiurnal tide. For tidal stations lacking observation data, the tidal level interpolation formulas were adopted through relative distance with observed stations.

(6) Calibration and verification: Due to limitations of the evaporation data, only data from the years 1999, 2000, 2003,2008,2009,and 2010 were collected.In order to keep consistent with typical scenarios of existing basin and regional planning,the data of years 1999,2000,2008,and 2009 were chosen in the calibration period to represent extraordinarily abundant,abundant, normal, and scarce precipitation, respectively. The 2003 and 2010 data used in the verification period are two typical scenarios representing scarce precipitation and normal-topartially abundant precipitation. The Xin'anjiang model parameters that needed to be calibrated for hilly regions of the Zhexi sub-region included four modules: evapotranspiration,runoff generation, runoff separation, and runoff routing. The parameters of hydrodynamic models for plain regions included roughness, the free flow coefficient, and the submerged flow coefficient.The model parameters are listed in Table 1.

Nash-Sutcliffe efficiency (NSE) was used to evaluate model performance. WhenNSEapproaches 1.0, the model simulates the measured data perfectly. WhenNSEis negative,the model is a worse predictor than the measured mean value.The equation forNSEis as follows:

whereZoiandZsiare the observed and simulated water levels at timei,is the mean observed data over the simulation period, andnis the total number of observations. For the calibration period at the daily level in this study,NSEwas 0.98. For the validation period,NSEwas 0.97.

Table 1 Parameters of Xin'anjiang model and river network hydrodynamic model.

Table 2 Land use changes in Taihu Basin from 1985 to 2010.

4. Results

From 1985 to 2010, the most noticeable change in land use was conversion of cropland to urban area. In 1985,cropland occupied the largest proportion of land (64%),followed by forest (14%), water area (13%), and urban area(10%). In 2000, the proportion of cropland decreased to 53%, while the proportion of urban area increased to 19%.In 2010, the proportion of cropland decreased to 48%,whereas the proportion of urban area increased to 24%. The land use change is shown in Table 2. In terms of the change rate, the urban area has undergone the most significant changes, by 151% (from 3 570 km2in 1985 to 8 943 km2in 2010), followed by cropland, with a decreasing rate of 25%(from 23 482 km2in 1985 to 17 675 km2in 2010). The change was more significant from 1985 to 2000. As shown in Fig. 4, small towns expanded around the former urban center, while the transport infrastructure was well developed, causing the urban boundary to blur and city growth to become scattered.

4.1. Trends of precipitation, water stage, diversion,inflow, and outflow

The increasing trends of annual precipitation and precipitation in the flood season (from May to September) over the Taihu Basin were not statistically significant. As shown in Fig.5,monthly basin precipitation decreased significantly at a significance level of 0.05 in April,May,and September,withZvalues of -2.54, -2.33, and -2.82, respectively. The subregion precipitation in the non-flood season (January to March,October to December)demonstrated increasing trends,but only precipitation in January increased significantly at a significance level of 0.05.Monthly sub-region precipitation in April, May, and September decreased significantly at a significance level of 0.1, except in HX.

The increasing trend of the highest annual Taihu Lake water stage(HWS)was not significant,with aZvalue of 1.45.However, theZvalue of the lowest annual Taihu Lake water stage (LWS) reached 5.08, and theZvalue of the average annual Taihu Lake water stage (AWS) reached 2.57, which indicated that LWS and AWS increased significantly at a significance level of 0.05. As shown in Fig. 6(a), the HWS,LWS,and AWS of all months had increasing trends,especially in January, February, March, April, August, and December.Apart from this, the LWS in May, July, September, and November also increased significantly, while the AWS in September and November increased significantly (Fig. 6(b)).Considering the 11 key projects completed and water diversion begun in 2002,the HWS,LWS,and AWS variations were compared before 2001 and after 2002. As shown in Fig. 6(b),from January to April and in December,the HWS increased by more than 0.18 m, the LWS increased by more than 0.19 m,and the AWS increased by more than 0.17 m. The largest increasing extent of the HWS, LWS, and AWS reached more than 0.29 m in March, whereas the least increasing extent reached more than 0.04 m in July.

Fig. 4. Land use changes in Taihu Basin from 1985 to 2010.

Fig. 5. Trend analysis of precipitation over Taihu Basin from 1954 to 2015.

Monthly net inflow into the Taihu Lake were defined as the daily accumulative amount of water diversion minus drainage. As shown in Fig. 7(a) and (b), net inflow from HX into Taihu Lake had significant increasing trends for all months,and inflow from the Wangting Sluice into Taihu Lake had insignificant increasing trends in non-flood seasons except for April. Compared with net inflow before 2001, net inflow from HX after 2002 increased by 120 × 106m3to 402 × 106m3, while the net inflow from the Wangyu Sluice increased by 5 × 106m3to 185 × 106m3. However, net inflow from ZX into Taihu Lake decreased by 36×106m3to 202 × 106m3in the flood season, and decreased by 1×106m3to 76×106m3in the non-flood season except in October.As shown in Fig.7(c)and(d),the net outflow of the Taipu Sluice had decreasing trends in the non-flood season except in March. In April, May, June, July, September,October, and November, the net outflow of YCDM had decreasing trends,whereas that of HJH had increasing trends.Compared with net outflow before 2001, net outflow of HJH after 2002 increased by 32×106m3to 155×106m3,and the Taipu Sluice decreased by 49×106m3to 72×106m3in the non-flood season except in March.The net outflow of YCDM decreased by 11 × 106m3to 121 × 106m3from April to November, and that of WCXY increased by 11 × 106m3to 34 × 106m3in the non-flood season.

The flood movement of the Taihu Basin changed a lot.During the period from 2002 to 2015, total water diversion from the Yangtze River through the Wangyu River reached 272.4 × 108m3, with net inflow into Taihu Lake being 124.2 × 108m3, and water moving downstream through the Taipu River reaching 182.8 × 108m3. Compared with the annual net flow (differences between inflow and outflow)before 2001, the annual net inflow from HX increased by 24.9 × 108m3, the value from the Wangting Sluice increased by 7.1 × 108m3, and the value from ZX decreased by 7.1 × 108m3. However, net outflow of HJH increased by 11.4×108m3,the value of WCXY increased by 1.9×108m3,the Taipu Sluice decreased by 3.0 × 108m3, and the value of YCDM decreased by 3.7 × 108m3.

4.2.Possible reasons for Taihu Lake water stage increase and flood risks in Taihu Basin

As mentioned above,the year of 2002 can be considered an important demarcation point to study hydrological regime variation in the Taihu Basin.In order to eliminate the influence of water diversion on the Taihu Lake water stage, we selected 218 flood events from 1986 to 2015 to study the relationships between the Taihu Lake water stage increase and precipitation under low net water diversion conditions. The Taihu Lake water stage increase has a strong relationship with precipitation,and the correlation coefficient was 0.86 before 2001,but 0.83 after 2002. Similarly, in order to eliminate the influence of precipitation on the Taihu Lake water stage increase, we selected 233 flood events to study the relationships between the Taihu Lake water stage increase and net water diversion from the Yangtze River under low-precipitation conditions.The correlation coefficient was 0.30 before 2001, but 0.39 after 2002.

Fig. 6. Trends and increase of HWS, LWS, and AWS.

Fig. 7. Trends and variation of net inflow and net outflow of different sub-regions.

For each period of water diversion from the Yangtze River into Taihu Lake after 2002, the Taihu Lake water stage demonstrated a very slow decreasing trend, taking five to six days to complete a 0.01-m decrease. However, for the same period before 2001,it took only two to three days to complete a 0.01-m decrease. Therefore, the Taihu Lake water stage decreased much more slowly after 2002, becoming steadier and generally higher after 2002.Therefore,the increase of the Taihu Lake water stage is still dominated by precipitation,while the decrease is affected by water diversion from the Yangtze River after 2002.

For the period before 2001, we selected 116 heavy flood events with an average basin precipitation of 57.8 mm, while the Taihu Lake water stage increased from 3.10 m to 3.22 m on average. For the period after 2002, we selected 39 flood events with an average precipitation of 55.9 mm, while the water stage increased from 3.27 m to 3.38 m on average.Compared with water stage increase for the same precipitation before 2001, shown in Table 3, water stage increased more after 2002 when precipitation exceeded 70 mm, while the water stage increased more significantly when precipitation exceeded 120 mm. For the same 0.01 m of the Taihu Lake water stage increase after 2002,5 mm of average precipitation was needed under the 50-mm precipitation scenario, 4 mm of average precipitation was needed under the 100-mm scenario,and only 3.8 mm of precipitation was needed for the 150-mm scenario.

The Wuxi urban flood control projects were put in use in 2007,and can be used as an example to analyze the effects of human activities on water stage in the WCXY sub-region.The Wuxi urban flood control projects are located in the WCXY sub-region, downstream of HX and upstream of YCDM.Before construction of urban flood control projects, the water stage of Wuxi inner cities was the same as that of suburbs.As shown in Fig. 8, the annual average water stage of the inner city was 0.16-0.40 m lower than that of suburbs in flood seasons after construction. We selected 36 intensive flood events to explore non-stationary relationships between area precipitation and regional average water stage differences in the WCXY sub-region. Before intensive precipitation occurred, the average water stage difference between inner cities and suburbs was 0.24 m. However, after precipitation,the average water stage difference increased to 0.46 m on average. With more intensive precipitation, the water stage of suburbs increased more quickly,and more differences between inner cities and suburbs arose. As shown in Fig. 9(a), with more precipitation, the maximum water stage differences ΔZmaxbetween inner cities and suburbs grew larger.A 10-mm precipitation increase may lead to a 0.054-m ΔZmaxincrease on average. Since an initial water stage difference ΔZintmayhave already existed before intensive precipitation occurred,we also built relationships between precipitation and the real increase of water stage differences. The real increase of water stage differences ΔZrealis defined as ΔZmaxminus ΔZint. As shown in Fig.9(b),a 10-mm precipitation increase may lead to 0.046 m of ΔZreal. Therefore, flooding in inner cities was transferred to suburbs through urban flood control projects,leading to higher and higher water stages in suburbs,which are especially severe in the Jiangnan Canal.Instead of flowing into WCXY, flooding upstream was forced from HX into Taihu Lake, increasing the inflow from HX into Taihu Lake.

Table 3 Taihu Lake water stage increase for different precipitation scenarios before 2001 and after 2002.

Fig. 8. Average water stage differences of Wuxi inner cities and suburbs and precipitation distribution.

The peak lag time of inflow from HX into Taihu Lake was defined as the time duration between precipitation peak and inflow peak. Based on the flood fluctuation process under low net water diversion along the Yangtze River,the peak lag time of HX inflow into Taihu Lake was one to two days before 2001,while the peak lag time decreased to only one day after 2002.For the same inflow into Taihu Lake of 16×106m3,16 mm of precipitation was needed before 2001, while only 11 mm precipitation was needed after 2002. With a shorter peak lag time, more inflow was concentrated in shorter time intervals.Thus,the efficiency of flow concentration increased after 2002.

Possible reasons for Taihu Lake water stage increase can therefore be divided into three categories:

(1) The water diversion increase along the Yangtze River:The HWS, LWS, and AWS of April and May increased,although monthly precipitation of the same period decreased,which may have been due to the water diversion of the Wangyu Sluice from the Yangtze River. The HWS, LWS, and AWS in August and September demonstrated significantly increasing trends, although the precipitation of the same period decreased significantly, which may have been caused by the water diversion of the Wangyu Sluice and HX from the Yangtze River storing water resources in the non-flood season.The HWS, LWS, and AWS in November and December increased significantly, which may have been due to the precipitation increase as well as water diversion of the Wangyu Sluice and WCXY from the Yangtze River in order to satisfy water requirements and improve the regional water environment.

(2) Increase of net inflow into Taihu Lake: The increase of multi-year average inflow was more significant than that of outflow, and it was particularly significant for HX and the Wangting Sluice. Compared with the multi-year average annual net inflow before 2001,the annual net inflow after 2002 increased by 18.1 × 108m3. The increase of net inflow from HX into Taihu Lake may be due to increasing water diverted from the Yangtze River and frequent use of Wuxi urban flood control projects. The Taihu Lake water stage decreased much more slowly after 2002,causing the Taihu Lake water stage to be steadier and higher after 2002. Flooding in Wuxi inner cities was transferred to suburbs through urban flood control projects,leading to a higher water stage in the Jiangnan Canal.Instead of flowing into WCXY, flooding upstream was forced from HX into Taihu Lake,increasing inflow of HX into Taihu Lake.

Fig. 9. Relationships between precipitation and maximum water stage differences and between precipitation and real increase of water stage differences.

(3)Land cover change with more impervious areas:With a significant decrease in cropland and an increase in urban area,both surface runoff and flow velocity from HX into Taihu Lake increased significantly.More inflow from HX into Taihu Lake was concentrated in a shorter time interval and the efficiency of flow concentration increased after 2002, leading to a quicker and higher Taihu Lake water stage increase,as well as a heavy flood control burden for both the basin and regions.

5.Discussion on optimized flood control strategy in Taihu Basin

With the drainage power of urban flood control projects along the Jiangnan Canal(especially in Changzhou,Wuxi,and Suzhou)increasing,more water in inner cities has been moved to suburbs, leading to higher water stages and severe flood risks in the Jiangnan Canal. Therefore, a certain incompatibility has already emerged in projects’ regulation mode. In order to increase basin and regional drainage and tackle high water stage problems in the Jiangnan Canal, the optimized regulation schemes of regional and urban projects were explored through water quantity numerical model computation. The control of annual average water stage increase(ΔZ)was the basis for the optimized regulation scheme.

One basic scheme and six optimized schemes with 1991 design precipitation were investigated in this study. The heavy precipitation in 1991 was caused by the Meiyu, and mainly concentrated in the northern part of the Taihu Basin.The homogenous frequency enlargement method was adopted to calculate the 1991 design precipitation of with a 50-year return period. For basic schemes, the regulation modes of regional and urban flood control projects are currently in use. As shown in Table 4, for urban optimized scheme 1(CO1), the water stage was increased by 50%ΔZon the drainage line of inner cities on the Jiangnan Canal.For urban optimized scheme 2 (CO2), the water stage was increased by 100%ΔZon drainage line of inner cities on the Jiangnan Canal. For regional optimized scheme 1 (RO1) along the Yangtze River,the water stage was reduced by 50%ΔZfor the regional drainage line. For regional optimized scheme 2(RO2), the water stage was reduced by 50%ΔZfor the regional diversion line,while the water stage was reduced by 100%ΔZfor the regional drainage line. The urban and regional optimized scheme 1 (CRO1) was a combination scheme of CO1 and RO1, while the urban and regional optimized scheme 2 (CRO2) was a combination scheme of CO2 and RO2.

As shown in Table 5, variation between basic and optimized schemes for average and highest water stage in the flood season was compared, with HX, WCXY, and YCDM representing regional water stage. The CO1 and CO2 had no clear effects on reducing the average Taihu Lake and regional water stages in the flood season. RO1 decreased the stage by 0.04-0.06 m, RO2 decreased the stage by 0.08-0.13 m, the effect of CRO1 was basically identical to that of RO1,and the effect of CRO2 was basically identical to that of RO2. The water stage decrease in WCXY was more significant,followed by the decreases in HX and YCDM.For reducing the average suburb water stage, RO1 was better than CO1, and RO2 was better than CO2, while CRO1 included comprehensive effects of CO1 and RO1, and CRO2 included the comprehensive effects of CO2 and RO2.The decrease of the suburb water stage was 0.05-0.06 m for RO1, while the decrease was 0.11-0.13 m for RO2, and more significant in Changzhou.The decrease of the suburb water stage was 0.06-0.08 m for CRO1, while the decrease was 0.13-0.18 m for CRO2, and more significant in Suzhou. For CRO1 and CRO2, the suburb water stage decrease was more significant than the city water stage increase.

For reducing the highest Taihu Lake and regional water stage, the combined urban and regional optimized scheme CRO1 (CRO2) was more effective than corresponding regional optimized scheme RO1 (RO2), followed by corresponding urban optimized scheme CO1 (CO2). Changes of water stages in WCXY were more significant than those in HX and Taihu Lake for CRO1 and CRO2. The urban optimized schemes decreased the suburb water stage by 0.01-0.04 m,the regional optimized schemes decreased by 0.03-0.14 m,and urban and regional optimized schemes decreased by 0.06-0.15 m. However, for RO1, RO2, CRO1, and CRO2 in Suzhou, the suburb water stage decrease was more significant than the city water stage increase.

Therefore, research on sustainable development of water safety should set the basin as the basic unit, and orchestrate flood control relationships among basin, region, and city. In order to achieve more harmonious regulation,the principles of a part being subordinate to the whole should be followed.From the aspect of the basin, regulation schemes should bemodified by adoption of more flexible methods based on regional and urban water regime variations. From the aspects of regional and city needs, the optimized regulation scheme should take overall consideration of highlighting flood control priorities and sharing risks, coordinating reasonable drainage lines of inner cities,and diversion and drainage lines along the Yangtze River.

Table 4 Basic and optimized computation schemes.

Table 5 Variations of average and highest water stages of optimized schemes.

6. Conclusions

In this study,precipitation,water stage,net water diversion along the Yangtze river,and inflow and outflow of Taihu Lake using long time series data were analyzed. The optimized regulation schemes of regional and urban projects were explored through a hydrodynamic model. The following conclusions were made:

(1)The HWS,LWS,and AWS of all months had increasing trends.The HWS increased by more than 0.18 m,the LWS by more than 0.19 m, and the AWS by more than 0.17 m in the non-flood season after 2002. Total net inflow into Taihu Lake increased by 18.1×108m3,with HX and the Wangting Sluice increasing significantly after 2002. Monthly HWS, LWS, and AWS increased, although precipitation in the same period decreased, which may be due to the water diversion of the Wangyu Sluice,the WCXY sub-region,and the HX sub-region from the Yangtze River.

(2) The Taihu Lake water stage decreased much more slowly after 2002, becoming steadier and higher after 2002.More inflow was concentrated in a shorter time interval,leading to higher inflow concentration efficiency into Taihu Lake.After construction of Wuxi urban flood control projects,the average water stage of the inner city was 0.16-0.40 m lower than that of suburbs in flood seasons. Flooding in inner cities was transferred to suburbs, leading to a higher water stage in suburbs of the WCXY sub-region. Instead of flowing into the WCXY sub-region, flooding upstream was forced from the HX sub-region into Taihu Lake, increasing HX inflow into Taihu Lake.

(3) Urban and regional optimized schemes were more significant than single regional or urban optimized schemes.All six optimized schemes demonstrated advantages in water stage decrease in WCXY, Taihu Lake, and suburbs along the Jiangnan Canal (Changzhou, Wuxi, and Suzhou), among which regional optimized schemes acquired more satisfying effects of without increasing inner city flood control burden.Regional schemes decreased the average Taihu Lake and regional water stages by 0.04-0.13 m, and decreased the highest water stage by 0.04-0.09 m in the flood season.Future research on sustainable development of water safety should set the basin as the basic unit, and orchestrate flood control relationships between the basin, region, and city. Decreasing diversion and drainage lines along the Yangtze River could take an active role in flood control.

Declaration of competing interest

The authors declare no conflicts of interest.