Impacts of the Three Gorges Reservoir on its immediate downstream hydrological drought regime during 1950–2016

Abstract

Impacts of the Three Gorges Reservoir (TGR) on the downstream hydrological extremes always draw public attentions, particularly during its (quasi) normal water storing period. To investigate the impact of the TGR on its downstream hydrological drought regime during different operation stages, Yichang and Cuntan stations were selected as the case study site with Cuntan station as a reference. On the basis of the recent updated long-term flow discharge data series from 1950 to 2016, drought occurrence, duration and frequency of the hydrological droughts downstream of the TGR during pre- and post-dam periods were analyzed and compared, respectively. Abrupt changes were also identified for hydrological drought at different timescales. Results showed that TGR’s operation altered its downstream seasonal variation patterns of dry/wet spells, i.e., more severe drought from middle summer to middle winter while much wetter condition in late winter and spring; impacts on the downstream hydrological drought intensify with upgrading reservoir water storing levels; significant abrupt changes of hydrological droughts were identified, and the abrupt changing years were greatly correlated with the impoundment stages of the TGR. The study could provide valuable references for the ecological security and integrated water resources management of the Yangtze River basin and other large rivers severely impacted by anthropogenic activities over the world.

1 Introduction

Hydrological drought is associated with the effects of periods of precipitation (including snowfall) shortfalls on surface or subsurface water supply, i.e., streamflow, reservoir and lake levels, groundwater (Heim 2002; Keyantash and Dracup 2002). The frequency and severity of hydrological drought are often defined on a watershed or river basin scale. The link between hydrological drought and human activities is an essential field research that is highlighted by the International Association of Hydrological Science in the new scientific decade of ‘Panata Rhei’ (Montanari et al. 2013). Thus, it is of vital significance in exploring the long-term evolution patterns of hydrological drought which might be impacted by large-scale key infrastructures by human kind. Reservoirs, built for water supply, navigation, hydropower generation, flood control and drought relieves, destroy river continuum into pieces, and always induce alterations in river hydrological and hydraulic regimes as a result of operational conflict between the main purpose of the reservoirs and their functions in hazard mitigation or environmental protection (López-Moreno et al. 2009).

The Three Gorges Reservoir (TGR), located over the Yangtze River main stream, is the largest hydropower project over the world. The TGR region has a subtropical monsoon humid climate, with an annual average temperature of 18.9 °C (2013) and an annual precipitation between 817.1 and 1360.5 mm (Gao et al. 2016). The TGR management practices and the impact on its downstream hydrological extremes, river ecological health and climate have been investigated continually since its initial impoundment in June 2003 (Wu et al. 2004; Dai et al. 2008; Li et al. 2011, 2012a, 2014b; Sun et al. 2012; Chen et al. 2016). During the past decade, severe and persistent drought occurred in 2004, 2006, 2007 and 2011, respectively, in the middle and downstream of the Yangtze River basin after the TGR period, which exerted detrimental impacts on local economics, society and ecological system (Zhang et al. 2016; Guan et al. 2017; Shan et al. 2017). For the drought from October in 2010 to June in 2011, the severe autumn–winter–spring persistent drought hit the middle and the downstream of the Yangtze River basin widely including Hubei, Hunan, Jiangxi, Anhui and Jiangsu provinces, causing severe shrink of lake water area and ecological deteriorations (Wang and Duan 2012; Chen et al. 2014; Wang et al. 2016). The issue of whether to build dam or not in the Poyang Lake once again put the TGR under highly public attentions. Recently, the flow discharge entering the Poyang Lake from the Yangtze River reduced significantly and resulted in water level falling during dry season, what’s more, the dry season advanced with prolonged duration which coincided with the normal water storing period of the TGR (Tang et al. 2016; Liu et al. 2017a, b; Zhang et al. 2017).

Therefore, drought management has become an unavoidable key issue in the risk-based integrated reservoir operation of the TGR for its downstream, particularly during period of water scarcity. Wan et al. (2017) investigated the impacts of local water extraction and reservoir regulation on hydrological drought in the USA. Zhang et al. (2015) explored the evolution of hydrological drought under the regulation of two reservoirs in the Huaihe River. Wu et al. (2016) studied the response of hydrological drought to meteorological drought under the influence of Sanmenxia Reservoir in the Yellow River. Li et al. (2012b) investigated the effects of the TGR on the hydrological droughts at the downstream Yichang station during 2003–2011 by comparing the observed runoff series with the naturalized one which was simulated based on a water balance model, and demonstrated that the hydrological droughts at Yichang are slightly aggravated by the TGR’s initial operation. To the best of our knowledge, however, limited attention was focused on the impacts of the TGR’s refilling operation during normal stage on the hydrological drought spells in the middle and downstream of the Yangtze River basin. Considering the TGR is currently in its normal operation phase, an assessment of long-term effects of the TGR on its downstream hydrological droughts is of critical importance and necessary based on more recent observed data series.

The aim of the study is to provide a complex investigation into the impact of the TGR’s refilling operation on its immediate downstream hydrological drought regime with the most updated hydrological data series (1950–2016). In the study, the Standardized Streamflow Index (SSI) (Vicente-Serrano et al. 2011) was adopted to evaluate the hydrological drought in the upper and downstream of the TGR since it was restrictedly mathematical demonstrated. The SSI was widely used over the world in recent years due to its low data requirements and simplicity comparable in both time and space (Lorenzo-Lacruz et al. 2013; Li et al. 2014a; WMO and GWP 2016). It should be noted that the hydrological drought here is referred to streamflow drought. The paper is organized as follows. The study region, data and method are presented in Sect. 2. The results are provided in Sect. 3. Section 4 is the discussion, and Sect. 5 is the conclusions. The study could provide valuable references for the ecological security and the integrated water resources management of the Yangtze River basin and other large rivers which are severely impacted by anthropogenic activities over the world.

2 Study region, data and method

2.1 Study region

The TGR, built for flood control, navigation and hydropower generation, is located on the main stream 44 km above the end of the upper Yangtze River. The TGR operation experienced four stages: The initial stage started in June 2003, with the reservoir water level being raised to 135 m above sea level; the transitional stage began in October 2006 by increasing the water level up to 156 m; the reservoir water level rose to 173 m in November 2008 during the quasi-normal stage and then to the designed reservoir water level of 175 m in October 2010 during the normal stage (Yang et al. 2014). According to the current refill rules of TGR proved by the Ministry of Water Resources (Li et al. 2014b; MWR 2009), the TGR stores 221.5 × 10⁸ m³ water to reach normal water level of 175 m from 145 m in the end of flood season (generally from September 15 to October 31, lasting 47 days; it begins refilling from September 10 provided the year is relatively dry). During the refill operation period, the mean outflow discharge will be reduced by 5460 m³/s. To assess the impacts of the TGR on its immediate downstream hydrological drought regime, two key hydrological stations of Cuntan and Yichang were selected as case study sites (Fig. 1). Cuntan station, the inflow control point of the upper Yangtze River into the TGR, is located 608 km above the TGR. Yichang station, an important monitoring station for the outflow from the TGR, is the control point of the upper Yangtze River and is located at the start of the middle Yangtze River, immediately downstream of the TGR without influence from large tributaries in the middle and lower reaches of the Yangtze River, approximately 44 km below the TGR.

Fig. 1

Sketch map of the Yangtze River basin

2.2 Data collection and processing

The daily time series of flow discharge (m³/s) from 1950 to 2016 at Yichang and Cuntan stations were collected from the Changjiang Water Resources Commission (CWRC), China. The homogeneity and reliability of the flow discharge data have been checked and firmly controlled by the CWRC before its release.

Based on the facts of about 86% runoff at Yichang coming from Cuntan and the alteration of flow regime being limited induced by human activities in the region between Cuntan and Yichang, Cuntan station just above the TGR was selected as a reference station to investigate the impacts of the TGR on its downstream hydrological drought regime in the Yangtze River basin. According to the daily flow discharge, the monthly flow discharge was obtained by averaging the daily ones. We carried out our analysis by using monthly data due to the characteristics of hydrological droughts. Hydrological drought episodes are recognized to be difficult to pinpoint in time and space, the effects grow slowly in magnitude, and their impact can accumulate over long periods of time (Lorenzo-Lacruz et al. 2013). For these reasons, we considered monthly data more suitable for drought characterization than daily data. Based on the monthly flow discharge, the monthly SSI at Yichang and Cuntan was calculated to assess the hydrological drought evolution alterations in the downstream of the TGR. Since the Gezhouba Reservoir—the first biggest reservoir built on the Yangtze River was operated in 1981, and Yu et al. (2013) disclosed that it has influences on its downstream monthly and daily flow regime, to investigate the impact of the TGR on its downstream hydrological drought, the study period was divided into two periods, i.e., 1981–2002 for the pre-TGR and 2003–2016 for the post-TGR, respectively. Variation patterns of the hydrological drought in the upper and lower TGR basin were carried out. Drought occurrence and drought duration during pre-dam and post-dam period were also analyzed thereafter. Drought occurrence (duration) ratio is defined as the drought occurring times (duration) during post-TGR being divided by that during pre-TGR. On the basis of the computed frequency of the monthly SSI falling within different drought severity ranges, drought frequency changing patterns for dry and wet spells over different sub-periods at Yichang and Cuntan were explored, respectively. Furthermore, abrupt changes for hydrological drought and flow discharge at both annual and monthly scales during 1950–2016 were identified by adopting the sequential version of Mann–Kendall test (Sneyers 1975).

2.3 Sequential version of Mann–Kendall test

Here is a brief introduction of the sequential version of Mann–Kendall test (SQMK). It is used to test assumptions about the start of a trend within the sample $X_{1} , \ldots ,X_{n}$ from set of random variable $X$ based on rank series of progressive and retrograde rows of the sample. The magnitudes of $X_{j}$ annual mean time series and $j = 1, \ldots ,n$ are compared by $X_{k}$, where $k = 1, \ldots ,j - 1$. For each comparison, the number of cases $X_{k} > X_{j}$ is counted and denoted by $n_{j}$. The test statistic $t_{j} = \mathop \sum_{1}^{j} n_{j} n\ \left( {j = 2,3, \ldots ,n} \right)$ is normally distributed with mean given as

$$ E\left( t \right) = \frac{{n\left( {n - 1} \right)}}{4} $$

(1)

and variance as

$$ {\text{Var}}\left( {t_{j} } \right) = \frac{{\left[ {j\left( {j - 1} \right)\left( {2j + 5} \right)} \right]}}{72} $$

(2)

The sequential values of the statistic $U\left( t \right)$ are calculated as

$$ {\text{UF}}\left( t \right) = \frac{{t_{j} - E\left( t \right)}}{{\sqrt {{\text{Var}}\left( {t_{j} } \right)} }} $$

(3)

which is the forward sequence and the backward sequence ${\text{UB}}\left( t \right)$ is calculated using the same equation but in the reverse data series. In two-sided trend test, a null hypothesis is accepted at a significance level if $\left| {{\text{UF}}\left( t \right)} \right| \le {\text{UF}}\left( t \right)_{1 - \alpha /2}$, where ${\text{UF}}\left( t \right)_{1 - \alpha /2}$ is the critical value of the standard normal distribution with a probability exceeding $\alpha /2$. A positive ${\text{UF}}\left( t \right)$ denotes an upward trend, while the reverse denotes a downward trend (i.e., ${\text{UB}}\left( t \right)$ is similar to ${\text{UF}}\left( t \right)$). In this study, $\alpha$ is set at 0.05 significant level. The sequential Mann–Kendall test enables detection of the approximate time of occurrence of a trend from the intersection point of the forward and backward curves of the test statistic. If the intersection point is significant at $\alpha = 0.05$, then the critical point of change is at that period (Moraes et al. 1998; Gerstengarbe and Werner 1999). Hence, the sequential Mann–Kendall test is an efficient way by which the starting time of a trend is pinpointed.

The SQMK test is selected to detect the potential changing points of hydrological droughts and flow discharge. Based on the generated progressive and the retrograde time series of the SSI and flow discharge difference between Yichang and Cuntan ($\Delta {\text{SSI}},\Delta Q$), we can conclude that the intersection of curves ${\text{UF}}$ and ${\text{UB}}$ localizes the changing timing of hydrological drought and flow discharge, provided this intersection is situated between the critical values of the 5% level of significance (Goossens and Berger 1987). Moreover, the existence of a drying (wetting) trend may also be put forward if the intersection of the two curves ${\text{UF}}$ and ${\text{UB}}$ is localized above the critical value of the 5% level of significance.

2.4 Standardized Streamflow Index

The Standardized Streamflow Index (SSI) developed by Vicente-Serrano et al. (2011) was adopted to assess the hydrological drought condition at Yichang and Cuntan stations. Here, Pearson III distribution was chosen to fit the streamflow series since it has also been illustrated the best-fitting distribution for the discharge series in China. The probability density function of the Pearson III distributed variable is written as

$$ f\left( x \right) = \frac{1}{{\beta^{\alpha } \varGamma \left( \alpha \right)}}\left( {x - \gamma } \right)^{\alpha - 1} e^{{ - \left( {x - \gamma } \right)/\beta }} $$

(4)

where $\alpha , \beta , \gamma$ are shape, scale and location parameters for discharge values $Q$ > 0, respectively. Γ(a) is the gamma function of $\alpha$. The parameters were estimated using the L-moment method (Hosking 1990).

The cumulative probability distribution function of the Pearson III distributed variable is thus written as

$$ F\left( x \right) = \frac{1}{{\beta^{\alpha } \varGamma \left( \alpha \right)}}\mathop \int \limits_{\gamma }^{x} \left( {x - \gamma } \right)^{\alpha - 1} e^{{ - \left( {x - \gamma } \right)/\beta }} $$

(5)

Following the approach formulated by Abramowitz and Stegun (1965), the SSI is finally obtained as

$$ {\text{SSI}} = W - \frac{{C_{0} + C_{1} W + C_{2} W^{2} }}{{1 + d_{1} W + d_{2} W^{2} + d_{3} W^{3} }} $$

(6)

$$ W = \sqrt { - 2\ln \left( P \right)} ,\quad P \le 0.5 $$

(7)

It should be noted that the hydrological droughts referred in the study affecting flow discharge are streamflow droughts. The hydrological drought begins when the SSI is less than − 0.5, and ends when the SSI is more than 0.5. The drought severity classification of the SSI is shown in Table 1.

Table 1 Drought severity classification of the SSI

3 Results

3.1 Temporal variation of the hydrological droughts

Figure 2 shows the temporal variations of the SSI in the upstream and downstream of the TGR in the Yangtze River basin. Comparison of the two hydrological series demonstrated that the variation patterns of dry/wet conditions in the downstream of the TGR were generally correspondent to those in the upstream of the TGR with the Spearman rank correlation coefficient being of 0.87. For the extremely drought events, they were both recorded in 1960, 1970, 1979, 1994–1995, 2006 and 2011 at Yichang and Cuntan, respectively. Other similarities in the hydrological drought at these two stations are also apparent for the rest drought periods. From Fig. 3, it can be found that generally the green area increased continuously after 2003 for both late winter and whole spring, while the red area increased significantly for both late summer and whole autumn at both Yichang and Cuntan. It demonstrated that the dry conditions in the TGR area (between Cuntan and Yichang) were much alleviated during late winter and whole spring, whereas dry conditions became much more serious during late summer and whole autumn after 2003. Nevertheless, it cannot be attributed to the TGR’s refill operations since it occurred at Yichang and Cuntan simultaneously.

Fig. 2

Temporal variations of the SSI at Yichang and Cuntan stations

Fig. 3

Heatmap of the monthly SSI at Yichang and Cuntan stations. The blue frame marks the post-TGR period

Despite the similarities listed above between Yichang and Cuntan, differences are also obvious in terms of drought severity for most drought events. From Fig. 3, lighter red grids were found at Yichang during a drought than those at Cuntan before 2003; in contrast, these lighter red grids changed to darker red ones at Yichang than those at Cuntan after 2003. This indicated that the hydrological droughts at Yichang after TGR were generally more severe than the droughts before 2003 in comparison with those at Cuntan. Moreover, the grids at Yichang in October being totally covered by red after TGR disclosed that the hydrological drought at Yichang aggregated the most in October by the TGR’s water storing to the high water level of 175 m. Boxplot of the monthly $\Delta {\text{SSI}}$ during post-dam and pre-dam periods also verified the conclusion above (Fig. 4). Most significant $\Delta {\text{SSI}}$ decrease was detected in October with the main box of $\Delta {\text{SSI}}$ during post-dam period totally below the box during pre-dam period. Remarkable reduction in $\Delta {\text{SSI}}$ was also identified in September, November, August, July, December and January, while obvious increases in $\Delta {\text{SSI}}$ were found in February, March, May and June over post-dam period. Two-sample t test also revealed that the changes of $\Delta {\text{SSI}}$ were statistically significant at 95% confidence level in May, July, September and October during post-TGR period with the p values being of 0.006, 0.047, 0.011 and 0.000, respectively.

Fig. 4

Boxplots of the monthly SSI difference between Yichang and Cuntan during pre- and post-TGR periods. The upper and lower parts of the boxes represent the 75th and 25th percentiles, respectively; the whiskers indicate the 90th and 10th percentiles; and the lines within the boxes are the median values

3.2 Impacts of the TGR on the drought frequency and duration

The drought occurrence, drought duration and drought frequency distribution of moderate drought, severe drought and extremely drought during pre-TGR period and post-TGR period are presented in Tables 2 and 3, respectively.

Table 2 Drought occurrence and duration statistics during pre-TGR and post-TGR periods

Table 3 Drought frequency statistics during pre-TGR and post-TGR periods

For the extremely hydrological drought during post-TGR period, the drought occurrence at Yichang decreased, while the drought duration increased. For the severe hydrological drought, both the drought occurrence and duration at Yichang increased remarkably. For the moderate hydrological drought, both the drought occurrence and duration showed a reduction at Yichang.

The dry/wet spell frequency distributions at Yichang and Cuntan during pre-TGR and post-TGR were also investigated (Table 3). Extremely drought percentage at Yichang rose from 1.11% during pre-TGR to 1.13% during post-TGR with the ratio of 1.02, while that at Cuntan reduced from 2.26 to 1.86% with the ratio of 0.82. The severe drought percentage at Yichang even increased much more than the extremely droughts, i.e., the percentage at Yichang augmented from 4.36 to 9.57% with the ratio of 2.19, while the percentage at Cuntan declined from 7.17 to 3.67% with the ratio of 1.95. The moderate drought percentage at Yichang also increased from 7.65 to 7.84% with the ratio of 1.02, while that at Cuntan decreased from 10.57 to 9.85% with the ratio of 0.93. In contrast, the percentages of moderate wetness, severe wetness and extremely wetness at Yichang dropped in comparison with that at Cuntan.

3.3 Detecting abrupt changes of hydrological droughts

3.3.1 Alterations in annual and monthly ΔSSI

Statistics of the annual/monthly ΔSSI and $\Delta Q$ during 1950–2016 were carried out by the SQMK test. To save space, related figures can be referred to Appendix. Figure 5 shows the MK statistical values of annual $\Delta {\text{SSI}}$ and $\Delta Q$. A clear intersection of curves ${\text{UF}}$ and ${\text{UB}}$ between the 5% levels of significance was observed which localized the abrupt change of the annual $\Delta {\text{SSI}}$ around 2009–2010, whereas significant change was detected around 2003–2004 in annual $\Delta Q$.

Fig. 5

MK statistics of annual $\Delta {\text{SSI}}$ and $\Delta Q$. The dashes lines indicate the value of the 0.05 significance levels

For the monthly $\Delta {\text{SSI}}$, abrupt changes that are significant at 95% confidence level were identified around 2008–2009 in January and May, 2009–2010 in July, 2005–2006 in October and 2010–2011 in November, respectively, while not significant in September which was around 2008–2009. Comparatively, abrupt changes of monthly flow discharge difference significant at 95% confidence level were observed between 2009 and 2010 in January and July, 2004–2005 in October, while 2007–2008 in September but not significant. Although the abrupt changes of the $\Delta {\text{SSI}}$ and the $\Delta Q$ in July both occurred between 2009 and 2010, the changing point of $\Delta Q$ was somewhat earlier than that of $\Delta {\text{SSI}}$, with the former changing point in early 2009–2010 while the latter in late 2009–2010. The fact that the abrupt change years of $\Delta Q$ occurred always earlier than those of $\Delta {\text{SSI}}$ in January, July, September and October suggested the flow being more sensitive than the SSI in response to human disturbance, particularly the TGR operation. In general, the changing points can be divided into four periods, i.e., 2003–2006, 2006–2008, 2008–2010 and after 2010, respectively, which were consistent with the initial stage, transitional stage, quasi-normal stage and normal stage of the TGR, respectively. It can be concluded that refill operation of the TGR has affected the downstream hydrological regime, and the impacts varied with operational schemes, i.e., the higher the reservoir water storing level, the stronger the influences.

3.3.2 Alterations of hydrological droughts in September and October

To better understand the effect of the TGR on its downstream hydrological droughts, behaviors of the SSI during September and October were further investigated since the TGR started storing water from middle September to late October or very early November (Fig. 6).

Fig. 6

Temporal variations of the monthly SSI and discharge difference in September and October. a SSI in September, b flow discharge difference in September, c SSI in October, d flow discharge difference in October

Before 2009, the SSI at Yichang (0.11) was generally higher than that at Cuntan (− 0.02) in September. After 2009, however, the SSI at Yichang (− 0.80) dropped and fell below that at Cuntan (− 0.47) generally, and the Yichang–Cuntan discharge average difference also decreased dramatically to − 304 m³/s during 2009–2016 from 3497 m³/s during 1950–2008. More significant reduction in Yichang–Cuntan differences of SSI and discharge can be found in October. The SSI at Yichang was higher than that at Cuntan generally before 2005; however, the SSI both at Yichang and Cuntan fell below X-axis after 2006 and the SSI at Yichang even fell below Cuntan thereafter (except 2011 due to the extremely drought in the upper TGR basin). Correspondingly, the mean annual SSI after 2005 at Yichang (− 1.34) and Cuntan (− 0.90) descended sharply from values before 2005 with particularly larger decrease at Yichang. Meanwhile, the related mean annual discharge difference also dropped greatly from 3451 to only 1050 m³/s over the period.

4 Discussion

Operating scheme of ‘storing clear water and releasing sediment’ adopted by the TGR has resulted remarkable effects on the downstream intra-annual distribution of wet/dry conditions, drought duration and drought frequency. The pronounced increase in seasonal drought severity and risk in middle summer, autumn and early-middle winter is probably due to the water storing from September to the end of wet season. The hydrological drought upgraded particularly significant in September and October since the TGR came into refill operation. The impact by the TGR was even worse since October is located in late wet seasons. In contrast, the drought condition in late winter and spring alleviated because of the larger outflow than normal to the downstream from the TGR. Although the TGR is in its normal operational period, it still should be noted that the current dominated drought would probably propagate to early summer and autumn drought since the TGR would advance water storing in early September if it were a dry year. Multi-year persistent drought might also occur but not frequently if the Yangtze River basin confronting a long period of severe water scarcity since the TGR is only a seasonal regulated reservoir. The abrupt changing points being consistent with different operating stages of the TGR imply that the impact by reservoir is significant positively correlated with the reservoir water storing level. The majority abrupt changes falling within the (quasi) normal stage further confirm that storing water level to 175 m or near 175 m exerts the most significant effects of the TGR on the downstream hydrological droughts.

More severe hydrological droughts are prone to occur in the downstream of the TGR with longer drought duration and higher drought occurrence due to the high and persistent reservoir water storing level. The drought frequency distribution was also altered by expanding the percentages of moderate drought, severe drought and extremely drought while lessening the proportion of moderate wetness, severe wetness and extremely wetness in the downstream of the TGR. The more frequent and severe hydrological droughts emphasize the importance and urgency of the drought risk management during the TGR’s integrated operation. Effective early drought monitoring and forecasting will greatly minimize social and economic losses by drought events. Generally, earlier period of abrupt change point for flow discharge than SSI suggested the flow discharge is capable of providing a certain time period (from several months to one year) advancing for the future hydrological drought forecasting. Such forecasting period will be of great significance for the hydrological drought downstream of the TGR since the hydrological drought forecasting at present mostly focused on the seasonal scale. It provides a possible way to predict hydrological drought in regions strongly affected by anthropogenic activities, particularly for dam construction.

Although the discharge decrease is not sufficient to cause problems for navigation or water supply in the Yangtze River (Chen et al. 2016), the environmental flow satisfying degree downstream the TGR might be weakened or even insufficient during post-dam period. Since more and more large-scale reservoirs and water diversion projects are still built and operated in the Yangtze River, the ecological water requirement downstream should be of highly cautiousness, particularly during water scarcity periods. It was reported that the outflow of the Danjiangkou Reservoir for the Hanjiang River was severely lower than planned ecological flow values in 2015 because of the drought event. More importantly, the Chinese sturgeon spawning is closely associated with the flow characteristics and just takes place in October and November downstream the Gezhouba Reservoir and the TGR (Guo 2008; Yi et al. 2016; Song et al. 2018). Although influencing factors of fish spawning and habitat are complicated, their related ecological flow should be firstly guaranteed for the target river. The substantial increase in the intensity and magnitude of droughts downstream of the TGR might also lead unfavorable impacts on the wetland ecosystem such as plants’ seed germination, seeding development, species composition and their spatial distribution, and availability of forging sites for migratory birds (Wu and Liu 2015). To minimize the negative influences induced by the hydrological drought on the downstream ecological system of the TGR during dry years, drought management should be considered in the optimal reservoir operation of the TGR, and the strategies of drought management should be adapted to ecological system health of the Yangtze River.

5 Conclusions

The main purpose of the study was to investigate the impacts of the TGR on the downstream hydrological drought regime. Based on the long-term and most recent flow discharge data series, temporal variation patterns of hydrological drought spells, drought frequency and duration in downstream and upstream during pre- and post-TGR periods were examined, and abrupt changes of the hydrological series were also conducted. Main conclusions are summarized as follows: (1) TGR’s impoundment and operation altered its downstream seasonal distribution of dry conditions in the Yangtze River basin. The hydrological drought severity and risk increased in middle summer, autumn and early-middle winter, while decreased in late winter and spring; (2) two-sample t test disclosed that the SSI difference between Yichang and Cuntan was statistically significant in May, July, September and October. The hydrological drought upgraded particularly significant in September and October with the most severe condition occurring in October. The ecological water requirement downstream of the reservoirs should be of highly cautiousness since the Chinese sturgeon spawning mainly taking place in October and November just downstream of the reservoirs; (3) significant abrupt change of the SSI difference was also identified in January, May, July, September, October and November, and the changing years were highly in line with different stages of the TGR.

Appendix

See Fig. 7.

Fig. 7

MK statistics of monthly SSI difference between Yichang and Cuntan