Hydrological drought in two largest river-connecting lakes in the middle reaches of the Yangtze River, China

Drought is among the most common and serious natural disasters that occur worldwide (Wong et al. 2013; Chen & Sun 2015). With a global warming background, the frequency and intensity of drought events are increasing significantly, which has become one of the important factors restricting sustainable economic development and social and ecological environments (Vandenberghe et al. 2011; Dai 2013; She et al. 2013; Song et al. 2021). The Yangtze River Basin is one of the areas in China with the most serious drought occurrences due to its complex climatic conditions and a large spatiotemporal variation of precipitation (Dai et al. 2008; Zhang et al. 2018, 2019; Sun et al. 2019; Jiao et al. 2020; Li et al. 2020). The rapid economic development and the construction of large water conservancy and hydropower projects in the Yangtze River basin further aggravate the drought problem in the region (Guo et al. 2012; Sun et al. 2012; Wang & Yuan 2018; Ma et al. 2019). Statistically, more than 30 hydrological drought events occurred in the Yangtze River basin from 1979 to 2012, and the drought in the middle and lower reaches was more serious than that in the upper reaches (Zhang et al. 2019).

Poyang and Dongting Lakes, located in the south bank of the middle reaches of the Yangtze River, are the two largest freshwater lakes in China. Their huge volumes and vast water areas play an important role in water resource supply, ecosystem stability, and industrial and agricultural production (Zhang et al. 2022). However, these two lakes have been under threat of hydrological drought in recent years. In particular, the construction and operation of the Three Gorges Reservoir (TGR) in 2003 changed the hydrological regime of the Yangtze River, and this change greatly influenced the hydrological connections and water balances between lakes and the Yangtze River and increased the seriousness of the drought problem (Li et al. 2013; Yu et al. 2019). Frequent hydrological drought events have seriously affected local industrial and agricultural production, ecological security, and even threatened people's lives and health.

Numerous studies have analyzed and investigated hydrological drought in the Poyang Lake basin and the Dongting Lake basin (Sun et al. 2012; Liu et al. 2013; Zhang et al. 2014, 2016, 2019; Fan et al. 2017; Li et al. 2018; Yu et al. 2019; Ye et al. 2020). For example, Xue et al. (2014) analyzed the temporal and spatial characteristics of drought in the Dongting Lake basin based on the improved comprehensive drought index and wavelet theory and found that the autumn drought was the most serious at the seasonal scale. Huang et al. (2014) analyzed the detection, characteristics and challenges of hydrological drought at Dongting Lake in association with the TGR. Chen et al. (2013) identified the characteristic variables of hydrological drought based on run theory and analyzed the joint probability of the duration and intensity of hydrological drought in the Poyang Lake basin. Zhang et al. (2017) further constructed a hydrological drought index based on river runoff and lake water level and analyzed the joint probability of hydrological drought in Poyang Lake, the catchment and the Yangtze River using the copula function. In terms of the formation mechanism of hydrological drought in Poyang and Dongting Lakes, more scholars have focused on the TGR impact on the lake water level, water surface area, water resources and water exchange capacity with the Yangtze River (Li et al. 2013; Yu et al. 2019). For example, Yuan et al. (2014) used MODIS image data to extract the water surface area of Dongting Lake and found that the lake water area showed a shrinking trend following the operation of the TGR, and the impoundment behavior of the TGR in late summer and early autumn aggravated the continuous drought in Dongting Lake. Li et al. (2013) analyzed the evolution characteristics of water exchange capacity between Dongting Lake and the Yangtze River at different time scales. Their results showed that the spill division ability of the three outlets of the Yangtze River weakened after the operation of the TGR, and the amount of water entering the lake was reduced. Similarly, many scholars have also studied the impact of TGR on droughts in Poyang Lake. Zhang et al. (2014) investigated the role of the inflows from the Poyang Lake basin and the Yangtze River discharges on lake hydrology based on a hydrodynamic model. The hydrodynamic simulations showed that the changes in the hydrological regime of the Yangtze River caused by the TGR had a greater impact on seasonal drought in Poyang Lake, especially during autumn. Lai et al. (2014) established a hydrodynamic model for the middle reaches of the Yangtze River, and their simulations showed that the impact of the TGR on the Yangtze River discharge was highly spatially uneven. In autumn, impoundment accelerated the outflow from Poyang Lake and thus significantly decreased the lake water level. Other scholars have analyzed the impact of climate change and water cycle on hydrological drought in the middle reaches of the Yangtze River (Zhang et al. 2015; Li et al. 2020; Liu et al. 2020). The results showed that the decrease in precipitation and meteorological drought in the Dongting Lake basin were the main driving factors causing the decrease in runoff and the decline in lake water level at the annual scale (Cheng et al. 2016; Li et al. 2020). Similar results were found for the Poyang Lake basin (Liu et al. 2013; Ye et al. 2013; Zhang et al. 2014). Overall, these previous studies have promoted the understanding of hydrological drought in the middle reaches of the Yangtze River and the impact of the TGR.

However, due to the different distances from the ocean, the influences of atmospheric circulation on the Poyang and Dongting Lake basins and their spatial and temporal distributions of precipitation are varied. More importantly, Dongting Lake is approximately 300 km downstream from the TGR and 500 km upstream from Poyang Lake, and changes in the water regime caused by the TGR gradually attenuate from Yichang to Datong, resulting in very different strengths of river–lake interactions between the Yangtze River and Poyang and Dongting Lakes (Guo et al. 2012). Previous studies have focused little on comparing hydrological droughts in Poyang Lake and Dongting Lake and have not analyzed the differences in the impacts of the Yangtze River and the local catchment on hydrological drought events in Poyang and Dongting Lakes. This lack of research is not conducive to mitigating drought or providing early drought warnings in the middle reaches of the Yangtze River. Therefore, the objectives of the study are (1) to analyze the characteristics of hydrological drought in Poyang and Dongting Lakes during the period of 1964–2016, including the annual lowest water level, the average lake level in the dry season, the start date and duration of the dry season, and the intensity of hydrological drought events; (2) to investigate and compare the joint probability of concurrent hydrological drought in the lake–local catchment–Yangtze River system during different seasons based on the copula function and (3) to explore the relationship between the large-scale climate indices and hydrological drought in the two lakes. The results of this study can provide useful references and deep insights into hydrological droughts in the middle reaches of the Yangtze River.

The data used in this study mainly include hydrological data for Poyang Lake, Dongting Lake and their basins, as well as the discharges of the main stream of the middle reaches of the Yangtze River. The observed water level at Hukou, Xingzi, Duchang and Kangshan stations from 1964 to 2016 was selected to represent the water level changes of Poyang Lake, and the monthly runoff data from Waizhou, Lijiadu, Meigang, Hushan, Dufengkeng and Wanjiabu hydrological stations in the Poyang Lake basin were selected to reflect the water inflow into the lake. The water level records at Nanzui, Xiaohezui and Chenglingji stations during the same period were selected to describe the changes in the water level of Dongting Lake, and the runoff data at Taojiang, Taoyuan and Xiangtan hydrological stations in the Dongting Lake basin were used to reflect the changes in water inflow into Dongting Lake. The discharges of the Yangtze River at Yichang and Hankou stations were used to quantify hydrological drought in the main stream of the Yangtze River. These data were collected from the Hydrology Bureau of Changjiang Water Resources Commission, China, and the locations of these stations are shown in Figure 1. Moreover, these data have been widely used in various hydrological studies (Ye et al. 2013; Li et al. 2014, 2015, 2017a, 2017b) and their qualities have been tested and reliable. Additionally, regional climate indices, including the Southern Oscillation Index (SOI), the Arctic Oscillation (AO) index, the North Pacific Decadal Oscillation (PDO) and the El Niño 3.4 region index (NINO3.4), were collected from the National Centers for Environment Information (NCEI), National Oceanic and Atmospheric Administration (NOAA) of the United States (https://www.ncei.noaa.gov/access/monitoring/products) to analyze the causes of hydrological drought events.

In this study, the time-varying standardized water level/runoff index (SWIt/SRIt) was used to identify the characteristics of hydrological drought and quantify the degree of drought. The definition of the SWIt/SRIt is similar to that of the standardized precipitation index (SPI), but it is based on nonstationary Gamma distribution (Sutanto et al. 2020). In order to consider the time-varying characteristics of the distribution parameters of the water level and runoff series in the changing environment, time-varying nonstationary Gamma models were established for the water level series of Poyang and Dongting Lakes, the runoff series of the two lake basins and the discharges of the Yangtze River at the seasonal scale (Sun et al. 2020). Finally, the cumulative distribution calculated by the optimal model was transformed into the normal distribution by the equal probability transformation method, and then the SWIt and SRIt were obtained (Rashid & Beecham 2019; Song et al. 2020). When the values of the SWIt and SRIt are greater than 0, there is no drought, and when they are less than 0, there is drought, and the smaller SWIt and SRIt values indicate greater drought intensity. This study mainly focuses on hydrological drought events with the SWIt/SRIt ≤ −1.

Copula theory was first proposed by Sklar (1959) and has been widely used in the analysis of bivariate or multivariate probability distributions in hydrometeorological research. In this study, the Copula function was applied to quantify the joint probability of concurrent hydrological drought events in the lake–river system of the Poyang and Dongting Lake basins. Specifically, Gumbel, Clayton, Frank, Gaussion, t and Plackett Copula functions were used to fit the SWIt and SRIt series, and the relationship between the Kendall rank correlation coefficient τ and the parameter θ of the Copula function was established to estimate the parameters (Ravens 2000).

Cross-wavelet analysis is an extension of the wavelet analysis method, which is used to analyze the correlation between two time series [x(t), y(t)] in the time and frequency domains. In this study, in order to identify the key factors affecting hydrological drought at different times, cross-wavelet analysis of the hydrological drought index and the large-scale climate index was carried out to depict their phase structure and detailed characteristics in the time and frequency domains. The cross-wavelet energy spectrum can show the same energy spectrum region of x(t) and y(t) after wavelet transformation, which reflects the structural characteristics of the two series after wavelet transformation in the time–frequency domain. The advantage of the cross-wavelet energy spectrum is that the main oscillation period of the signal can be observed, and the evolution of the oscillation period with time at different time scales can also be more clearly observed.

Additionally, the start date and duration of the dry season in the two lakes were compared and analyzed, as shown in Figure 2. The water level of Poyang Lake is lower than 12.0 m at the Hukou station and the water level of Dongting Lake is lower than 24.5 m at the Chenglingji station, indicating that they have entered the dry season. The start date of the dry season was depicted by the Julian day number, which is a continuous count of days from the first day of the year. Figure 2(b) and 2(d) shows that the Julian day numbers presented long-term decreasing trends, which means that the two lakes began their dry season earlier. Especially, Poyang and Dongting Lakes began to exhibit low water in October (Julian day <300) in the past 10 years, which was obviously earlier than the average times historically. At the same time, the dry season has become longer, even the duration has exceeded 200 days in the recent year.

The hydrological drought of Poyang and Dongting Lake is affected by both the inflows of the local catchments and discharges of the Yangtze River, but the dominant factors change with time (Yuan et al. 2016). This study revealed that the joint probability of concurrent hydrological drought between Dongting Lake and its catchment was higher than that between Dongting Lake and the Yangtze River in spring, which indicated that the hydrological drought of Dongting Lake in spring was synchronous with the hydrological drought in the local catchment. The decrease in water inflow from the local catchment was the main factor driving hydrological drought in Dongting Lake in spring. This result was consistent with the study of Sun et al. (2015), and they believed that the spring drought in Dongting Lake was basically caused by the lack of water inflow from the local catchment. He et al. (2021) also revealed that the water volume of Dongting Lake was mainly affected by the inflow of the local catchment. While, in autumn, this study found that the hydrological drought of Dongting Lake was mainly affected by the discharge of the Yangtze River, with the high joint probability of simultaneous hydrological drought in the Dongting Lake–Yangtze River system (15.38%). Previous studies have shown that after the operation of the TGR, the water level–discharge relationship of the Yangtze River changed, and the impoundment of the reservoir during autumn significantly reduced the water level and discharge of the middle and lower reaches of the Yangtze River (Gao et al. 2013), which weakened the jacking effect of the Yangtze River on the two lakes and aggravated the low water phenomenon in the lakes (Ou et al. 2012; Guo et al. 2018). Lai et al. (2014) also found that the impoundment of the TGR in autumn caused the water level of Dongting Lake to drop rapidly and the low water level to appear ahead of schedule. On the other hand, the reduction of precipitation in the Dongting Lake basin also plays an important role. For example, the study of Ou et al. (2012) showed that the extreme drought event in Dongting Lake in autumn 2006 was caused by the sharp decrease in discharge of the Yangtze River and water inflow from the four rivers in the basin, with an average decrease by 74.7 and 30.4%, respectively.

A similar situation exists also in Poyang Lake. Fang et al. (2012) found that the operation of the TGR made the dry season of Poyang Lake start approximately 1 month ahead of schedule. Zhang et al. (2012) found that the average water level of Poyang Lake dropped by 2.0 m at the Hukou station and the jacking effect of the Yangtze River weakened after impoundment of the TGR, which increased the possibility of hydrological drought in Poyang Lake in summer and autumn.

In addition, the effects of other human activities on the hydrological drought of the two lakes cannot be ignored. According to the Statistic Bulletin on China Water Activities, there were 10,798 reservoirs with a total storage capacity of 32.8 billion m³ in Jiangxi Province, including 30 large reservoirs (storage capacity > 100 million m³) and 251 medium reservoirs (storage capacity > 10 million m³), and 14,098 reservoirs with a total storage capacity of 51.4 billion m³ in Hunan Province, including 45 large reservoirs and 359 medium reservoirs. The operation of these reservoirs directly changed the seasonal distribution of runoff of the five rivers in the Poyang Lake basin and the four rivers in the Dongting Lake basin. Additionally, with the prohibition of sand mining in the main stream of the Yangtze River in 2000, many sand dredgers entered Poyang Lake and Dongting Lake to conduct sand mining. Continuous sand mining has caused erosion of the lake basin, resulting in an increase in the gradient of the channel into the Yangtze River in the lake area (Yao et al. 2018), which accelerated lake water drainage to the Yangtze River and further aggravated the hydrological drought in the lake area (Yao et al. 2019). In addition, forestation in the two lake basins began in the 1980s, with the forest coverage of the Poyang and Dongting Lake basins increasing from 41 and 36.7% in 1989 to 61.2 and 59.6% in 2016, respectively (Ye et al. 2013; Yu et al. 2018). The increase in forest coverage led to an increase in regional evapotranspiration and interception, which undoubtedly resulted in a significant decrease in runoff on a monthly or seasonal scale. Moreover, the two lake plains are important grain bases in China, with large agricultural water consumption. For example, the agricultural water consumption in the Dongting Lake plain was 10.12 billion m³ in 2016, accounting for more than 70% of the total water consumption. In particular, mid-September is the peak period of irrigation water for late rice and other crops, and a large number of agricultural irrigation water demand further aggravated the water shortage problem in Dongting Lake in autumn (Liang et al. 2021).

Overall, the impact process and mechanism of human activities on the hydrology and water resources of Poyang and Dongting Lakes are complex. In the context of the changing relationships of Yangtze River–Lake, it is necessary to strengthen the ecological environment construction and optimize the scheduling scheme of water conservancy projects to meet the demand of production and domestic and ecological water use (Yang et al. 2016). On the other hand, water-saving agriculture should be developed to improve the efficiency of water use in agricultural production. Quantitative assessment of the impact of different human activities on drought in Poyang and Dongting Lakes is the focus of the next step and further studies need to be conducted with quantitative hydrological and hydrodynamic simulations.

This study analyzed and compared the characteristics and differences in hydrological drought in Poyang and Dongting Lakes during the period of 1964–2016 and further investigated the correlation between hydrological drought and large-scale climate indices. The results of the study showed that the hydrological drought of Poyang and Dongting Lakes has become increasingly serious and their average lake level in the dry season decreased. The two lakes entered the dry season earlier; furthermore, the duration of the dry season was prolonged and the intensity of hydrological drought events increased. The occurrence probabilities of hydrological drought at Poyang and Dongting Lakes were 16.40 and 15.58% in spring, 27.65 and 32.46% in autumn, and 14.26 and 4.60% in winter, respectively. The hydrological drought of Poyang Lake was more serious than that of Dongting Lake in spring and winter, while, in autumn, it was not as serious as that of Dongting Lake. The joint probabilities of concurrent hydrological drought in the Lake–catchment systems were the highest in spring (10.84 and 9.52%) and the joint probabilities in the Lake–Yangtze River systems were the largest in autumn (26.39 and 27.76%). Finally, the changes in large-scale climate indices had an important effect on the hydrological drought of Poyang and Dongting Lakes. The SWIt of both lakes was negatively correlated with the SOI (−0.34 and −0.30) and positively correlated with NINO3.4 (0.38 and 0.31) at the annual scale. At the seasonal scale, the hydrological drought in Poyang Lake was more closely related to the changes in climate indices, while their correlation relationships were weak for Dongting Lake, especially in autumn and winter. In addition, the impoundment of the TGR and other human activities, such as sand mining, reservoir construction, agricultural production and forestation in the local catchment, played important roles in aggravating the hydrological drought that occurred in the two lakes.

This work was jointly funded by the National Key Research and Development Project of China (Grant No. 2018YFE0206400), the National Natural Science Foundation of China (Grant No. 41871093 and 42071028) and the Strategic Priority Research Program of the Chinese Academy of Sciences (Grant No. XDA23040202).

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.