Diversion rivers are important water and biological connection channels, which provide valuable ecological services. However, diversion rivers are degraded owing to the construction and unreasonable operation of large-scale water conservancy projects. Quantifying and guaranteeing the ecological flow of diversion rivers is necessary to sustain the riverine ecosystems. Due to the lack of methods to verify the rationality of the existing hydrological series length for ecological flow calculation by hydrological methods, a Monte Carlo-length of record (MC-LOR) algorithm was proposed to analyze the accuracy level of the existing hydrological series. Taking the Dongjing River (DJR) in the lower Han River as an example, the result shows that the 15-year hydrological series can meet the accuracy level of 5/90 and 10/95. The flow changes and hydrological relationships between the Qianjiang hydrological station (QJS) and the Tianguan River (TGR) under the influence of the Middle Route of South-to-North Water Transfer (MSNW) project and the Yangtze–Han Water Diversion (YHWD) project were analyzed. It is found that the flow variation of QJS is large and shows a downward trend while TGR displays a significant upward trend through the M-K trend test. And the flow of QJS is drier and less than the water replenishment of TGR in the non-flood season after the operation of the MSNW project. This paper established an ecological flow threshold calculation method based on kernel density curves and quantiles. Compared with the minimum average monthly flow (MAMF) method, the monthly minimum ecological runoff (MMR) method, improved Q90 method, Tennant method, Tessmann method, distribution flow method (DFM), the method proposed in this paper can reduce the influence of extreme flow events and is suitable for the diversion river. Finally, the ecological water replenishment demands of TGR to meet the ecological flow target of DJR under different flow conditions in the Han River were analyzed by Bayesian network (BN) models. The results show that the minimum water requirement thresholds of DJR are 31.3m3/s and 15.85m3/s, 55.35m3/s and 30.85m3/s when the flows of Han River are < 654m3/s and 654 ~ 770m3/s in non-flood season and < 875m3/s and 875 ~ 1210m3/s in flood season, respectively. This research provides references for the ecological stability and sustainable development of diversion rivers.
Research Article
Research on ecological flow and water replenishment threshold for diversion river based on MC-LOR
https://doi.org/10.21203/rs.3.rs-1526138/v1
This work is licensed under a CC BY 4.0 License
You are reading this latest preprint version
Ecological flow
Diversion River
KDCQ method
Monte Carlo-length of record algorithm
Ecological water replenishment
To efficiently utilize limited water resources and reduce water disasters in China, 23,841 dams (with a height of more than 15m) have been built by 2020 according to International Commission on Large Dams (ICOLD), accounting for 40.6% of the world. However, dams have led to fragmentation and hindered the free flow of rivers. Only 37% of rivers longer than 1000km in the world are not affected by reservoirs and dams (Grill et al. 2019), and 41% of rivers with an annual mean flow (AMF) > 0.1m3/s cease to flow at least one day per year (Messager et al. 2021), which poses a challenge to the ecological health of rivers (Liu et al. 2011). Research on ecological flow is never more needed for evaluating river health (Guan et al. 2021; Liu et al. 2020). A scientific and reasonable estimation of ecological flow is of great significance to promote the rational utilization of water resources and achieve sustainable development of rivers.
There are more than 200 ecological flow calculation methods globally, which can be roughly divided into hydrological, hydraulic, habitat simulation, and holistic methods (Cheng et al. 2019; Wang et al. 2016). However, the hydraulic, habitat simulation and holistic methods require a large amount of data as support, which cannot be obtained in data-scarce areas. The hydrological method has been widely used, which is simple and easy to handle based on historical flow (Liu et al. 2011). For example, Richter et al. (Richter et al. 1996) constructed the indicators of hydrologic alteration (IHA) method including 5 groups of hydrological indicators, including flow magnitude, timing, duration, frequency, and change rate, and further proposed the range of variability (RVA) method based on the IHA method (Richter et al. 1998), and the functional relationship between fish abundance and hydrological indicators was obtained by data mining method (Yang et al. 2008). Five hydrological methods were used to calculate the ecological flow at the global scale (Pastor et al. 2014), and found that the Tessmann method and the variable monthly flow (VMF) method had good calculation results. A multi-level ecological flow calculation method was proposed and used to assess river health (Ma et al. 2019). However, the hydrological method requires a sufficient hydrological sequence to guarantee accuracy. Richter et al. (Richter et al. 1996) analyzed three rivers with different hydrological conditions and believed that the calculation of the IHA index requires at least 20 years of hydrological data, and some scholars believed that it needed 10 to 40 years (Huh et al. 2005), i.e. 7Q10 method generally uses the lowest average flows that occur for a consecutive 7-day period at the recurrence interval of 10 years (Ouyang 2012). Due to the uncertainty in the required hydrological sequence, the length of record (LOR) method was proposed to quantitatively determine the length of data to calculate ecological flow based on the existing hydrological series (Timpe et al. 2017), but it cannot verify the rationality of the length of existing hydrological series.
Diversion rivers are branches of the main rivers, which are the result of water and sediment movement in plains or deltas. At present, the researches on diversion rivers mainly focus on the diversion ratio (Meselhe et al. 2012; Sun et al. 2019) and flood control safety (Zhang et al. 2014), and little attention is paid to ecological flow. The Dongjing River (DJR) is a diversion river connecting the Han River and the Yangtze River with various functions such as flood control and drainage, water supply, and ecological services. Its diversion ratio increases with the increase of flow in the Han River and can reach 1/4 ~ 1/6 in flood period (Wang et al. 2016). However, the construction of the cascade reservoirs and a large amount of water diversion from the Han River have reduced the flow in the middle and lower reaches of the Han River (Palmer et al. 2019; Song et al. 2018; Yin et al. 2020; Zhang et al. 2022), and the scouring of the Han river due to the reservoir's function of storing muddy water and releasing clear water has lowered the water level (Bin et al. 2020), which all have caused a continuous flow reduction of DJR. The ecological water replenishment is effective for river health (Yan et al. 2018), and the Yangtze–Han Water Diversion (YHWD) project supplies water to DJR through the Tianguan River (TGR), but no scholars have quantitatively analyzed the water demand of DJR under different flow conditions of the Han River. Because the Bayesian network (BN) model can clearly describe the uncertainty of the system, it is gradually used in the field of water resources management, such as water supply, groundwater, and irrigation (Shi et al. 2020), but few scholars have applied it to ecological water replenishment.
Based on previous studies, the main objectives of this study are: (1) to construct an algorithm to assess the accuracy level of the existing hydrological series for the calculation of ecological flow hydrology method; (2) to analyze the influence of the Middle Route of South-to-North Water Transfer (MSNW) project and YHWD project on the diversion flow of the Qianjiang Hydrological Station (QJS) and the water replenishment process of TGR respectively; (3) to propose an ecological flow threshold calculation method suitable for the diversion river; (4) to analyze the ecological water replenishment requirements of DJR under different flow scenarios of the Han River.
2.1 Research area
DJR is a diversion river connecting the Han River and the Yangtze River with a total length of 173 km. It is located in the hinterland of the Jianghan Plain, north of the middle reaches of the Yangtze River and south of the lower reaches of the Han River, and enters the Yangtze River at Sanhe protective embankments, Wuhan City, Hubei Province. DJR is an important source of water supply, irrigation and ecology along the river. The location and related water conservancy projects of DJR are shown in Fig. 1.
QJS is a basic hydrological station monitoring diversion flow from the Han River, which is located about 7 km downstream of the source of DJR in Qianjiang City, Hubei Province. The AMF of QJS from 2006 to 2020 is 95.18 m3/s. In addition to the diversion from the Han River to DJR, TGR supplies water to DJR with AMF of 29.67 m3/s through the Tianguan sluice (or Tianguan pumping station) at 8.8km downstream of the QJS. The daily-scale hydrological data used in this study are obtained from the Hubei Clairvoyant Water and Rain Inquiry System (http://113.57.190.228:8001/#!/web/Report/RiverReport).
2.2 Monte Carlo-length of record (MC-LOR) method
LOR method is a quantitative method to determine the length of computational data required to calculate the hydrological index based on the existing hydrological series (Zhang et al. 2019; Zhou et al. 2020). However, it is impossible to verify the rationality of the length of the existing hydrological series. This study innovatively combines the Monte Carlo stochastic simulation (MC) and LOR method to verify the accuracy level of the existing hydrological series for the calculation of the ecological flow hydrology method. The specific calculation steps are as follows:
1) AMF is calculated based on the measured daily flow;
2) The common hydrological probability distribution functions (PDFs) are tested for the AMF, and the best PDF is selected. MC is used to randomly generate a long series of flow data X that fits the best PDF (parameter errors all less than 1%);
X=(X 1 , X2,……, Xn) (1)
where n is the simulation length.
3) Based on the generated data X, i data are extracted randomly and repeatedly for 5000 times, totaling (n-1)\(\times\)5000 times, and the mean value of each extracted data is calculated to form matrix Y. 85%, 90%, and 95% confidence intervals (CI) for the i-year data are calculated based on Y;
Y = \(\left[\begin{array}{ccc}{y}_{21}& \cdots & {y}_{2m}\\ ⋮& \ddots & ⋮\\ {y}_{n1}& \cdots & {y}_{nm}\end{array}\right]\) (2)
Upper confidence limit: \({CI}_{iku}=\frac{1}{m}\sum _{j=1}^{j=m}{y}_{ij}+{\mu }_{k}\frac{{\sigma }_{i}}{\sqrt{i}}\) (3)
Lower confidence limit: \({CI}_{ikl}=\frac{1}{m}\sum _{j=1}^{j=m}{y}_{ij}-{\mu }_{k}\frac{{\sigma }_{i}}{\sqrt{i}}\) (4)
where, Y is the mean matrix; \({y}_{ij}\) is the mean value of each extracted i data; i = 2, 3, ..., n; j = 1, 2, ..., m; m = 5000; \({CI}_{iku}\) is the upper confidence limit corresponding to the i-year data and confidence level k;\({CI}_{ikl}\) is the lower confidence limit corresponding to the i-year data and confidence level k; k = 1, 2, 3, represent 85%, 90% and 95% confidence levels respectively; \({\mu }_{k}\) represents the quantile corresponding to different confidence levels; \({\sigma }_{i}\) is the standard deviation of (\({y}_{i1}, {y}_{i2}, \dots , {y}_{im}\)).
4) Data lengths corresponding to errors of 5%, 10% and 15% of the measured AMF on CIiku and CIikl are calculated respectively. MC-LOR results with a 10% error and a 90% CI can be abbreviated as 10/90 (Timpe et al. 2017), and so on.
LOR αku =\(i+1\) \(if (1+\alpha )\times {X}_{mean}\)<\({CI}_{iku}\) and \(（1+\alpha ）\times {X}_{mean}>{CI}_{(i+1)ku}\) (5)
LOR αkl =\(i+1\) \(if (1-\alpha )\times {X}_{mean}\)>\({CI}_{ikl}\) and \(（1-\alpha ）\times {X}_{mean}<{CI}_{(i+1)kl}\) (6)
LORαk = max{LORαku
LORαkl} (7)
LORαkl} (7)where, Xmean is the measured AMF; α is the error; LORαku is the result calculated based on the upper confidence limit; LORαkl is the calculation result of the lower confidence limit; LORαk is final the result corresponding to error α and confidence level k; max is the function of taking the maximum value.
2.3 Ecological flow threshold calculation method based on kernel density curves and quantiles
The traditional hydrological methods use the percentages of average river flow as the standard for calculating ecological flow, which is easily affected by extreme hydrological events that will damage or impede river ecosystems in most situations (Li et al. 2014; Reich and Lake 2015). Researchers have used parameter estimation to fit the probability distribution for ecological flow research, but it has shortcomings such as low accuracy and poor flexibility (Tan et al. 2018). Many studies have proved that non-parametric estimation, especially kernel density estimation, has better application effects (Tan et al. 2018). Thus, this paper proposes an ecological flow calculation method based on the combination of kernel density curves and quantiles (KDCQ), the calculation steps are as follows.
Most suitable ecological flow (Q0): The most suitable ecological flow is the flow process that can maintain species diversity and ecosystem stability (Ma et al. 2019). In the selected daily flow sequence, the flows that correspond to the highest points of the monthly mean flow kernel density curve and the median monthly average flow are compared. The smaller of these two flows is treated as the most suitable ecological flow for the month, and the most suitable ecological flow of the whole year is composed of the monthly most suitable ecological flow.
Lower threshold for the most suitable ecological flow (Q1): In the selected daily flow sequence, the average of the monthly flow at the 25th quantile and the most suitable ecological flow is taken as the lower threshold of the suitable ecological flow. The lower threshold of annual suitable ecological flow is composed of the lower threshold of most suitable ecological flow in each month.
Upper threshold for the most suitable ecological flow (Q2): In the selected daily flow sequence, the average value of the monthly flow at the 75th quantile and the most suitable ecological flow is taken as the upper threshold of the most suitable ecological flow. The upper threshold of the annual most suitable ecological flow is composed of the upper threshold of the most suitable ecological flow in each month.
Minimum ecological flow (Q3): The minimum ecological flow is the most basic demand flow that the river can maintain in the short term, but the long-term existence will cause damage to the aquatic ecosystem (Guan et al. 2021). In the selected daily flow sequence, the monthly flow at the 10th quantile is selected as the minimum ecological flow, and the minimum ecological flow per month is used to establish the minimum annual ecological flow process.
Maximum ecological flow (Q4): The maximum ecological flow is the flow process that provides energy and materials for plains, lakes and wetlands connecting the rivers (Bond et al. 2014; Talukdar et al. 2019). In the selected daily flow sequence, the monthly flow at the 90th quantile is selected as the maximum ecological flow, and the maximum ecological flow per month is used to form the maximum annual ecological flow process.
2.4 BN models
BN models were originally proposed by Pearl in 1988 by combining graph theory and the Bayesian probability formula (Shi et al. 2020). BN models are composed of a directed acyclic graph (DAG) and a conditional probability table (CPT) based on the causal relationship between variables (Cheng et al. 2021; Owusu et al. 2022). BN models originated from artificial intelligence (Uusitalo 2007), and have been widely used present in uncertain fields, such as ecological environment and hydraulic engineering (Morrison and Stone 2014) because of the ability to model complex causal relationships qualitatively and quantitatively (Uusitalo 2007). There is no time for TGR to operate when flow of QJS is small as they are next to each other (Fig. 1). The flow of DJR is mainly affected by the Han River even there exists some uncertainty. Therefore, BN models were used to analyze the ecological water replenishment demands of TGR to meet the ecological flow target of DJR under flow conditions of the Han River. The advantages of BN models (Shenton et al. 2011): (1) they have an obvious causal relationship and are easy to construct, expand and optimize; (2) they are able to handle missing data; (3) they have high prediction accuracy even with small sample sizes; (4) they can build models simultaneously using observational data and expert experience; (5) they can avoid overfitting effectively.
2.4.1 Determination of the structure of BN models
The determination of the structure of BN models is mainly based on expert experience or structure learning from a large amount of measured data. Studies have shown that the BN models obtained based on expert experience are significantly better than those established by data learning (Huang 2013). Therefore, the BN models were established by analyzing the main factors affecting the flow of DJR as shown in Fig. 2, in which the Xiantao hydrological station (XTS) closest to the source of DJR represented the flow of Han River.
2.4.2 Determination parameters of BN models
To reduce the amount of calculation, the node parameters were discretized. The flow of XTS and QJS were divided into five grades equally from low to high according to the flow quantiles, A1, A2, A3, A4, A5 and B1, B2, B3, B4, B5 respectively. Given that the water replenishment of TGR is regulated by Tianguan sluice, the scenario of no water replenishment was set as C1, and the other four levels were also bounded by quantiles equally. Based on the multi-level ecological flow of the KDCQ method, the flow of DJR was divided into 5 levels from low to high, as shown in Fig. 2.
2.4.3 Evaluation of BN models
K-fold cross-validation is an effective way to evaluate the performance of fitted models (Fushiki 2011). K-fold cross-validation divides the data into K groups, and each group is reserved for testing the model, and the remaining K-1 groups are used to train the model. K-fold cross-validation is performed K times in total, and the confusion matrices of the K times are averaged to evaluate the overall performance (Shi et al. 2020). This paper used 5-fold cross-validation to evaluate the overall performance of BN models by comparing the consistency of the posterior probability state with the actual probability state.
3.1 Calculation of MC-LOR
Eight commonly used hydrological probability distribution functions (PDFs), including gamma distribution (gamma), pearson III distribution (pearson3), generalized extreme value distribution (genextreme), uniform distribution (uniform), normal distribution (norm), exponential distribution (expon), generalized Pareto distribution (pareto) and lognormal distribution (lognorm) (Guan et al. 2021; Yisehak et al. 2020) were fitted based on the measured AMF of DJR by Python. The fitted PDFs were tested and evaluated for goodness by K-S test and root mean square error (RMSE) (Table 1). All the seven PDFs except lognorm pass the K-S test. According to the P-value and RMSE, the pearson3 is selected as the best PDF for the AMF of DJR. The 50-year, 100-year, and 200-year flow series that obeyed the same pearson3 are randomly generated. The simulated data are all located near to y = x in the Q-Q diagram (Fig. 3), which confirms the simulation effect is perfect.
|
分布类型 |
Pearson3 |
gamma |
genextreme |
uniform |
norm |
expon |
pareto |
lognorm |
|---|---|---|---|---|---|---|---|---|
|
P |
0.995 |
0.995 |
0.982 |
0.376 |
0.926 |
0.42 |
0.09 |
0.001 |
|
RMSE |
0.03008 |
0.03008 |
0.03009 |
0.03025 |
0.03029 |
0.03046 |
0.03188 |
0.03156 |
The CIs of the simulated annual flow of DJR were calculated by using the method established in Section 2.2, as shown in Fig. 4. The 85%, 90%, and 95% CIs gradually decrease with the increase of the years, and the gaps between CIs are decreasing and coincide with the mean value ultimately. Table 2 showed the MC-LOR results of different simulation years that satisfied 85%, 90% and 95% CIs and 15%, 10% and 5% error levels. Under the same error conditions, the greater the CI, the greater the result of MC-LOR, but CI has no significant effect on MC-LOR results (P > 0.05). The MC-LOR result increases rapidly with the reduction of the error with certain CI, and the error level has a significant influence on the MC-LOR result (P < 0.05), and the 5% error level is significantly different from the 10% and 15% error level (P < 0.01), respectively. From the annual perspective, the 15-year hydrological series in this study can meet the accuracy of 5/90 and 10/95 for all simulation years.
|
Year |
15/85 |
10/85 |
5/85 |
15/90 |
10/90 |
5/90 |
15/95 |
10/95 |
5/95 |
|---|---|---|---|---|---|---|---|---|---|
|
50 |
3 |
5 |
11 |
4 |
6 |
12 |
4 |
7 |
14 |
|
100 |
3 |
5 |
11 |
4 |
6 |
14 |
5 |
8 |
16 |
|
200 |
4 |
7 |
13 |
5 |
7 |
15 |
6 |
9 |
18 |
The optimal PDF of each monthly flow was determined (Fig. 5), and a random series of certain length that conformed to the optimal PDF was generated, respectively. Only 100-year is selected as the monthly simulation series in this study because the increase in the simulation length has little effect on MC-LOR results. Similar to the annual perspective, the CI has a small impact on the monthly MC-LOR results (P > 0.05), while the error level has a large impact (P < 0.01). The study finds that the MC-LOR result is larger in July, September, October, and December because the coefficient of variation of the monthly mean flow in these four months is larger. Some scholars find that the greater the coefficient of variation, the larger the number of years required for LOR (Zhang et al. 2019), which is highly consistent with the result of this study. The 15-year hydrological data in this study can meet the accuracy of 10/95 from the monthly perspective.
3.2 Analysis of flow change of DJR
From Fig. 6, it can be seen that the intra- and inter-annual variation of the flow at QJS is drastic, and the flow in the flood season is relatively large with a maximum flow of 3010m3/s, but the flow in the non-flood season is insufficient obviously. The water supply of TGR is relatively stable, but also large in flood season and small in the non-flood season. The YHWD project began to replenish DJR in 2014 through TGR with an average of 600 million m3 per year, and the stability of water replenishment of TGR in the non-flood season has increased significantly.
From 2006 to 2020, it can be seen that the diversion flow of QJS shows a downward trend (z=-1.39) while the water replenishment of TGR displays a significant upward trend (z = 2.28) respectively through the M-K trend test (Fig. 7). The annual average water replenishment of TGR is 14.32 ~ 53.03m3/s, and the annual average diversion flow of QJS is 14.23 ~ 223.83m3/s with the inter-annual extreme value ratio reaching 15.73. And the coefficient of variation of AMF of QJS is 0.678, which is much higher than that of the four hydrological stations in the mainstream of the Huaihe River (0.53 ~ 0.58) (Pan et al. 2013) and the West, North, and East rivers of the Pearl River (0.196 ~ 0.273) (Yang et al. 2019). From 2006 to 2020, the flow from May to October in the flood season of QJS accounts for about 80% of AMF, which is much larger than the ratio of the six hydrological stations in the middle and lower reaches of the Han River during the same period from 1999 to 2013 (63%~66%) (Li et al. 2015). And the flow from April to October at the three hydrological stations in Yangtze River during 1960 ~ 2015 accounted for about 80% of the AMF (Xu et al. 2020), which is smaller than the proportion of QJS (84.6%). So, the diversion flow of QJS has a large intra- and inter-annual variability.
Before the operation of the MSNW project (2006 ~ 2014), the diversion flow of QJS is 5.29 times as large as the water replenishment of TGR, but the years of diversion flow of QJS less than water replenishment of TGR accounts for 1/3 after the operation of the MSNW project (2015 ~ 2020). The diversion flow of QJS from November to April of the following year decreased significantly after the MSNW project, which is only 54.58% of that before the MSNW project and 68.29% of the water replenishment of TGR. The possible reasons are that the MSNW project reduces the flow in the middle and lower reaches of the Han River (Yu et al. 2020), and the function of storing muddy water and releasing clear water of Xinglong Water Control Project has lowered the elevation of the river bottom of the Han River and resulted in the gradual shrinking of the right bank where the source of DJR is located (Bin et al. 2020), which all reduces the diversion ratio of DJR and reduces the diversion flow finally. Under the background of the increasing water diversion from the Han River, the diversion flow of QJS will inevitably show a decreasing trend. DJR maybe only play the role of flood diversion of the Han River during the flood period, and the ecological flow will mainly rely on TGR in the future. Thus, this paper focuses on the research on the ecological flow of DJR with the combination of diversion flow of QJS and water replenishment of TGR.
3.3 Ecological flow of DJR based on the KDCQ method
The ecological flow of DJR based on the KDCQ method was illustrated in Fig. 8. The Q0 of DJR in January, June, and July is the median monthly average flow and the remaining is the flow corresponding to the highest point of the kernel density curve. It can be seen from Fig. 8 that the difference between the multi-level ecological flow from November to April of the following year in the non-flood season is relatively small compared with that in the flood season. Q4 and Q3 are the maximum and minimum flow of the multi-level ecological flow in each month, respectively. The maximum values of the ecological flow of Q0 ~ Q3 appear in July and August, while Q4 appears in October which is affected by the extreme flood in October 2017. At the same time, Q4 far exceeds Q0 ~ Q3 with the AMF of 253.52m3/s, which is 4.01 times that of Q0 (63.16m3/s). Only Q3 has no change all the year, which is only 23.39% of Q0. Therefore, the ecological flow of DJR is easily affected by the extreme flow.
The interval from May to October is considered as the flood season of DJR and the interval between July and September is the main flood season. It can be seen from Table 3 that the Q0 ~ Q3 from May to October and Q0 ~ Q4 from July to September in the flood season account for 39.3 ~ 42.96% and 63.19%~72.71% of the corresponding annual ecological flow respectively, which are less than the corresponding proportion of the flow of DJR. January through March is the dry season before the flood and prone to water blooms (Liang et al. 2012), and Q0 ~ Q3 accounts for 11.26 ~ 20.82% of the annual ecological flow of the corresponding grade during this period, which is greater than the proportion of flow during the same period (10.03%). Only Q4 has abnormal phenomena, mainly because the diversion proportion of DJR increases with the flow increase of the Han River. Large-scale floods occurred in the Han River from August to September 2010, September 2011, and October 2017 (Fig. 6), causing the Q4 of DJR from August to October much larger than others (Fig. 8). Therefore, the proportion of ecological flow calculated by the KDCQ method is smaller in the flood season and greater in the non-flood season than the proportions of the flow of DJR, which reduces the impacts of extreme hydrological events. Therefore, the KDCQ method is suitable for diversion rivers with large intra- and inter-annual variability.
|
Period |
Q0 |
Q1 |
Q2 |
Q3 |
Q4 |
DJR |
|---|---|---|---|---|---|---|
|
May ~ October |
69.04 |
68.83 |
72.71 |
63.19 |
77.43 |
75.30 |
|
January ~ March |
14.52 |
14.09 |
11.26 |
20.82 |
8.4 |
10.03 |
|
July ~ September |
39.83 |
39.95 |
42.96 |
39.3 |
41.14 |
46.73 |
3.4 Comparing the KDCQ method with other hydrological methods
To further assess the rationality of the proposed KDCQ method, the result of multi-level ecological flow is compared with other hydrology methods (Fig. 9). Comparing Q3 with the minimum average monthly flow (MAMF) method, the monthly minimum ecological runoff (MMR) method, and improved Q90 method (Q90, 90% guarantee rate based on flow duration curve), which are widely recognized minimum ecological flow calculation methods (Tan et al. 2018), it is found that the change trends of Q3 and Q90 method represent consistency and the proportions in the AMF are 11.46% and 10.7% respectively, and the proportions of MAMF and MMR method are 17.83% and 9.68% of AMF, respectively. The Tennant method uses 10% of the multi-year average flow as the critical value to maintain the short-term survival of aquatic organisms (Tennant 1976). Q3 accounts for slightly more than 10% of the MAF and the result is located between MAMF and MMR methods.
The suitability of the river ecosystem and biodiversity reaches the best when the river flow is the most suitable (Ma et al. 2019). In this study, the smaller value of the flow corresponding to the highest point of the monthly flow kernel density curve and the median monthly average flow is regarded as the most suitable ecological flow. The monthly range of Q0 of the KDCQ method is 29.76 ~ 154.2m3/s, with an average value of 63.16m3/s, which is far lower than MAF, accounting for 50.59% of MAF only. The Tessmann method (Pastor et al. 2014), as a typical hydrological ecological flow calculation method, calculates that the ecological flow of DJR accounted for 47.18% of AMF, which is close to the results of Q0. The Tennant method believed that 40% of AMF can provide better habitat conditions for most aquatic organisms (Tennant 1976), and the annual suitable ecological flow threshold (Q1 ~ Q2) of the KDCQ method accounts for 41.18%~80.96%. The distribution flow method (DFM) method (Tan et al. 2018) also divides the ecological flow into different levels and calculates separately, and the suitable ecological flow threshold of the DFM method accounts for 37.48%~89.31% of AMF of DJR. The proportional range of the suitable ecological flow of the KDCQ method is within the DFM method.
Flood is an important power for river sediment transport and the maintenance of riparian ecosystems (Bond et al. 2014; Tonkin et al. 2018). Research has regarded the maximum value of the monthly average flow of many years as the maximum ecological flow (Ma et al. 2019), and this method reaches 380.64m3/s in DJR. The Tennant method takes 200% of the average annual flow as the maximum critical threshold of river ecological flow, and it is not conducive to the development of ecosystems if the flow exceeds this threshold (Tennant 1976). Q4 is 2.03 times AMF. In a word, the multi-level ecological flow calculated by the KDCQ method proposed in this paper is scientific and reasonable and can be used to calculate the ecological flow of diversion rivers.
3.5 Multi-scenario analysis of ecological water replenishment in DJR
Due to the discernible decrease of diversion flow of QJS, the minimum ecological flow guarantee rates (the proportion exceeding the minimum ecological flow) were only 33.33% (2014), 41.67% (2016), and 16.67 (2019) without ecological water replenishment of TGR. Ecological water replenishment of TGR is regulated by Tianguan sluice (or Tianguan pumping station) and YHWD project. Therefore, the ecological flow replenishment of DJR under different water flow conditions of the Han River need to be calculated. According to the change of the multi-level ecological flow of DJR (Fig. 7 and Fig. 8), it can be seen that the ecological flow of DJR is quite different in flood season and non-flood season. Therefore, the ecological water demands of DJR to maintain the optimal state under different flow scenarios of the Han River in the non-flood season and flood season were calculated separately by BN models. Five-fold cross-validation showed that the accuracy rates of the BN models were 72.2% and 74.2% in the non-flood season and flood season respectively, and the wrongly predicted samples were basically located in the adjacent interval of the actual sample, which proved that BN models had high accuracy (Shi et al. 2020).
Figure 10 depicted the proportions of the flow of DJR between multi-level ecological flow in non-flood season and flood season based on the KDCQ method under multi-scenario. The flow of DJR increases with the increase of flow of the Han River and TGR, and the changes of non-flood season and flood season are highly consistent. When the flow of the Han River is less than 562m3/s in non-flood season and 653m3/s in flood season (A1), the probabilities of flow less than Q1 are as high as 98.6% and 97.3% respectively. Although the C3 water replenishment level (15.85 ~ 31.3 m3/s in non-flood season and 30.85 ~ 55.35m3/s in flood season) can meet the minimum ecological flow demand, they are basically at D2 lower than the lower threshold of the most suitable ecological flow, which is not conducive to the healthy development of the river. Therefore, it is recommended that the water replenishment should reach C4 (31.3 ~ 49.3 m3/s in non-flood season and 55.35 ~ 83.2m3/s in flood season). When the flow of the Han River is 562 ~ 654m3/s in non-flood season and 653 ~ 875m3/s in flood season (A2), the probabilities of diversion flow of QJS less than Q1 during the non-flood season and flood season are 82.6% and 66.76% respectively. Under this condition, if the ecological water replenishment of TGR reaches C4, the flow of DJR will be at the optimal state. 60.7% of the non-flood season and 52.9% of the flood season are still less than Q1 when the flow of the Han River is 654 ~ 770m3/s in non-flood season and 875 ~ 1210m3/s in flood season (A3), respectively. When the water replenishment of TGR reaches C3, the minimum ecological flow guarantee rates of DJR rise to 98.71% and 98.96% respectively, and both the flows exhibit normal distribution centered on the most suitable ecological flow. The flows of QJS can meet the minimum ecological flow requirements and the distributions are reasonable when the flows of the Han River are more than 770m3/s in non-flood season and 1210m3/s in flood season (A4 and A5), so there is no need to consider water replenishment. The results discussed above can provide scientific references for the management of DJR.
Owing to the lack of data, this research was carried out based on water quantity only. The ecological flow and water replenishment of DJR needs further research, such as considering the impact of water quality and water intake and drainage.
Aiming at the continuous degradation of the diversion river in the lower Han River, this paper constructed a MC-LOR method to quantitatively analyze the accuracy of the existing hydrological sequence for calculating ecological flow based on the hydrological method and proposed an ecological flow threshold calculation method combined kernel density curve and quantiles. And through the BN models, the ecological water replenishment of DJR under different flow scenarios of the Han River was analyzed. The main conclusions are as follows:
(1) The error level has a significant effect on the MC-LOR result (P < 0.05), while the CI and the number of simulation years have less effect. From the annual (monthly) scale, the 15-year hydrological series of DJR can meet the accuracy level of 5/90 and 10/95 (10/95) for the calculation of ecological flow.
(2) By analyzing the impact of the MSNW project and YHWD project on the hydrological process of QJS and TGR, it is found that the diversion flow of the QJS has remarkable intra- and inter-annual variability and decreasing trend, and DJR may only play the role of flood diversion of the Han River during flood period and ecological water replenishment will mainly rely on TGR in the future.
(3) The suitable ecological flow threshold of the KDCQ method accounts for 41.18%~80.96% of the AMF of DJR, and the annual most suitable ecological flow is 63.16m3/s, accounting for 50.59%. Compared with various hydrological ecological flow calculation methods, the KDCQ method can reduce the influence of extreme flow events.
(4) When the flows of the Han River are greater than 770m3/s in non-flood season and greater than 1210m3/s in flood season, the flows of QJS can both meet the ecological flow requirements. The best ecological water replenishment thresholds of DJR are 31.3 ~ 49.3m3/s and 15.85 ~ 31.3m3/s, 55.35 ~ 83.2m3/s and 30.85 ~ 55.35m3/s respectively when the flows of Han River are < 654m3/s and 654 ~ 770m3/s in non-flood season and < 875m3/s and 875 ~ 1210m3/s in flood season.
Acknowledgments
The authors would like to thank the reviewers and the editors for their valuable suggestions and contributions which significantly help to improve this article. This study was supported by the Variation Characteristics of Water Transformation and its Water Saving and Emission Reduction Adaptation Mode in Jianghan Plain Under Changing Environment (U21A20156).
No conflict of interest
- Bin B, Yu X (2020) Analysis of the evolution and dry water level change of the Xinglong zekou reach of the Han River. China Rural Water and Hydropower 09105–111. https://doi.org/10.3969/j.issn.1007-2284.2020.09.021 (in Chinese)
- Bond N, Costelloe J, King A, Warfe D, Reich P, Balcombe S (2014) Ecological risks and opportunities from engineered artificial flooding as a means of achieving environmental flow objectives. Front Ecol Environ 12(7):386–394. https://doi.org/10.1890/130259
- Cheng B, Li H, Yue S, Huang K (2019) A conceptual decision-making for the ecological base flow of rivers considering the economic value of ecosystem services of rivers in water shortage area of northwest China. J Hydrol 578:124126. https://doi.org/10.1016/j.jhydrol.2019.124126
- Cheng B, Li H (2021) A bayesian network approach for determining optimal ecological base flow of rivers in water shortage areas of northwest China. Environ Sci Pollut R 28(28):37768–37780. https://doi.org/10.1007/s11356-021-13384-w
- Fushiki T (2011) Estimation of prediction error by using k-fold cross-validation. Stat Comput 21(2):137–146. https://doi.org/10.1007/s11222-009-9153-8
- Grill G, Lehner B, Thieme M, Geenen B, Tickner D, Antonelli F et al (2019) Mapping the world's free-flowing rivers. Nature 569(7755):215–221. https://doi.org/10.1038/s41586-019-1111-9
- Guan X, Zhang Y, Meng Y, Liu Y, Yan D (2021) Study on the theories and methods of ecological flow guarantee rate index under different time scales. Sci Total Environ 771:145378. https://doi.org/10.1016/j.scitotenv.2021.145378
- Huang Y (2013) Survey on Bayesian network development and application. Trans Beijing Inst Technol 33(12):1211–1219. https://doi.org/10.15918/j.tbit1001-0645.2013.12.022 (in Chinese)
- Huh SH, Dickey DA, Meador MR, Ruhl KE (2005) Temporal analysis of the frequency and duration of low and high streamflow: years of record needed to characterize streamflow variability. J Hydrol 310(1–4):78–94. https://doi.org/10.1016/j.jhydrol.2004.12.008
- Li C, Kang L (2014) A new modified tennant method with spatial-temporal variability. Water Resour Manag 28(14):4911–4926. https://doi.org/10.1007/s11269-014-0746-4
- Li Y, Wang X, Zhou B, Yuan D, Chen J (2015) Characteristics and change trend of water quantity and quality in middle and lower reaches of Hanjiang River before South-to-North Water Diversion running. J China Hydrology 35(05):82–90. https://doi.org/10.3969/j.issn.1000-0852.2015.05.016 (in Chinese)
- Liang K, Wang X, Zhang D, Zhou Y (2012) Ecological conditions of diatom water bloom formulation in the middle and lower reach of the Hanjiang River and strategy for water bloom control. Environ Sci Technol 35(12J):113–116 (in Chinese)
- Liu C, Zhao C, Xia J, Sun C, Wang R, Liu T (2011) An instream ecological flow method for data-scarce regulated rivers. J Hydrol 398(1–2):17–25. https://doi.org/10.1016/j.jhydrol.2010.11.026
- Liu F, Qin T, Yan D, Wang Y, Dong B, Wang J, Nie H, He S, Liu S (2020) Classification of instream ecological water demand and crucial values in a semi-arid river basin. Sci Total Environ 712:136409. https://doi.org/10.1016/j.scitotenv.2019.136409
- Ma D, Luo W, Yang G, Lu J, Fan Y (2019) A study on a river health assessment method based on ecological flow. Ecol Model 401:144–154. https://doi.org/10.1016/j.ecolmodel.2018.11.023
- Meselhe EA, Georgiou I, Allison MA, Mccorquodale JA (2012) Numerical modeling of hydrodynamics and sediment transport in lower Mississippi at a proposed delta building diversion. J Hydrol 472–473:340–354. https://doi.org/10.1016/j.jhydrol.2012.09.043
- Messager ML, Lehner B, Cockburn C, Lamouroux N, Pella H, Snelder T, Tockner K, Trautmann T, Watt C, Datry T (2021) Global prevalence of non-perennial rivers and streams. Nature 594(7863):391–397. https://doi.org/10.1038/s41586-021-03565-5
- Morrison R, Stone M (2014) Spatially implemented Bayesian network model to assess environmental impacts of water management. Water Resour Res 50(10):8107–8124. https://doi.org/10.1002/2014WR015600
- Ouyang Y (2012) A potential approach for low flow selection in water resource supply and management. J Hydrol 454–455:56–63. https://doi.org/10.1016/j.jhydrol.2012.05.062
- Owusu A, Mul M, Strauch M, van der Zaag P, Volk M, Slinger J (2022) The clam and the dam: a Bayesian belief network approach to environmental flow assessment in a data scarce region. Sci Total Environ 810:151315. https://doi.org/10.1016/j.scitotenv.2021.151315
- Palmer M, Ruhi A (2019) Linkages between flow regime, biota, and ecosystem processes: implications for river restoration. Science. 365(6459). https://doi.org/10.1126/science.aaw2087
- Pan Z, Ruan X, Zhu Y, Li R (2013) Analysis of variation in runoff of the main stream of Huaihe River in the past 50 years. J soil water conserv 27(01):51–55. https://doi.org/10.13870/j.cnki.stbcxb.2013.01.007 (in Chinese)
- Pastor AV, Ludwig F, Biemans H, Hoff H, Kabat P (2014) Accounting for environmental flow requirements in global water assessments. Hydrol Earth Syst Sci 18(12):5041–5059. https://doi.org/10.5194/hess-18-5041-2014
- Reich P, Lake P (2015) Extreme hydrological events and the ecological restoration of flowing waters. Freshw Biol 60(12):2639–2652. https://doi.org/10.1111/fwb.12508
- Richter BD, Baumgartner JV, Braun DP, Powell J (1998) A spatial assessment of hydrologic alteration within a river network. Regul River-Res Manage 14(4):329–340
- Richter BD, Baumgartner JV, Powell J, Braun DP (1996) A method for assessing hydrologic alteration within ecosystems. Conserv Biol 10(4):1163–1174. https://doi.org/10.1046/j.1523-1739.1996.10041163.x
- Shenton W, Hart BT, Chan T (2011) Bayesian network models for environmental flow decision-making: 1. Latrobe river Australia. River Res Appl 27(3):283–296. https://doi.org/10.1002/rra.1348
- Shi H, Luo G, Zheng H, Chen C, Bai J, Liu T (2020) Water use analysis of Syr Darya river basin: Based on "Water-Energy-Food-Ecology" nexus and Bayesian network. Scientia Geogr Sinica 75(05):1036–1052. https://doi.org/10.11821/dlxb202005011 (in Chinese)
- Song X, Zhuang Y, Wang X, Li AE (2018) Combined effect of Danjiangkou Reservoir and cascade reservoirs on hydrologic regime downstream. J Hydrol Eng 23(6). https://doi.org/10.1061/(ASCE)HE.1943-5584.0001660
- Sun Z, Fu X, Chu X (2019) Assessing the impacts of reservoir operation on downstream water diversions using a simplified flow model. Hydrol Sci j 64(12):1488–1503. https://doi.org/10.1080/02626667.2019.1649410
- Talukdar S, Pal S (2019) Effects of damming on the hydrological regime of Punarbhaba river basin wetlands. Ecol Eng 135:61–74. https://doi.org/10.1016/j.ecoleng.2019.05.014
- Tan G, Yi R, Chang J, Shu C, Yin Z, Han S, Feng Z, Lyu Y (2018) A new method for calculating ecological flow: distribution flow method. AIP Adv 8(4):45118. https://doi.org/10.1063/1.5022048
- Timpe K, Kaplan D (2017) The changing hydrology of a dammed Amazon. Sci Adv 3(11):e1700611. https://doi.org/10.1126/sciadv.1700611
- Tonkin JD, Merritt DM, Olden JD, Reynolds LV, Lytle DA (2018) Flow regime alteration degrades ecological networks in riparian ecosystems. Nat Ecol Evol 2(1):86–93. doi:10.1038/s41559-017-0379-0
- Uusitalo L (2007) Advantages and challenges of Bayesian networks in environmental modelling. Ecol Model 203(3–4):312–318. https://doi.org/10.1016/j.ecolmodel.2006.11.033
- Wang B, Shao D, Mu G, Wang Z, Li X (2016) An eco-functional classification for environmental flow assessment in the pearl river basin in Guangdong. China Sci China-Technol Sci 59(2):265–275. https://doi.org/10.1007/s11431-015-5926-9
- Wang D, You X, Tang J, Weng Z, Wang L (2016) Study of flood control in the downstream of Dongjing River. Yangtze River 33(07):39–45. https://doi.org/10.11988/ckyyb.20150450 (in Chinese)
- Xu X, Wu Y, Sheng D, He H (2020) J water Resour water Eng 31(04):57–62. https://doi.org/10.11705/j.issn.1672-643X.2020.04.08 (in Chinese) Evolution characteristics of water resources at Dongting Lake from 1960 to 2015
- Yan Z, Zhou Z, Sang X, Wang H (2018) Water replenishment for ecological flow with an improved water resources allocation model. Sci Total Environ 643:1152–1165. https://doi.org/10.1016/j.scitotenv.2018.06.085
- Yang Y, Wang Y, Cai S, Liu F (2019) Interannual and intra-annual variation characteristic of runoff in downstream area of Pearl River basin during 1960–2017. Bull soil water Conserv 39(05):23–31. https://doi.org/10.13961/j.cnki.stbctb.2019.05.004 (in Chinese)
- Yang YE, Cai X, Herricks EE (2008) Water Resour Res 44(4). https://doi.org/10.1029/2006WR005764. Identification of hydrologic indicators related to fish diversity and abundance: a data mining approach for fish community analysis
- Yin X, Zhang J, Chen J (2020) The impact of multi-projects on the alteration of the flow regime in the middle and lower course of the Hanjiang River, China. Water 12(8):2301. https://doi.org/10.3390/w12082301
- Yisehak B, Adhena K, Shiferaw H, Hagos H, Abrha H, Bezabh T (2020) Characteristics of hydrological extremes in Kulfo River of southern Ethiopian Rift Valley basin. SN Appl Sci 2(7). https://doi.org/10.1007/s42452-020-3097-1
- Yu M, Wood P, van de Giesen N, Liu X, Li Q, Wang G, Zhang J (2020) Enhanced potential ecological risk induced by a large scale water diversion project. Stoch Environ Res Risk Assess 34(12):2125–2138. https://doi.org/10.1007/s00477-020-01861-6
- Zhang J, Guo L, Huang T, Zhang D, Deng Z, Liu L, Yan T (2022) Hydro-environmental response to the inter-basin water resource development in the middle and lower Han River, China. Hydrol Res 53(1):141–155. https://doi.org/10.2166/nh.2021.090
- Zhang J, Yu M (2014) Experimental study of flood diversion in the middle and lower Han River, China. Can J Civ Eng 41(5):381–388. https://doi.org/10.1139/cjce-2013-0374
- Zhang W, Jia Y, Ge J, Huang X, Ni G, Hou J, Wang H (2019) Multi-index data dimension reduction approach and its applicability in the calculation of indicators of hydrological alteration. Hydrol Res 50(1):231–243. https://doi.org/10.2166/nh.2018.068
- Zhou X, Huang X, Zhao H, Ma K (2020) Development of a revised method for indicators of hydrologic alteration for analyzing the cumulative impacts of cascading reservoirs on flow regime. Hydrol Earth Syst Sci 24(8):4091–4107. https://doi.org/10.5194/hess-24-4091-2020
- Zhou, X., Huang, X., Zhao, H., Ma, K., 2020. Development of a revised method for indicators of hydrologic alteration for analyzing the cumulative impacts of cascading reservoirs on flow regime. Hydrol. Earth Syst. Sci. 24(8), 4091–4107. https://doi.org/10.5194/hess-24-4091-2020.