Water Flow Regulation and Scheme Optimization in the Haihe River

Abstract

To improve water quality in the Haihe River, 6 cycle routes and 16 water diversion schemes were simulated, and their influence on water quality was analyzed. First, a one-dimensional hydrodynamic and water quality model of the Haihe River was established. Then, according to an orthogonal experiment, 16 cycle schemes were proposed, and an orthogonal experimental table, including 3 factors and 4 levels, was used to optimize the 16 schemes. Average pollutant concentration from Sanchakou to the Erdao Sluice served as the index for orthogonal experiments. Index values were obtained from the model and the influence of three factors on chemical oxygen demand, total nitrogen, total phosphorus, and nitrogen ammonia was evaluated by range analysis. In addition, optimal schemes to reduce pollutants in each route were also obtained. After comparing the simulated results of the optimal schemes in each route, an optimal route was obtained. A combination of the one-dimensional hydrodynamic and water quality models with orthogonal experimental design was found to provide an efficient method to simulate and optimize the schemes, thereby improving water quality by water flow regulation in plain river network regions by taking into account multiple factors such as water cycle route, the operation of sluice, and the water diversion project.

Introduction

Plain river network regions are common in delta areas, and their formation is affected primarily by estuarine sedimentation (Nanson et al., 1996; Makaske, 2011; Meng et al., 2014). Plain river network regions have two significant characteristics as follows: (1) sluices and embankments are widely built to form artificial catchment areas; therefore, there are few natural catchment areas; (2) there is substantial interference with runoff generation and flow concentration processes. For example, in the Haihe River Basin, flow direction is reciprocating and affected by the tide (Putty and Prasad, 2000; Wang, 2007).

According to previous studies (Zhou et al., 2013; Julian et al., 2015; Zhang, 2015; Zhao et al., 2015), plain river network regions have superior natural conditions, abundant water resources, developed social economies, and large populations, which are beneficial for social and economic development. However, water resources are subject to high environmental pressures. For example, plain river network regions have slow water velocity and weak water discharge, resulting in a series of environmental and ecological problems. With the development of urbanization and the continuous intervention of human activities, the number of sluices and pump stations in plain river network regions is increasing; thus, rivers are currently divided into different sections, worsening river connectivity. In addition, point source and nonpoint source pollutants are discharged into rivers, which cause severe water-based environmental problems. Therefore, it is important to perform comprehensive water quality and quantity control research on plain river network regions.

Water flow regulation is a method to measure water quality regulation indirectly, and on this basis, technological measures, such as resource recovery and water cycling schemes, have been developed (Mccarty et al., 2011; Jin et al., 2015). These regulations were performed early in developed countries, and some effective measures were carried out. Water diversion projects have also been very effective to improve water quality (Pyka et al., 2016). Mathematical modeling also provides a useful tool to simulate water environment and demonstrate the potential for significant improvements in river water quality (Bach et al., 2014).

In China, rivers have been dredged to improve water circulation, and the total amount of pollutants discharged into rivers have been regulated; in addition, water diversion projects have been carried out (Zhou, 2011; Sheng et al., 2013; Chen et al., 2015). However, although the continuous deterioration of the water environment is currently controlled in local watersheds, the water quality of rivers, lakes, and reservoirs needs long-term improvement.

Water flow regulation is an efficient measure to improve water quality; however, there are many rivers, sluices, and dams in plain river network regions. The number of water cycle schemes will increase if the number of factors considered in the schemes increases. Therefore, it is necessary to select a method to formulate and optimize schemes. In this study, the combination one-dimensional hydrodynamic and water quality model with orthogonal experimental design was used to address this problem.

The Haihe River Basin, a typical plain river network region, represents an important economic, political, and cultural development center in China. In recent years, urbanization, human disturbance, riverbed hardening, discharge of point source and nonpoint source pollutants, and a massive loss of biodiversity have caused problems in the Haihe River mainstream. Although water pollution control in the Haihe River has achieved some goals in recent years, water quality is still poor and poses a threat to the ecological security and development of the economy (Chen et al., 2016; Rong et al., 2016). Therefore, it is imperative to improve the water environment and enhance water self-purification capacity in the Haihe River Basin.

Coupled water quality–quantity technology is based on the hydrodynamic and water quality model, which reflects the mutual influence between the qualitative and quantitative attributes of water resources. It is a comprehensive method that describes water quality and quantity control and has been used widely in water quality and quantity simulation (Zuo and Xia, 2002; Zhao et al., 2003, 2013). To address river blockage, water quality deterioration, and the destruction of the ecological environment, the Haihe River mainstream was selected as the research area and a one-dimensional hydrodynamic and water quality model was established. By using different diversion discharges, diversion durations, and the operations of the Erdao Sluice, 16 water diversion schemes were simulated. The effects of water cycle routes and water diversion on water quality were analyzed by range analysis, and optimal schemes were obtained from an orthogonal experimental design. This study will help improve water quality in the Haihe River mainstream and also provides a scientific basis and technical examples for protecting water ecosystems in plain river network regions.

Study Region

Geography conditions

The Haihe River Basin, located in the northern part of China, lies on longitude 112°–120° E and latitude 35°–43° N, with a drainage area of 318,200 km². The Haihe River crosses Tianjin City and has many functions, such as drainage, water supply, shipping, and tourism. Its mainstream is the channel into the sea of five major rivers as follows: the Ziya, Daqing, Nanyun, Beiyun, and Yongding Rivers. The reach between the Qujiadian Sluice and Sanchakou in the Beiyun River, the reach between Xihe Sluice and Sanchakou in the Ziya River, and the reach between Sanchakou and Haihe Sluice in the Haihe River were included in the study area (Fig. 1). The Beiyun and Ziya Rivers flow into Haihe River in Sanchakou.

FIG. 1.

Study area location in Haihe River Basin.

The Beiyun River in the study area has a length of 16.7 km and 24 cross-sections in the model, and the first cross-section in the study area was located in Qujiadian Sluice; the Ziya River in the study area has a length of 15.7 km and 16 cross-sections in the model, and the first cross-section in the study area was located in Xihe Sluice; the Haihe River has a length of 71.5 km and 81 cross-sections in the model, and the first cross-section was located in Sanchakou. The first cross-sections in the Beiyun River and Ziya River in the study area were not the initial points in the two rivers. The distance from initial points in each river to each cross-section was named chainage, and the unit of chainage was meter. The cross-sections were named after chainage, and the location of the cross-sections was assigned based on chainage.

In this study, the first cross-section in the Beiyun River in the study area was named the 55,000 cross-section, which meant the distance from the cross-section to the initial point was 55,000 m; the last cross-section in the Beiyun River in the study area was named the 70,700 cross-section, which meant the distance from the cross-section to the initial point was 70,700 m. These naming methods of cross-sections were applied to the Ziya River and Haihe River as well. The chainage ranges in each river were 55,000–70700, 11,000–27,740, and 0–71,500.

Current state of water pollution

Environmental quality standards for surface water (GB 3838-2002) were published in 2002 in China. According to the standards, water at the Level-I water quality standard should be used as source water and for national nature reserves; water at the Level-II water quality standard should be used for concentrated drinking water of first-class reserves and for spawning grounds for fish and shrimps; water at the Level-III water quality standard should be used for concentrated drinking water of second-class reserves and aquaculture area; water at the Level-IV water quality standard should be used for industrial production and entertainment with no direct human contact; water at the Level-V water quality standard should be used for agriculture and common landscapes.

The Erdao Sluice divides the Haihe River mainstream into two sections; water quality in the two sections has varied in recent years. In 2010, the water quality upstream of the Erdao Sluice reached the Level-IV water quality standard, but the water quality downstream of the Erdao Sluice reached the Level-V water quality standard (Table 1). The water quality of branches was even poorer, and 40% of the water body was black and odoriferous, owing to the cyanobacteria bloom in summer.

Table 1.

Quality Standard for Surface Water

	COD	TP	TN	NH₃-N
Standard	Concentration (mg/L)
V	40	0.40	2.00	2.00
IV	30	0.30	1.50	1.50
III	20	0.20	1.00	1.00
II	15	0.10	0.50	0.50
I	15	0.02	0.20	0.20

COD, chemical oxygen demand; NH₃-N, nitrogen ammonia; TN, total nitrogen; TP, total phosphorus.

According to statistical reports of water quality in the Haihe River from 2010 to 2015, the major pollutants in the Haihe River were total nitrogen (TN) and chemical oxygen demand (COD), followed by total phosphorus (TP) and nitrogen ammonia (NH₃-N). The current state of pollution is shown in Table 2.

Table 2.

Statistical Results of Water Quality in Haihe River from 2010 to 2015

Pollutant	COD	TP	TN	NH₃-N
	Concentration (mg/L)
Range	11–146	0.01–1.85	0.50–13.0	0.05–5.80
Mean	47.90	0.33	3.89	0.97
Level-V water quality standard	<40	<0.40	<2.00	<2.00
Over standard rate (%)	57.30	29.20	88.10	10.80

Methodology

One-dimensional hydrodynamic model

MIKE 11 HD applied with the dynamic wave description solves the vertically integrated equations of the conservation of continuity and momentum (the Saint-Venant equations), on the basis of the following assumptions:

1. Water is incompressible and homogeneous.

2. The slope of the bottom is small, thus the cosine of the angle may be taken as 1.

3. Wave length is far larger than the water depth.

4. The flow is subcritical.

The specific forms of the equations are as follows: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} { \frac { \partial A } { \partial { \rm { t } } } } + { \frac { \partial Q } { \partial x } } = q , \tag { 1 } \end{align*} \end{document} \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} { \frac { \partial Q } { \partial { \rm { t } } } } + \frac { \partial } { { \partial { \rm { x } } } } \left( { \alpha \frac { { { Q^2 } } } { A } } \right) + gA { \frac { \partial h } { \partial x } } + g { \frac { Q \left\vert Q \right\vert } { C_z^2AR } } = 0 , \tag { 2 } \end{align*} \end{document}

where Q is the discharge, m³/s; A represents the flow area, m²; q stands for lateral inflow, m³/s; h is water deep, m; C_z is Chezy resistance coefficient; R represents the hydraulic or resistance radius, m; t is the time coordinate, s; x is the space coordinate, m; α is the momentum distribution coefficient.

One-dimensional water quality model

The one-dimensional advection–dispersion equation for matter in water is as follows: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} { \frac { \partial AC } { \partial { \rm { t } } } } + { \frac { \partial QC } { \partial x } } - \frac { \partial } { { \partial x } } \left( { AD { \frac { \partial C } { \partial x } } } \right) = - AKC + { C_2 } q , \tag { 3 } \end{align*} \end{document}

where C is the concentration of pollutants, mg/L; D stands for dispersion coefficient, m²/s; K is comprehensive decay coefficient, s⁻¹; C₂ represents the source or sink concentration, mg/L.

The advection equation is the second term on the left in Equation (3) and the dispersion equation is the third term on the left in Equation (3). Two mechanisms of material transport are expressed by Equation (3) as follows: advective transport is in proportion to discharge, so it is also in proportion to flow velocity. Dispersive transport is related to concentration gradients. Advective transport plays a dominant role in mass transport process in lotic environments. Dispersive transport cannot be neglected in lentic environments. Equation (3) has three basic assumptions as follows: (1) the material is completely mixed at the cross-section, (2) the material is conserved or decays linearly, and (3) Fick's law is satisfied.

Mike 11 AD was used to simulate advective and dispersive transport processes. In this module, changes of nonconservative matter concentrations in time coordinates due to the effect of water flow and concentration gradient were reflected and simulated by the comprehensive decay coefficient as the concentrated expression of complex processes. In fact, a comprehensive coefficient simulated the change, not taking into account the specific changing processes of pollutants, such as denitrification, nitrification, mineralization. The complex pollutant processes influencing concentration were replaced with the comprehensive decay coefficient. The value of the comprehensive decay coefficient was obtained from calibration and validation processes; it was found to be reliable.

Establishment of the One-Dimensional Hydrodynamic and Water Quality Model

One-dimensional hydrodynamic model

Calculation conditions

The basic calculation conditions for the model were as follows:

1. Terrain conditions:

Topographic data were cross-section based on the 1985 National Elevation Standard, and the interval between cross-sections was ∼1,000 m.

2. Hydrological conditions:

Discharge and water level data during the flood season (July 1 to October 1) in 2015 and 2014 were used to calibrate and validate the parameters.

3. Boundary conditions:

Boundaries (Fig. 2) included three outer boundaries, two inner boundaries, and nine lateral inflow processes.

FIG. 2.

Boundary conditions of hydraulic model.

The upstream boundary conditions were the inflow process of the Qujiadian Sluice (Figs. 3A and 4A) and the outflow process of the Xihe Sluice (Figs. 3B and 4B); the tidal level process downstream of the Haihe Sluice (Figs. 3C and 4C) was the downstream boundary condition.

FIG. 3.

(A) Inflow of Qujiadian Sluice in 2015. (B) Outflow of Xihe Sluice in 2015. (C) Tidal level downstream of Haihe Sluice in 2015. (C1) Partial process of tidal level. Conditions of model calibration at three outer boundaries.

FIG. 4.

(A) Inflow of Qujiadian Sluice in 2014. (B) Outflow of Xihe Sluice in 2014. (C) Tidal level downstream of Haihe Sluice in 2014. (C1) Partial process of tidal level. Conditions of model validation at three outer boundaries.

Erdao Sluice and Haihe Sluice were the inner boundaries. The operation of Erdao Sluice was based on measured discharge data (Figs. 5 and 6). Haihe Sluice was closed to prevent sea water intrusion except for when the up-sluice water level was higher than the down-sluice water level.

FIG. 5.

Calibration discharge.

FIG. 6.

Validation discharge.

4. Lateral inflow:

The Storm Water Management Model (SWMM) includes a hydrological model, a hydrodynamic model, and a water quality model. It primarily simulates precipitation and runoff generation and is widely used for the simulation of storms and floods in the city.

SWMM was established to study runoff and pollutants generated by rainfall in Tianjin City (Li, 2016). Based on terrain slope data and drainage network conditions, Tianjin City was divided into 415 subcatchment areas (Fig. 7A). To show subcatchment areas clearly, a part of the division diagram, subcatchment areas in city center, is shown in Fig. 7B.

FIG. 7.

(A) Division diagram of SWMM in Tianjin. (B) Division diagram in the city center. Division diagram of SWMM. SWMM, Storm Water Management Model.

Some parameters in the model were obtained from measurement, such as the area and width of the subcatchment. Other parameters were obtained using measured data of calibration and validation, such as roughness, washoff coefficient, and buildup constant. The model was used to study runoff and nonpoint pollutants in different rainfall conditions, and the results were found to be reliable.

In this study, measured nonpoint pollutant data during the simulation period were short, so the rainfall data during the simulation period were input into Li's (2016) SWMM to obtain runoff and water quality processes of lateral inflows. Although runoff and water quality data of lateral inflows were obtained by simulation, the error was acceptable according to Li (2016), and the results reflected the characteristics of runoff and nonpoint pollutants in Tianjin City.

For example, the runoff processes of the Dazhigu Pump Station, marked with a red triangle (Fig. 7B), are shown in Figs. 8 and 9.

FIG. 8.

Runoff of the model calibration.

FIG. 9.

Runoff of model validation.

5. Initial conditions:

Data measured on July 1, 2014, and July 1, 2015, were used as the initial water level and discharge (Table 3).

Table 3.

Initial Conditions of Hydrodynamic Model

	2014		2015
Location	Initial water level (m)	Initial discharge (m³/s)	Initial water level (m)	Initial discharge (m³/s)
Down-sluice of Qujiadian Sluice	−0.24	0	−0.53	0
Down-sluice of Xihe Sluice	−0.35	0	−0.66	0
Sanchakou	−0.35	0	−0.67	0
Erdao Sluice	−0.37	0	−0.66	0
Haihe Sluice	−1.30	0	−0.79	0

6. Parameter setting:

The main parameter was riverbed roughness. Roughness was obtained by trial and error. First, an initial roughness was set based on relevant references. Then, the simulated water level values were compared with the measured water level values at control sections. If the differences between them were beyond the acceptable range (the relative error could not exceed 13% in this study), the roughness was adjusted. This step was repeated until the differences were within a reasonable range.

Calibration

Data from July to October in 2015 were used for the model. The comparison between simulation values and measured values at control sections, including the Qujiadian Sluice, the Xihe Sluice, Sanchakou, the Erdao Sluice, and the Haihe Sluice, was obtained. The water level from Sanchakou is as an example (Fig. 10).

FIG. 10.

Calibrated water levels at Sanchakou.

Simulated results given in Fig. 10 show the basic trends of variations in measured values. Root-mean-squared error (RMSE) and correlation coefficient (r) were used to evaluate the model accuracy. The smaller the RMSE is, the smaller the difference between simulated values and measured values is; the closer the r is to 1, the better the agreement between simulated values and measured values is.

The two statistical variables are calculated as follows: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} RMSE = \sqrt { \mathop \sum \limits_{i = 1}^n {{ \left[ {{{ \left( {{X_m}} \right) }_i} - {{ \left( {{X_s}} \right) }_i}} \right] }^2} / n} , \tag{4} \end{align*} \end{document} \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} r = { \frac { \sum \nolimits_ { i = 1 } ^n \left[ { { { \left( { { X_m } } \right) } _i } - { { \left( { { { \bar X } _m } } \right) } _i } } \right] \left[ { { { \left( { { X_s } } \right) } _i } - { { \left( { { { \bar X } _s } } \right) } _i } } \right] } { \sqrt { \sum \nolimits_ { i = 1 } ^n { { \left[ { { { \left( { { X_m } } \right) } _i } - { { \left( { { { \bar X } _m } } \right) } _i } } \right] } ^2 } \cdot \sum \nolimits_ { i = 1 } ^n { { \left[ { { { \left( { { X_s } } \right) } _i } - { { \left( { { { \bar X } _s } } \right) } _i } } \right] } ^2 } } } } , \tag { 5 } \end{align*} \end{document}

where m and s represent measured and simulated values, respectively, n represents the number of measured values, and \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} $${ \overline X _m}$$ \end{document} and \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} $${ \overline X _s}$$ \end{document} represent the average measured and simulated values, respectively.

The statistical analysis results are shown in Table 4.

Table 4.

Results of Measured and Simulated Water Level at Each Cross-Section in 2015

Location	RMSE (m)	r
Down-sluice of Qujiadian Sluice	0.038	0.973
Down-sluice of Xihe Sluice	0.036	0.982
Sanchakou	0.037	0.976
Up-sluice of Erdao Sluice	0.036	0.980
Up-sluice of Haihe Sluice	0.027	0.988

RMSE, root-mean-squared error.

The deviation between simulated values and measured values is small; therefore, the error is acceptable and the parameters are reasonable (Table 5).

Table 5.

Calibration Results of Hydrodynamic Model Parameters

Location	Roughness
Ziya River (11,000–27,740)	0.0225
Beiyun River (55,000–70,700)	0.0225
Haihe River (0–15,000)	0.0300
Haihe River (15,000–71,500)	0.0250

Validation

Based on the calibrated parameters, the model was validated by using data measured from July 1 to October 1, 2014, and the simulated results and values measured at control cross-sections were compared. The water level at Sanchakou was used as an example (Fig. 11).

FIG. 11.

Validated water levels at Sanchakou.

The statistical analysis of validated results is shown in Table 6.

Table 6.

Results of Water Level at Each Cross-Section in 2014

Location	RMSE (m)	r
Down-sluice of Qujiadian Sluice	0.024	0.977
Down-sluice of Xihe Sluice	0.031	0.968
Sanchakou	0.028	0.974
Up-sluice of Erdao Sluice	0.035	0.966
Up-sluice of Haihe Sluice	0.024	0.992

RMSE is between 0.024 and 0.035 m; correlation coefficients indicate that the results reach a 95% significance level. Therefore, the riverbed roughness is reasonable and reliable.

One-dimensional water quality model

Calculation conditions

1. Terrain conditions:

The study area and topographic conditions were the same as those used in the hydrodynamic model.

2. Water quality conditions:

Water quality data measured from July 1 to October 1, 2015, owing to their higher integrity and reliability, were used to calibrate water quality parameters in the model; data from July 1 to October 1, 2014, were used to validate the parameters.

3. Boundary conditions:

Water quality conditions of the three outer boundaries based on data from 2015 and 2014 are shown in Table 7.

Table 7.

Water Quality Conditions of Three Outer Boundaries in 2015 and 2014

Location	Time	Concentration of COD (mg/L)	Concentration of TP (mg/L)	Concentration of TN (mg/L)	Concentration of NH₃-N (mg/L)
Down-sluice of Xihe Sluice	2015.07	16	0.18	1.37	0.46
	2015.08	16	0.10	2.26	1.17
	2015.09	28	0.05	1.94	0.49
	2014.07	17	0.12	1.58	0.29
	2014.08	21	0.14	1.03	0.74
	2014.09	12	0.17	2.03	0.90
Down-sluice of Qujiadian Sluice	2015.07	21	0.14	1.78	0.31
	2015.08	14	0.07	1.68	1.11
	2015.09	26	0.04	1.74	0.28
	2014.07	18	0.09	2.16	0.21
	2014.08	10	0.09	2.47	0.45
	2014.09	11	0.05	1.24	0.09
Down-sluice of Haihe Sluice	2015.07	102	0.58	1.81	0.35
	2015.08	75	0.46	2.46	1.98
	2015.09	34	0.51	6.51	2.01
	2014.07	57	0.34	3.58	0.77
	2014.08	57	0.29	1.97	0.74
	2014.09	45	0.35	2.78	0.94

Water quality processes of lateral inflows generated were calculated using the SWMM. The pollutant concentrations in the Dazhigu Pump Station are shown in Figs. 12 and 13.

FIG. 12.

(A) COD concentration. (B) TP concentration. (C) TN concentration. (D) NH₃-N concentration. Pollutant concentrations of model calibration. COD, chemical oxygen demand; NH₃-N, nitrogen ammonia; TN, total nitrogen; TP, total phosphorus.

FIG. 13.

(A) COD concentration. (B) TP concentration. (C) TN concentration. (D) NH₃-N concentration. Pollutant concentrations of model validation.

4. Initial conditions:

The initial water level and discharge were the same as those used in the hydrodynamic model. Concentrations of water quality components from July 1, 2014, to July 1, 2015, were set as the initial water quality parameters. The values at control cross-sections are shown in Table 8.

Table 8.

Initial Water Quality Concentrations

Location	Concentration of COD	Concentration of TP	Concentration of TN	Concentration of NH₃-N
	Pollutant concentration on July 1, 2015 (mg/L)
Down-sluice of Qujiadian Sluice	21	0.140	1.780	0.310
Down-sluice of Xihe Sluice	16	0.180	1.370	0.460
Sanchakou	23	0.140	1.300	0.414
Erdao Sluice	35	0.250	1.820	0.075
Haihe Sluice	102	0.580	1.810	0.350
	Pollutant concentration on July 1, 2014 (mg/L)
Down-sluice of Qujiadian Sluice	18	0.090	2.160	0.210
Down-sluice of Xihe Sluice	17	0.120	1.580	0.290
Sanchakou	8	0.050	2.160	0.819
Erdao Sluice	16	0.260	2.390	1.200
Haihe Sluice	57	0.340	3.580	0.770

5. Parameter settings:

The one-dimensional water quality model was established based on the pollutant advection–dispersion equation; the main parameters of the equation were the dispersion and decay coefficients.

The Haihe River mainstream is located in the plain river network region, and the entire terrain is flat; the dispersion equation could not be neglected. However, flow velocity was high due to the inflow discharge upstream and lateral inflow generated by rainfall during the flood season. Therefore, advective transport should play a dominant role in mass transport process. In addition, dispersion coefficient changes only had a small effect on the simulated results by sensitivity analysis. Wang et al. (2008) reported similar findings in the Hanjiang Plain. Therefore, the dispersion coefficient was set at an experimental constant (20 m²/s).

The comprehensive decay coefficient was used to characterize the complex pollutant processes in water by describing the water's self-purification capacity against pollutants and the daily decay rate of pollutants in water (Wang et al., 2006; Guo et al., 2008). The comprehensive decay coefficient was affected by many factors, including flow rate, discharge, pH, water temperature, and concentration gradients. Therefore, comprehensive decay coefficient was uncertainty and dynamic.

The comprehensive decay coefficients of COD, TP, TN, and NH₃-N were obtained by the trial method due to this uncertainty.

Calibration

The model was calibrated with data from July 1 to October 1, 2015, and parameters were adjusted repeatedly such that the errors between simulated values and measured values fell within an acceptable range. Reasonable comprehensive decay coefficients of pollutants were obtained. The water quality calibration results at Sanchakou are shown as an example (Fig. 14).

FIG. 14.

(A) COD concentration. (B) TP concentration. (C) TN concentration. (D) NH₃-N concentration. Calibrated results at cross-section of Sanchakou in 2015.

Figure 14 shows that variations at each pollutant concentrations calculated from the model are consistent with the measured concentration. The calibrated comprehensive decay coefficients of the pollutants are shown in Table 9.

Table 9.

Comprehensive Decay Coefficients of Pollutants

Pollutant	Decay coefficient (per day)
COD	0.015
TP	0.005
TN	0.020
NH₃-N	0.063

Validation

The model parameters obtained in the calibration in the one-dimensional water quality model (Table 9) were validated using data from July 1 to October 1, 2014. The validated results of water quality at Sanchakou are shown as an example (Fig. 15).

FIG. 15.

(A) COD concentration. (B) TP concentration. (C) TN concentration. (D) NH₃-N concentration. Validated results at the cross-section of Sanchakou in 2014.

Simulated pollutant concentrations are consistent with the measured values. Although some deviation exists, the trends of the simulated and the measured values are essentially identical. Therefore, the comprehensive decay coefficients are found to be reasonable and reliable.

In conclusion, this model can be used to simulate the hydrodynamic and water quality processes and can also be used as a computational tool for further study of water quality.

Use of Orthogonal Experiments to Study Optimization of Water Quality and Quantity Regulation

Factors and levels in orthogonal experiments

Water diversion discharge and water diversion duration of the Qujiadian Sluice were two main factors that influenced the water cycle of the mainstream and branches of Haihe River; the operation of Erdao Sluice was a secondary factor. Different levels of these 3 factors were combined to form 16 water cycle schemes according to the orthogonal table.

Level choice of water diversion discharge

Water diversion discharge levels were selected by ranking the measured discharge values in the Qujiadian Sluice in 2014 and 2015 to get the values of water cycle schemes in orthogonal experiments close to those of the actual situation; all values were between 0 and 30 m³/s. The range was divided into four groups as follows: 0–7.5, 7.5–15, 15–22.5, and 22.5–30 m³/s (Fig. 16); the average values of each group were selected as the four level values (6, 10, 20, and 24 m³/s).

FIG. 16.

Distribution and average value of water division discharge.

Level choice of water diversion duration

According to the diversion data in Qujiadian Sluice in 2014 and 2015, the duration of water diversion varied from 10 to 15 days to several months. Long water diversions restrict water self-purification if the water quality of the water diversion is poor. Therefore, 60 days was selected as the upper limit of water diversion duration. The four levels of water diversion duration were 15, 30, 45, and 60 days.

Level choice of the operation of Erdao Sluice

To guarantee drainage function during flood season and landscape water level during the nonflood season, the operation of Erdao Sluice is described according to four assumptions as follows:

1. To guarantee flood control function during flood season and prevent seawater encroachment, the sluice is opened when the up-sluice water level is higher than the flood control water level (−0.61 m) and the down-sluice water level.

2. According to the measured data of Erdao Sluice in 2015, the sluice was opened when the up-sluice water level was higher than −0.5 m and the down-sluice water level.

3. According to the lower limit of landscape water level during the nonflood season, the sluice is opened when up-sluice water level is higher than −0.11 m and the down-sluice water level.

4. According to the upper limit of landscape water level, the sluice is opened when up-sluice water level is higher than 0.39 m and the down-sluice water level.

The levels of each factor in the orthogonal experiments are shown in Table 10.

Table 10.

Factor Levels in the Orthogonal Experiment

	Factors
Level	(A) Water diversion discharge (m³/s)	(B) Water diversion duration (day)	(C) The highest water level of the up-sluice of Erdao Sluice (m)
1	6	15	−0.61
2	10	30	−0.50
3	20	45	−0.11
4	24	60	0.39

Orthogonal experiments

Water cycle schemes for the mainstream and branches of Haihe River

The water diversion cycle schemes were obtained using the orthogonal experimental table L₁₆(4³) (Table 11).

Table 11.

Orthogonal Experiment Table L₁₆(4³)

Case	Combination	Water diversion discharge (m³/s)	Water diversion duration (day)	The highest water level of the up-sluice of Erdao Sluice (m)
1	A1B1C1	6	15	−0.61
2	A1B2C2	6	30	−0.50
3	A1B3C3	6	45	−0.11
4	A1B4C4	6	60	0.39
5	A2B1C2	10	15	−0.50
6	A2B2C1	10	30	−0.61
7	A2B3C4	10	45	0.39
8	A2B4C3	10	60	−0.11
9	A3B1C3	20	15	−0.11
10	A3B2C4	20	30	0.39
11	A3B3C1	20	45	−0.61
12	A3B4C2	20	60	−0.50
13	A4B1C4	24	15	0.39
14	A4B2C3	24	30	−0.11
15	A4B3C2	24	45	−0.50
16	A4B4C1	24	60	−0.61

In the mentioned 16 schemes, basic conditions were the same, except for water diversion discharge, water diversion duration, and the operation of Erdao Sluice.

Boundary conditions and initial conditions for the Haihe River mainstream

1. Boundary conditions:

Discharge and duration of water diversion in the Qujiadian Sluice were based on different combination schemes. The discharge of the Xihe Gate was 0. The downstream water level boundary in the down-sluice of Haihe Sluice was the tide level in 2015. Haihe Sluice is opened when the up-sluice water level is higher than down-sluice water level to prevent seawater encroachment.

Quality of the diverted water was set according to the data in the down-sluice of the Qujiadian Sluice in 2015 as follows: COD concentration was 14.7 mg/L, TP concentration was 0.13 mg/L, TN concentration was 1.3 mg/L, and NH₃-N concentration was 0.47 mg/L. Diverted water may improve water quality in the Haihe River, as the pollutant concentrations of the diverted water were lower than the average concentrations in the Haihe River.

As the main functions of rivers during the flood season are flood and waterlogging control, water quality requirements are not as strict as those during the nonflood season. Therefore, this study only evaluated ways to improve water quality during the nonflood season.

2. Initial conditions:

Water quality measured on April 23, 2015, was the poorest in 2015, and was selected as the initial water quality value (Table 12).

Table 12.

Initial Water Quality Concentrations of Water Cycle Schemes

Cross-section	Concentration of COD (mg/L)	Concentration of TP (mg/L)	Concentration of TN (mg/L)	Concentration of NH₃-N (mg/L)
Beiyun River Bridge	15	0.060	0.910	0.247
Qinjian Bridge	17	0.070	0.680	0.064
Ziya River Bridge	38	0.480	17.200	3.800
Yongle Bridge	20	0.080	1.140	0.182
Haijin Bridge	20	0.050	1.600	0.025
Haihe Bridge	79	0.090	1.450	0.456
Zhangjiazui Dock	31	0.150	2.170	0.510
Erdao Sluice	31	0.150	3.500	0.927
Weijin Sluice	38	0.250	5.030	1.570

COD and TN concentrations both exceeded Level-V water quality standard; water quality in the Ziya River Bridge was the poorest. Overall, the concentrations of TP and NH₃-N were relatively lower, and those of COD and TN were higher.

The initial discharge of control cross-sections was 0 in all schemes; initial water levels were set with the data measured on April 23, 2015 (Table 13).

Table 13.

Initial Water Levels of Water Cycle Schemes

Location	Initial water level (m)
Down-sluice of Qujiadian Sluice	−0.33
Down-sluice of Xihe Sluice	−0.50
Sanchakou	−0.48
Erdao Sluice	−0.47
Haihe Sluice	−1.03

Boundary conditions and initial conditions of branches

1. Boundary conditions:

To improve the water quality, this study summarized six cycle routes (Fig. 17).

FIG. 17.

(A) Route 1. (B) Route 2. (C) Route 3. (D) Route 4. (E) Route 5. (F) Route 6. Six cycle routes of mainstream and branches.

The Mingzhu pump station and Dongchangyin pump station are used to transfer mainstream water into branches as the water in the Haihe River mainstream scarcely flows into the branches. When only one pump station was used in the route, the discharge was 6 m³/s; when both pump stations were used, the discharge of each pump station was 3 m³/s.

2. Initial conditions:

According to the data, the initial water level of the branches was −0.6 m. The initial water quality was at Level-V.

Range analysis and scheme optimization of the orthogonal experiment

Average concentrations of pollutants from the Sanchakou to Erdao Sluice at the end of the simulation were selected as evaluation indexes to demonstrate the effects of water diversion schemes. Taking route 1 as an example, at the beginning of the simulation, the average concentrations of COD, TP, TN, and NH₃-N were 39.66, 0.12, 2.27, and 0.53 mg/L, respectively. The average concentrations at the end of the simulation for 16 orthogonal experimental schemes are shown in Fig. 18.

FIG. 18.

Index values of 16 schemes in route 1.

As shown in Fig. 18, the average concentration of COD is dramatically reduced, with the decay rate ranging from 63.5% to 78.7%, as the COD concentration of the diverted water is 14.7 mg/L, which is better than initial water quality. The comprehensive decay coefficient of COD is 0.015 per day, indicating that the water is at self-purification capacity. Under the double effect of water diversion and self-purification capacity, improvement by the 16 schemes on the COD concentration is excellent. However, the average concentration of TP is not significantly reduced. In contrast, the average concentration of TP is higher than its initial concentration in some schemes, because the TP concentration of diverted water (0.13 mg/L) is slightly higher than the initial TP concentration (0.12 mg/L), and the comprehensive decay coefficient of TP is 0.005 per day, resulting in very limited self-purification capacity of TP. Therefore, the 16 schemes have little effect on the average TP concentration. The average concentration of TN is reduced, with a decay rate ranging from 51.1% to 73.3%; this behavior is similar to that of COD. The concentration of NH₃-N is decreased, with the decay rate ranging from 48.3% to 97.0%, which is mainly due to NH₃-N having a large comprehensive decay coefficient (0.063 per day); however, the decrease in the NH₃-N concentration is lower when the water diversion duration is much longer. For example, the diversion duration of scheme 1, scheme 2, scheme 3, and scheme 4, as well as the average concentration values of NH₃-N, increased gradually at the end of the simulation because long-term water diversion weakened the water's capacity to self-purify NH₃-N. To further analyze the effects of water diversion discharge, water diversion duration, and the operation of sluices on each water quality index, range analysis was used to analyze the average concentration of COD (Fig. 18). Analysis results are shown in Table 14.

Table 14.

Range Analysis of the Chemical Oxygen Demand Average Concentration in Route 1

Case	Factor A	Factor B	Factor C	Average concentration of COD (mg/L)
1	1	1	1	14.461
2	1	2	2	13.613
3	1	3	3	13.503
4	1	4	4	13.784
5	2	1	2	13.138
6	2	2	1	11.215
7	2	3	4	11.731
8	2	4	3	11.601
9	3	1	3	9.770
10	3	2	4	8.532
11	3	3	1	10.145
12	3	4	2	12.343
13	4	1	4	9.294
14	4	2	3	8.441
15	4	3	2	10.401
16	4	4	1	12.809
Y1	55.361	46.662	48.629
Y2	47.686	41.800	49.495
Y3	40.789	45.781	43.315
Y4	40.944	50.537	43.341
y 1	13.840	11.666	12.157
y 2	11.921	10.450	12.374
y 3	10.197	11.445	10.829
y 4	10.236	12.634	10.835
R	3.643	2.184	1.545
Optimal level	A3	B2	C3

The effect of the three factors on the average concentration of COD follows the order of A > B > C; water diversion discharge is the most important, followed by water diversion duration and operation of the sluice. The ranges of the other water quality indexes are obtained through the same method, and the different effects of three factors on the concentrations are obtained. As seen in Fig. 19, the average concentrations of TP, TN, and NH₃-N are mostly affected by water diversion duration (50%, 55.7%, and 52.7%, respectively). In general, the effect of water diversion on Haihe River water quality is more important than that of the sluice operation. Therefore, management departments should give priority to water diversion discharge and water diversion duration when they choose water cycle schemes.

FIG. 19.

Percentage of range analysis results of four water quality indexes.

As seen in Table 14, A3B2C3 represents the optimal combination of COD. Theoretically, the average COD concentration should be minimized when the optimal combination is simulated. To validate this result, the optimal combination is simulated and the average concentration of COD is found to be 8.418 mg/L, the smallest value observed across the 16 schemes.

Optimal combinations of TP, TN, and NH₃-N in route 1 are A2B1C4, A2B1C4, and A1BAC4, and the average concentration values are 0.107, 0.723, and 0.018 mg/L, respectively.

The average concentrations of TP, TN, and NH₃-N calculated from each optimal combination, however, are not the smallest, as model simulation is a very complex process. Water diversion discharge, water diversion duration, and sluice operation have interdependent relationships in the model. Nevertheless, the orthogonal experiments are mathematical methods that do not fully take into account all relationships among factors and physical processes.

To overcome the shortcomings of the orthogonal experiment, the scheme that makes the concentrations of pollutants lower is selected as the optimal scheme by comparing the results of optimal combinations with those of the 16 schemes.

The optimal schemes are shown in Table 15.

Table 15.

Optimal Schemes of Each Index in Each Route

		Water diversion discharge (m³/s)	Water diversion duration (days)	The highest water level of the up-sluice of Erdao Sluice (m)
Route	Pollutant	Optimal case			Average concentration (mg/L)
1	COD	20	30	−0.110	8.418
	TP	10	15	−0.500	0.101
	TN	24	15	0.390	0.601
	NH₃-N	6	15	−0.610	0.016
2	COD	20	30	0.390	8.213
	TP	10	45	0.390	0.102
	TN	24	15	0.390	0.608
	NH₃-N	6	15	−0.610	0.017
3	COD	20	30	−0.110	8.334
	TP	10	30	−0.610	0.106
	TN	24	15	0.390	0.641
	NH₃-N	10	15	−0.500	0.016
4	COD	20	30	0.390	8.278
	TP	6	15	0.390	0.104
	TN	24	15	0.390	0.608
	NH₃-N	6	15	0.390	0.016
5	COD	10	45	0.390	5.603
	TP	10	45	0.390	0.072
	TN	10	15	0.390	0.168
	NH₃-N	6	15	0.390	0.016
6	COD	10	30	0.390	4.801
	TP	6	60	0.390	0.076
	TN	6	30	0.390	0.223
	NH₃-N	6	15	−0.610	0.018

According to the statistical results, the optimal schemes of water quality and quantity that minimize the concentrations of TP, TN, and NH₃-N all occur in cycle route 5, and the optimal scheme that minimizes the concentration of COD occurs in cycle route 6. Although the concentration of COD in route 5 is not smaller than that in route 6, the difference is small; the concentration of COD in route 5 is far smaller than those in other four routes; as a consequence, route 5 is considered as the optimal cycle route.

Discussion

Previous studies on the optimization of hydrodynamic and water quality simulation focused on situations in which a single factor was controlled (Belayneh and Bhallamudi, 2012; Yang et al., 2013; Kunz and Wüest, 2014). Although the single-factor control schemes make simulation and parameter adjustment more feasible, the potential for improving water quality is far less compared with schemes that take into account multiple factors. Therefore, we study the influence of multiple factors on water quality optimization for more precise results.

This study combined the one-dimensional hydrodynamic and water quality models with orthogonal experimental design and evaluated water diversion discharge, water diversion duration, and the Erdao Sluice operation. This approach has rarely been applied to hydrology and water resources field in China. The one-dimensional hydrodynamic and water quality model was established based on measured data. Orthogonal experimental design was additionally used to optimize 16 schemes, representing an innovation. The orthogonal experiment simultaneously studied the effects of multiple factors on water quality and chose representative schemes from all the combinations objectively, thereby decreasing work load and increasing efficiency.

Results of this study have practical implications for water quality improvement. Management departments can prioritize the control of certain severe pollutants according to the findings reported here to improve Haihe River water quality to a greater degree in a limited time. These results may also provide a scientific basis and technical examples to ensure a circle of water quality and aquatic ecosystem in plain river networks.

Although the combination of models and the mathematical methods offers an effective way to optimize cycle schemes, it also requires a large amount of data. Data precision and density have a direct influence on the degree of accuracy of pollution simulations. Through data collection in this study, we found that water quality data in China were available for short periods of time, with long intervals between collections (2 weeks to 2 months) (Chang et al., 2016). In addition, water quality is greatly affected by endogenous pollution (Xing-Hua et al., 2010; Linden et al., 2012); however, this study did not fully consider the effects of endogenous pollution, owing to a lack of data. The optimal scheme mainly controls for external pollutant concentrations to improve water environment. A more comprehensive study in this respect should be undertaken in the future.

Conclusion

This study analyzed water quality and proposed how water diversion projects and water cycle routes could improve the water quality of the mainstream and branches in Haihe River.

The one-dimensional hydrodynamic and water quality model of the Haihe River was established using measured values. In the validation, simulated values were found to be in good agreement with measured values.

The orthogonal table (L₁₆ (4³)), which took into consideration water diversion discharge and duration and the operation of Erdao Sluice, was used to optimize water quality and quantity cycle schemes.

According to range analysis of the evaluation index, water diversion discharge and duration were the main factors that affected average concentrations of pollutants; the operation of Erdao Sluice was a secondary factor.

The optimal water diversion scheme was obtained from the orthogonal experimental table and the model, and route 5 was determined to be the optimal route by comparing concentrations in the optimal schemes.

The combination of one-dimensional hydrodynamic and water quality model with orthogonal experimental design can provide an efficient method to simulate and optimize schemes to improve water quality through water flow regulation in plain river network regions.

Footnotes

Acknowledgments

The Major Science and Technology Program for Water Pollution Control and Treatment of China (2014X07203-009) and the Natural Science Foundation of China (51779165) supported this study.

Author Disclosure Statement

No competing financial interests exist.

References

Bach

P.M.

, Rauch

, Mikkelsen

P.S.

, Mccarthy

D.T.

, and Deletic

(2014). A critical review of integrated urban water modelling-urban drainage and beyond. Environ. Model. Softw., 54, 88.

Belayneh

M.Z.

, and Bhallamudi

S.M.

(2012). Optimization model for management of water quality in a tidal river using upstream releases. J. Water Resour. Prot. 4. 149.

Chang

, Sun

, Feng

, Zhang

, and Ge

(2016). Impacts of nonpoint source pollution on water quality in the yuqiao reservoir. Environ. Eng Sci., 34, 418.

Chen

H.Y.

, Shi

C.L.

, Wang

, and He

A.Z.

(2015). Discussion on water quality change trend and pollution control in Chengdu. Admin. Techn. Environ. Monit., 27, 35.

Chen

, Sun

R.H.

, Kong

P.R.

, and Chen

L.D.

(2016). Spatial distribution characteristics of carbon, nitrogen and phosphorous and pollution status evaluation of sediments in the Haihe River Basin, China. Chin. J. Appl. Ecol., 27, 2679. (in Chinese with English abstract)

Environmental quality standards for surface water. (2002). China Environmental Science Press. Beijing, China.

Guo

, Li

Y.B.

, and Fu

(2008). Controlling factors of degradation coefficient on organic polluant in river. J. Meteorol. Environ., 24, 56. (in Chinese with English abstract)

Jin

, Gong

, and Wang

(2015). Application of hybrid coagulation microfiltration with air backflushing to direct sewage concentration for organic matter recovery. J. Hazard. Mater., 283, 824.

Julian

J.P.

, Wilgruber

N.A.

, Beurs

K.M.D.

, Mayer

P.M.

, and Jawarneh

R.N.

(2015). Long-term impacts of land cover changes on stream channel loss. Sci. Total Environ., 537, 399.

10.

Kunz

M.J.

, and Wüest

(2014). Impacts by dams: From water quality modelling to management optimization. River Flow, 72 Suppl 1, 35.

11.

W.T.

(2016). Study on Storm Runoff Process and LID's Influence on it in Plain Tidal River Network Area. (Doctoral dissertation, Tianjin University) (in Chinese with English abstract)

12.

Linden

L.V.D.

, Hipsey

M.R.

, Burch

M.D.

, and Brookes

J.D

. (2012). Myponga Reservoir, South Australia: The influence of nutrients, phytoplankton, pathogens, and organic carbon on water quality. In: Bengtsson

, Herschy

R.W.

, and Fairbridge

R.W.

, Eds., Encyclopedia of Lakes and Reservoirs. New York: Springer-Verlag New York Inc.

13.

Makaske

(2001). Anastomosing rivers: A review of their classification, origin and sedimentary products. Earth Sci. Rev., 53, 149.

14.

Mccarty

P.L.

, Jaeho

, and Jeonghwan

(2011). Domestic wastewater treatment as a net energy producer-can this be achieved? Environ. Sci. Technol., 45, 7100.

15.

Meng

H.F.

, Xu

Y.P.

, Xu

G.L.

, and Zhang

X.Q.

(2014). Study on rivers connectivity evaluation in plain river network area. Resour. Environ. Yangtze Basin, 23, 626. (in Chinese with English abstract)

16.

MIKE 11 (2004). A Modeling System for Rivers and Channels Reference Manual. Danish Hydraulic Institute.

17.

Nanson

G.C.

, and Knighton

A.D.

(1996). Anabranching rivers: Their cause, character and classification. Earth Surf. Process. Landf., 21, 217.

18.

Putty

M.R.Y.

, and Prasad

(2000). Understanding runoff processes using a watershed model—A case study in the western ghats in south India. J. Hydrol., 228, 215.

19.

Pyka

, Jacobs

, Breuer

, Elbers

, Nacken

, and Sewilam

, et al. (2016). Effects of water diversion and climate change on the rur and meuse in low-flow situations. Environ. Earth Sci., 75, 1206.

20.

Rong

, Shan

B.Q.

, Lin

, Guo

, Zhao

, and Zhu

X.L.

(2016). Evolution of the nitrogen pollution in the Hai River Basin. Acta Sci. Circum. 36, 420. (in Chinese with English abstract)

21.

Sheng

, Xiang

, Guo

H.C.

, and Liu

(2013).Watershed optimal decision models for water-quality management. Acta Sci. Circum., 33, 1.

22.

Wang

(2007). The Runoff Yield and Concentration Simulation of River Net Region in the Plain. (Doctoral dissertation, Hohai University). (in Chinese with English abstract).

23.

Wang

Q.G.

, Zhao

X.H.

, Wu

W.J.

, Yang

M.S.

, Ma

, and Liu

(2008). Advection-diffusion models establishment of water-pollution accident in middle and lower reaches of Hanjiang River. Adv. Water Sci., 19, 500. (in Chinese with English abstract)

24.

Wang

X.Q.

(2015). Effect of the Three-Gorge Project im-poundment on the CODMn, NH3-N and TP fall coefficients. J. Saf. Environ. 15, 325 (in Chinese with English abstract).

25.

Wang

Y.L.

, Sun

Y.H.

, Zhou

Z.F.

, Zhang

J.K.

, and Zhang

Q.F.

(2006). Research on cod_(cr) degradation coefficient in the Lanzhou channel segment of the yellow river. Gansu Metal. 28, 27.

26.

Xing-Hua

D.U.

, Duan

D.X.

, Liu

, Jinan

, and Ming-Lei

, et al. (2010). Effects of bottom mud on water quality of ponds in yellow river delta. J. Yangtze Univ. Agric. Sci., 7, 32.

27.

Yang

, Qian

, Yubai

W.U.

, and Jin

(2013). Water quality optimization scheme for qinhuai river based on qual2k model. Water Resour. Prot., 29, 51.

28.

Zhang

(2015). Calculation of the water environmental capacity for plain river network area of Huzhou. Environ. Dev., 27, 85. (in Chinese with English abstract)

29.

Zhao

, Li

, Zou

, He

, Zhu

, and Liu

, et al. (2013). A three-dimensional water quality modeling approach for exploring the eutrophication responses to load reduction scenarios in lake Yilong (China). Environ. Pollut., 177, 13.

30.

Zhao

L.H.

, Li

T.L.

, and Lu

J.J.

(2003). 2-D depth averaged flow pollutions model for Jiangsu reaches in Yangtze River. J. Hydraul. Eng., 34, 72. (in Chinese with English abstract)

31.

Zhao

, Coles

N.A.

, and Wu

(2015). Status of heavy metals in soils following long-term river sediment application in plain river network region, southern China. J. Soils Sediments, 15, 2285.

32.

Zhou

, Xu

, Chen

, Xu

C.Y.

, Gao

, and Du

(2013). Hydrological response to urbanization at different spatio-temporal scales simulated by coupling of clue-s and the swat model in the Yangtze River delta region. J. Hydrol., 485, 113.

33.

Zhou

(2011). Study on Biological Restoration Technology in Drip Lake Diversion Channel. (Doctoral dissertation, East China Normal University). (in Chinese with English abstract)

34.

Zuo

Q.T.

, and Xia

(2002). Study on the modeling method of water quantity-water quality-ecology coupling system. J. Hydraul. Eng., 3, 61. (in Chinese with English abstract)