Water quality improvement measures at the Yagang cross-section in the Pearl River Delta based on the calculation of excessive pollutant fluxes

The aim of this study was to quantify the sources of pollution in the Yagang River Basin. A 1-D hydrodynamic model and a 1-D water quality model were combined with the excessive pollutant flux analysis method to calculate pollution data of the Yagang area. The results showed that upstream pollution was the primary cause of water quality degradation for the Yagang Basin, exceeding the water quality standards. In addition, the pollution problem ranking of the entire basin was as follows: the Yagang area (30.4%)> the Foshan area (23.2%)> the Baini River Basin (13.1%)> the Liuxi River Basin (0.6%). In addition, the rainy season had the greatest influence on pollution concentrations. It was also concluded that if the boundary water quality could meet the inspection requirements (class IV water), and the internal research area sewage collection rate reached 60%, the ammonia-nitrogen (NH3-N) in the river discharge would reach 35.7%. This would allow the water quality at the Yagang cross-section to reach standard class IV.


GRAPHICAL ABSTRACT INTRODUCTION
Due to flat terrain and comfortable climates, plain areas are typically characterized by high population densities, developed industries and agriculture, as well as high urbanization (Song et al. ). However, water environment degradation in plain river networks caused by urbanization is becoming increasingly worse (Feng et al. ). In addition, a large number of gates and dams have been constructed in order to ensure adequate water supply. This has resulted in low flow rates, decreased river system connectivities, and insufficient pollutant degradation abilities (Deng et al. ). As a result, water pollution is aggravated. Therefore, there is an urgent need to identify the pollution sources and to propose water quality improvement measures. This study uses the Yagang cross-section in the Pearl River Delta (PRD) as an example. Therefore, an analysis of the source of this pollution and the proposal of effective measures is very important.
Currently, the primary methods that have been used to determine pollution sources in water environments are the chemical mass balance (CMB) method (Miller et al. ), the isotope tracer technique, and water quality modeling.
The CMB method is widely used for the analysis of the contribution of related pollution sources in the atmosphere (Cesari et al. ; Wang et al. ), but has not been frequently applied for water environments (Bao et al. ). Success in using the CMB method requires the identification of all sources that potentially could contribute to the study object, and this necessitates a source composition profile for each source in order to build a library (Gleser ). In addition, all of the source compositions should be measured exactly.
Thus, it is a challenge to identify nitrogen (N) and phosphorus (P) sources in plain areas, where point sources and nonpoint sources are highly complex (Kaushal et al. ; Yi et al. ).
With the development of the isotopic-based technique, the tracing of stable isotopes has become a primary tool for the identification of pollution sources. Over the last 50 years, nitrogen isotopes have been considered an important tool for the study of point and non-point sources of N contamination in water bodies in different regions (Kruk et al. ). Additionally, the isotope tracer technique has also been used to identify the sources of heavy metals in soils and determine their chemical behaviors (Huang et al. ). However, the isotope technique is expensive and time-consuming, and it is applicable for specific elements (for example, N). Therefore, it is difficult to conduct such studies in a plain river network.
Water quality modeling was then introduced for the analysis of pollution sources when monitoring data were insufficient (Huishu et al. ). For example, the export coefficient model (ECM) is applied in areas with flat terrains and complex river networks in order to predict the annual nutrient inputs to lakes and rivers and to analyze the primary contributors of pollution loads (Li et al. ). In addition, some numerical models, such as MIKE, EFDC, and the Delft 3D Model, have also been applied to provide estimations of pollutant fluxes and to calculate the mass balance of nutrients (Heeb et al. ), which is the basis for pollution source identification.
This study is based on hydrological water quality data from 2017. This is used to construct a complex river network mathematical model of the water environment combined with the excessive pollutant flux analysis method to simulate water quality changes at the Yagang cross-section. The aim of this study is to predict different affects of the pollution sources in different seasons and to propose an effective plan to provide a scientific basis for local governance in the future.

Study area
The Yagang cross-section is located on the Xi Channel, which flows through the bordering zone between By considering the current major pollution sources and the current water environment, a zone surrounded by five cross-sections (M1, M2, M3, M4, and the Yagang cross-section) was considered the most polluted area. This area was then defined as the study area. It is situated in the zone between 23 2 0 and 23 42 0 N and 112 53 0 and 113 43 0 E. As the Yagang cross-section is located in a tidal reach, it receives pollutants from the upstream area, but also pollutants from the downstream area during a high tide period. Therefore, in order to identify the pollution sources, the area that could affect the water quality at the Yagang cross-section was divided into five areas based on a division of the catchment section and the water environmental control units (Leach et al. a).
Hydrologic and water quality data, the model boundaries, and the location of the precipitation station were supplied by the Guangzhou Environment Protection Bureau and the Hydrology Bureau of Guangdong Province. These data consisted of the upstream area and the downstream area, and the upstream area was divided into the study area, which was the upstream area of the Liuxi River, the upstream area of the Baini River, and the southwestern area of Foshan City.

Study methods
To simulate the characteristics of reciprocating flow in the river network more accurately, this study combined the Saint-Venant equations to solve the finite difference method, namely, the solution area was divided into continuous grids. Hence, the approximate solution of the partial were considered the dependent variables. In addition, the model considered lateral source inflows and lateral sinks.
The calculation equations can be expressed as follows: where x represents the distance coordinate; Q represents the discharge; B represents the width of the section; B W represents the regulated width of the section; Z represents the water level; t represents time; q represents the lateral inflow; u represents the average velocity of the section; g represents the gravitational acceleration; A represents the area of the section; n is the roughness coefficient; and R represents the hydraulic radius.

The 1-D water quality model
A 1-D water quality model was constructed on the basis of the mass equation and the momentum equation, which suggest that the pollutant is mixed completely in the section (Chen et al. ). It can be calculated using the following equation: where x represents the distance coordinate; t represents time; u represents the average velocity of the section; C represents the pollutant concentration; Ex represents the convection diffusion coefficient; and K represents the attenuation coefficient.

Excessive pollutant flux analysis
The Yagang section is situated on a tidal reach, and the pol- where W excessive represents the excessive pollutant flux (t/a); C i represents the average pollutant concentration over one day at the Yagang cross-section (mg/L); C s represents the water quality standard of the Yagang section (mg/L); and Q i represents the simulated discharge by the 1-D model at the Yagang cross-section (m 3 /d).
The weight influence of a pollutant was calculated according to the proportional relation, which can be expressed using the following formulae: where α downstream area and α upstream area represent the weight influence of the pollutants from the downstream area and the upstream area, respectively; α p represents the weight influence of the pollutant from the different areas in the upstream area, except the study area (the upstream area of the Liuxi River, the upstream area of the Baini River, and the southwestern area of Foshan City); α 0 represents the weight influence of the pollutant from the study area; W p, excessive represents the excessive flux of the pollutants from different areas except the study area (t/a); and W 0 represents the pollutant discharge exceeding the water environment capacity, which is equivalent to the difference between the current emissions and the water environment capacity (t/a).

Model setup
The primary rivers (the Liuxi River, the Baini River, the Xi Channel, the Xinan River, as well as the Qian Channel) were generalized using 42 cross-sections. All the cross-

Model calibration
The model was calibrated based on the water level data from the Yagang cross-section in January, April, and July 2017.

)
: where N is the number of times of the total simulation; i is the number of times of the simulation; S i is the value of the i th simulation; M i is the value of the i th measurement; S is the simulated average value; and M is the measured average value.
The assessment results of the four stations (M1, M2, M3, and M4 in Figure 1) showed that the simulated water levels fit well with the measured water levels, and the RMSE was less than 5 cm. In addition, R 2 of the water qualities were all greater than 0.90 (Table 1). The simulation results accounted for more than 90% of the actual situation. As a result, the constructed water quality model met the requirements for subsequent water quality research.

Results of the pollution flux for the different periods
In this study, the period from December to March was considered the dry season (less rainy season), the period from  The influence of the different areas in the upstream area on the water quality at the Yagang cross-section will be further discussed according to the method of excessive pollutant flux in the following section.

Weight influence of pollutants from the upstream area and the downstream area
The weight influence of pollutants from the upstream area and the downstream area were calculated according to   Equations (6) and (7). As shown in Figure 5, the influence of the upstream area during the wet season was much greater than that in the normal and dry seasons. Previous studies show that the climate in the Pearl River Delta resulted in highly seasonal variations in water discharges (Xuan et al. ). Furthermore, during the wet season (April to September), high precipitation and runoff resulted in a greater loading of nutrients (Ou et al. ). However, during the dry season (October to March), decreased river discharges and strong northeasterly winds led to a well-mixed water column (Ye et al. ). In this study, it was assumed that the inflow discharge was the greatest during the wet season, and it would then carry a large amount of nonpoint pollution, which would lead to a greater influence.
According to the 2017 water quality monitoring data, the primary pollution factor at the Yagang cross-section was NH 3 -N, and the excessive multiple was 0.32 (the water quality standard of NH 3 -N is equal to 1.5 mg/L). With regard to NH 3 -N, the pollutant flux from the upstream area accounted for 67.3% for the entire year, and the downtown stream area accounted for 32.7%. Therefore, the water quality of the Yagang section could not meet the standard due to the large pollution load from the upstream area, which was the same as discussed above.
Excessive pollutant fluxes of the different areas in the upstream areas The excessive pollutant fluxes in the four areas (the southwestern area of Foshan City, the upstream area of the Baini River, the upstream area of the Liuxi River, as well as the study area) were calculated based on Equation (5), and then the contributions from the areas were estimated.
According to the water quality monitoring data, the concentration of COD did not exceed the water quality standard, and the excessive pollutant flux of COD was equal to 0. The results for NH 3 -N and TP are shown in Figure 6.
The contribution of the different areas varied significantly depending on the water period. The excessive pollutant flux from the study area was the largest for NH 3 -N, which

Water quality improvement measures
According to the excessive pollutant flux analysis, the Yagang section suffers not only from pollutants from the upstream area but also from the downstream area. Detailed measures to improve the water quality are discussed in the following section. pipeline network and improving the sewage treatment process.

Pollution reduction in the study area
By considering the recent inlet and outlet concentrations of NH 3 -N from sewage treatment plants, the NH 3 -N reduction rate of the pollutant discharged into rivers was calculated for different sewage collection rates (Table 3).  In the different scenarios, the concentrations of COD and TP at the Yagang cross-section were always able to meet the

Water quality standard
Water quality standard

Maintaining current situation
Water quality standard study area (30.4%) > the southwestern area of Foshan City (23.2%) > the upstream area of the Baini River (13.1%) > the upstream area of the Liuxi River (0.6%). In order to propose water quality improvement measures, a series of scenarios were simulated using the 1-D model. The results indicated that, to meet the water quality standard of the Yagang crosssection, the water quality of the inflow from Foshan City, the Baini River, as well as the Xi Channel should be improved, and the NH 3 -N pollutant discharged to the rivers in the study area should be reduced by more than 35.7% by increasing the sewage collection and processing rate.
Additionally, much attention should be paid to reduce the non-point source pollution caused by farmlands in the upstream area of the Baini River. In general, the calculation of excessive pollutant fluxes can serve as a method to identify pollution sources and divide pollution treatment responsibilities. This modeling can assist in the environmental management of the plain river network.