Simulation Study on Water Quality of Paddy Field Ditches Considering the Effects of Rainfall and Sediment Release

Abstract

As the main channel for the drainage of paddy fields, the water quality of canals directly affects the water quality of surrounding water bodies. Factors such as rainfall and the release of accumulated pollutants from the sediment of the canals have a direct impact on the water quality of drainage ditches. Based on the measured hydrological and water quality data, a dynamic water quality model was built to simulate the runoff and water quality changes in a paddy field for different return periods (1, 2, 5, 10, 20, and 50 years), and to consider the effects of released drainage sediment on the water quality of the paddy field. The change in water quality in paddy fields and ditches in different periods was studied. The simulation results showed that under different return periods, the total nitrogen concentration and total phosphorus concentration in the water of the paddy field and ditch increased first, and then decreased with time in June and July–September, while the pollutant concentration remained basically stable after the end of rainfall. With the increase in return period, the total nitrogen concentration and total phosphorus concentration decreased gradually. The release of nitrogen and phosphorus from the sediment of ditches resulted in an increase in the total nitrogen concentration and total phosphorus concentration in the ditches, and an increase in the pollution load. Under the influence of rainfall and fertilizer, the total nitrogen concentration and total phosphorus concentration in paddy ditch drainage were generally higher and more polluted in June than in July–September, and under the influence of released sediment, the ditch drainage was most polluted in June when the return period was one year, with the total nitrogen concentration and total phosphorus concentration at the outlet of the paddy ditch reaching 21.63 mg/L and 0.88 mg/L, respectively. The research results can provide a theoretical basis and basic support for the interception and treatment of non-point source pollution in farmland.

1. Introduction

Since the early 1980s, the water of the Taihu Lake Basin has been seriously polluted, and the phenomenon of eutrophication has been frequent [1]; this is mainly caused by serious agricultural non-point source pollution in the Taihu Lake region from farmland drainage. Currently, there is abundant research being conducted on farmland drainage. Hatcho et al. [2] found that it is possible to reduce the nitrogen load in paddy field drainage by means of circular irrigation. Tootoonchi et al. [3] monitored the water quality of paddy field drainage under four different flooding treatments and found that all flooding treatments reduced the phosphorus concentration in paddy field drainage. Varol et al. [4] found that paddy field drainage is the main source of nutrients in the downstream water body, and suggested that the use of fertilizer be controlled to reduce the nutrient load in paddy field drainage. Ghane et al. [5] found that the amount of nitrate migration in the drainage of unfertilized farmland was significantly lower than that of fertilized farmland, and the nitrate and total nitrogen load in the drainage of farmland increased correspondingly with an increase in flow depth. Badrzadeh et al. [6] found that fertilizer is the main factor affecting the pollution of farmland drainage, and the content of nitrate and phosphate in farmland drainage is also reduced by reducing the consumption of urea and phosphate fertilizer. Liu L. et al. [7] simulated the drainage of paddy fields and found that increasing the maximum standing water level could reduce phosphorus loss. By simulating the situation of nitrogen interception in farmland ditches with or without vegetation, Li X. et al. [8] found that the nitrogen interception rate in vegetated ditches was higher than that in non-vegetated ditches, and the amount of nitrogen interception was affected by the species and growth stage of plants in the ditches. Wang X. et al. [9] conducted a comprehensive evaluation of the drainage of farmland ditches, and found that the nitrogen index had a great impact on the water quality of ditches. Li S. et al. [10] analyzed the drainage of ecological ditches in paddy fields in the Taihu Basin, and found that plants in ditches and permeable dams played an important role in reducing nitrogen pollutant concentrations. Currently, the research on farmland drainage focuses on the analysis, control, and treatment of pollutants, and there is more research on field monitoring and less research on the simulation of drainage in paddy fields. However, the real-time monitoring of farmland drainage usually requires significant manpower and material resources, the monitoring may also face various problems, and it is difficult to study different drainage conditions. The use of mathematical model tools can easily solve many problems, and can contribute to the safe and sustainable use of farmland drainage.

The drainage of paddy fields ditches is mainly affected by rainfall. He S. et al. [11] found that the total nitrogen (TN) concentration in runoff induced by rainfall (the annual rainfall is 1468 mm) from paddy ditches in the Yangtze River Delta region of China ranged from 0.31 to 7.89 mg/L, and the total phosphorus (TP) concentration from 0.01 to 1.18 mg/L. Li X. et al. [12] found, by monitoring natural rainfall for two consecutive years (the annual rainfall is 544.4 mm and 433.5 mm), that the average concentration ranges of TN and TP in the drainage of a paddy field ditch in Dezhou City were 2.63–18.84 mg/L and 0.81–13.14 mg/L, respectively, and that the dominant form of nitrogen in the runoff was NO₃⁻-N, and the dominant form of phosphorus was particulate phosphorus. Wang L. et al. [13] conducted 17 rainfall (ranging from 7–101.7 mm) monitoring experiments on the ditches of the Baisha paddy field in Fuzhou, Fujian Province, China, and found that the runoff flux of natural rainfall was closely related to NO₃⁻-N, TP, and other losses, and the loss of nitrogen was greater than that of phosphorus. Yu Y. et al. [14] observed two typical natural rainfall events (with rainfall of 19.8 mm and 20.1 mm) in paddy field ditches in Jiangsu Province, and found that the water quality of ditches was seriously polluted during rainfall; the runoff carried a large amount of nitrogen at the early stage of rainfall, increasing its concentration until it reached its peak. Currently, simulating different rainfall conditions is one of the hot spots in the study of non-point source pollution, but there are few studies on the simulation of water quality in paddy field ditches. The drainage of paddy fields is the erosion of paddy fields by rainwater. The discharged water can enter rivers, lakes, and other water bodies, which has a great impact on downstream water bodies. Therefore, it is of great significance to study the water quality of paddy fields under different rainfall conditions.

Sediment release is also an important factor affecting water quality. Zhang Y. et al. [15] studied the release rules of nitrogen and phosphorus in sediments in shallow waters of the Yellow River Delta under different hydraulic actions, and found that sediment release increases the concentration of pollutants, and has a great impact on water quality. Komissarov et al. [16] found that the lake bottom sediments of small lakes in the Miyagi Prefecture contained about 3–4 times the Cs content of the surrounding soils, giving the overlying waters of the catchment area a rather high concentration of Cs. Xiang W. et al. [17] studied nitrogen and phosphorus pollution in the surface sediments of Baitan Lake, and found that the sediment release had a significant impact on the increase in nitrogen and phosphorus concentrations. Zhu Y. et al. [18] analyzed the contents of total nitrogen and ammonia nitrogen in the surface sediments of a reservoir in East China, and found that the sediment release was closely related to the increase in nitrogen concentration in overlying water. The drainage ditch of a paddy field is the only way for the drainage of the paddy field to enter the surrounding water. Long-term accumulation leads to large amounts of nitrogen and phosphorus in the sediment of the ditch. Therefore, the release of nitrogen and phosphorus in the sediment is bound to have a certain impact on the water quality of the ditch. Currently, there are relatively few studies using mathematical models to simulate sediment release in paddy field ditches, most of which focus on lakes, rivers, reservoirs, and other fields. As the connection channel between farmland water and downstream water, pollutants in paddy field ditches accumulate very easily in sediment, and it is easy to cause secondary pollution. Therefore, conducting simulations to study the effect of sediment release on water quality in the ditches of paddy fields can be helpful to understand the drainage pollution situation of paddy fields more accurately, and to put forward reasonable treatment measures.

In summary, the erosion of nutrients in paddy fields by rainfall and the release of accumulated pollutants in the sediment have a significant impact on the water quality of paddy field ditches. However, there is currently a lack of research on the drainage water quality of paddy field ditches under different rainfall conditions and considering sediment release. Therefore, the paddy field ditch in Donghu Village, Zhoutie Town, Yixing City, Jiangsu Province was taken as the research object in this paper. Under return periods of 1, 2, 5, 10, 20, and 50 years, the paddy field ditch drainage in the paddy field at different growth stages was simulated, and the influence of the sediment release was considered in analyzing the drainage water quality in the paddy field.

2. Materials and Methods

2.1. Overview of the Study Area

The study area is located in Donghu Village, Zhoutie Town, Yixing City, Jiangsu Province (north latitude 31°25′~39°27′, east longitude 119°57′~120°01′), which is located northeast of Yixing City, east of Taihu Lake and adjacent to the upper reaches of Hejiabang. It is warm and humid all year round, and belongs to the oceanic monsoon climate zone at the southern edge of the northern subtropics. The heat conditions are good, with a multi-year average temperature of 16 °C and an average temperature of 28.97 °C in the hottest month of summer. The average frost-free period for many years has been 223 days, and there is abundant precipitation, with an average precipitation of 1368.77 mm for many years, with concentrated rainfall in spring and summer. The multi-year average monthly rainfall is shown in Figure 1. The regional irrigation water is the branch water of Taihu Lake, and a rice–wheat rotation system has been implemented. In this area, rice is planted in May and June and harvested in November every year, and fertilization is carried out several times during the period when the rice needs nutrients. The paddy field has a drainage port, which is closed under normal circumstances, and opened during rainfall or other conditions that require drainage to discharge the water from the paddy field into the ditch. The overview of the study area is shown in Figure 2.

According to the investigation, the planting system and farming methods of rice in Yixing City, Jiangsu Province have not changed significantly in the last ten years [19]. However, due to the uneven distribution of fertilization and rainfall during the growing period of rice in the study area, their influence on the water quality of ditches varies greatly in different periods. In this region, rice is usually fertilized in late June, with about 300 kg/hm² of base fertilizer (in terms of N), and in early August, with about 150–225 kg/hm² of urea (in terms of N). According to meteorological data and on-site monitoring in consecutive years, the rainfall is higher from June to August. In actual monitoring, heavy rainfall often occurs in June, and moderate and heavy rainfall occurs more often from July to September. Since there are fewer rainfall days in June and more rainfall days in July–August, resulting in less rainfall in June than in July–August. Rainfall begins to decrease in September onwards, but in October, rice is managed alternately in wet and dry conditions, i.e., after irrigation, the field is allowed to naturally fall dry, and then is irrigated again, and then naturally falls dry, is irrigated again, falls dry again, and so on, 2–3 times until harvesting. Rainfall runoff is rarely monitored in October. Therefore, considering the influence of rainfall, fertilization, and other factors, in order to more accurately simulate the water quality of paddy field ditches, water quality models of paddy field ditches were built in June and July–September to explore the pollution status of paddy field ditches in different periods. Considering the accuracy of the model, the model was calibrated and validated using the observed data of different growth periods of rice in 2013 and 2014.

2.2. Sediment Release Test

The drainage ditch of a paddy field is the only way for its drainage to enter the surrounding water. Long-term accumulation leads to the accumulation of large amounts of nitrogen and phosphorus in the sediment of the ditch, and the release of nitrogen and phosphorus into the sediment inevitably has certain impacts on the water quality of the ditch. Therefore, this study fitted the formula of nitrogen and phosphorus release from the sediment of ditches through laboratory testing of nitrogen and phosphorus release from the sediment, and considered the effect of their release from the sediment on the water quality of ditches during the simulation process.

The overlying water and 0–10 cm of the sediment at the bottom of the drainage ditch in the paddy field in the study area were collected as test water and sediment to conduct the nitrogen and phosphorus release test. Due to the slow water flow in the ditch, 6 sampling points were evenly distributed along the ditch, as shown in Figure 3. The collected overburden water and surface sediment were brought back to the laboratory, and the residue, branches, and other foreign matter in the overburden water were filtered out using filter paper [20]. The concentrations of TN and TP in the overburden water at this time were determined as the initial concentrations, and the overburden water was taken as the original water sample.

The experimental device was a 1000 mL Shu Niu (GG-17) low beaker with a height of 154 mm and an outer diameter of 113 mm. The sediment was laid evenly on the bottom of the beaker, and the mud thickness was 3 cm; the water sample was added to the 1000 mL scale line, with each sampling position 3 cm below the water level. Each sample was 15 mL, and the original water sample was supplemented to the 1000 mL scale line.

When the experimental equipment and samples were arranged and the samples were taken every 24 h thereafter, the concentrations of TN (Nessermann reagent spectrophotometry) and TP (ammonium molybdate spectrophotometry) of the water samples were determined. The difference between the concentrations of TN and TP on two consecutive days was used as the nitrogen and phosphorus release concentration. The release concentration of nitrogen and phosphorus was found to be basically stable on the 6th day of the release test, and the test was ended.

Based on the nitrogen and phosphorus concentrations in the overlying water obtained from the nitrogen and phosphorus release test in the sediment, the difference between the concentrations of TN and TP on two consecutive days was taken as the nitrogen and phosphorus release concentration, and the variation trend in the nitrogen and phosphorus release concentration in the sediment of the ditch was fitted, and the fitting equation of the nitrogen and phosphorus released in the sediment was obtained, as shown in Figure 4.

It can be seen that the relationship between the release concentrations of TN and TP and time is consistent with the trend of a logarithmic function, and the correlation coefficient R² values are 0.9393 and 0.9457, respectively. It can be seen that the fitting formula fits the nitrogen and phosphorus release law of sediment well, and the formula can be used to simulate the effect of nitrogen and phosphorus released from sediment on ditch water quality. The formula for nitrogen and phosphorus release from sediment was embedded in the Ecolab module of the water quality model, and the water quality model of the paddy ditch was constructed considering the nitrogen and phosphorus released from sediment. The effects of sediment release on water bodies in the paddy field gully under different return periods were analyzed.

2.3. Model

2.3.1. MIKE 21 Model

As a two-dimensional water quality simulation software, MIKE 21 has the characteristics of convenient use, scientific algorithm, flexible parameter modification, fast calculation speed, and high simulation accuracy. It can load rainfall files or source and sink items files during the modeling process, which can make the simulation more practical. Therefore, the MIKE 21 model was selected to realize the two-dimensional hydrodynamic simulation and water quality simulation of the study area.

The MIKE 21 hydrodynamic model integrates the incompressible Reynolds mean Navier–Stokes equation under the Boussinesk approximation and the hypothesis of hydrostatic pressure to form the following two-dimensional unsteady shallow water equation:

$$\frac{\partial h}{\partial t}+\frac{\partial h\overline{u}}{\partial x}+\frac{\partial h\overline{v}}{\partial y}=hs$$

(1)

$$\begin{array}{l}\frac{\partial h\overline{u}}{\partial t}+\frac{\partial h{\overline{u}}^2}{\partial x}+\frac{\partial h\overline{u}v}{\partial y}=fh\overline{v}-gh\frac{\partial \eta }{\partial x}-\frac{h}{{\rho }_0}\frac{\partial p_a}{\partial x}-\frac{g{h}^2}{2{\rho }_0}\frac{\partial \rho }{\partial x}+\frac{{\tau }_{sx}}{{\rho }_0}-\frac{{\tau }_{bx}}{{\rho }_0}-\\ \frac{1}{{\rho }_0}\left(\frac{\partial s_{xx}}{\partial x}+\frac{\partial s_{xy}}{\partial y}\right)+\frac{\partial }{\partial x}\left(h{T}_{xx}\right)+\frac{\partial }{\partial y}\left(h{T}_{xy}\right)+h{u}_s\end{array}$$

(2)

$$\begin{array}{l}\frac{\partial h\overline{v}}{\partial t}+\frac{\partial h{\overline{v}}^2}{\partial x}+\frac{\partial h\overline{u}v}{\partial y}=-fh\overline{u}-gh\frac{\partial \eta }{\partial y}-\frac{h}{{\rho }_0}\frac{\partial p_a}{\partial y}-\frac{g{h}^2}{2{\rho }_0}\frac{\partial \rho }{\partial y}+\frac{{\tau }_{sy}}{{\rho }_0}-\frac{{\tau }_{by}}{{\rho }_0}-\\ \frac{1}{{\rho }_0}\left(\frac{\partial s_{yx}}{\partial x}+\frac{\partial s_{yy}}{\partial y}\right)+\frac{\partial }{\partial x}\left(h{T}_{xy}\right)+\frac{\partial }{\partial y}\left(h{T}_{yy}\right)+h{v}_{s}S\end{array}$$

(3)

where x, y are Cartesian coordinate system coordinates; u, v are velocity components in the x, y directions, in m/s; t is time, in s; d is the resting water depth, in m; η is the water level, in m; h is the total water depth, in m; f is the coefficient of Kohl’s force, f = 2ωsinφ, with ω as the angular velocity vector of the Earth’s rotation, and φ as the local latitude; g is the acceleration of gravity, in m/s²; s_xx, s_xy, and s_yy are the radiative stress components, in N; ρ is the density of water, in kg/m³; S is the source term; and u_s and v_s are the flow velocity components of the source term, in m/s.

$\overline{u}$, $\overline{v}$ denote the mean values of the velocity components in the x, y directions, determined by Equation (4) as follows:

$$h\overline{u}={\displaystyle \underset{-d}{\overset{\eta }{\int }}udz}, h\overline{v}={\displaystyle \underset{-d}{\overset{\eta }{\int }}vdz}$$

(4)

T_ij is the horizontal viscous stress term, including fluid viscous stress T_xx, turbulent stress T_xy, and horizontal convection T_yy, which is determined by Equation (5) as follows:

$$T_{xx}=2A\frac{\partial \overline{u}}{\partial x}, T_{xy}=A\left(\frac{\partial \overline{u}}{\partial y}+\frac{\partial \overline{v}}{\partial x}\right), T_{yy}=2A\frac{\partial \overline{v}}{\partial y}$$

(5)

The main significance of the water quality module from the MIKE 21 model is to study the diffusion pattern of pollutants in water, and the same applies to the state variables in Ecolab, which can be expressed by differential equations as follows:

$$\frac{\partial \left(hc\right)}{\partial t}+\frac{\partial \left(MC\right)}{\partial x}+\frac{\partial \left(NC\right)}{\partial y}=\frac{\partial }{\partial x}\left(E_{x}h\frac{\partial c}{\partial x}\right)+\frac{\partial }{\partial y}\left(E_{y}h\frac{\partial c}{\partial y}\right)+S+F\left(C\right)$$

(6)

where h is the water depth, in m; C is the pollutant concentration, in mg/L; M is the transverse single-width flow, in m²/s; N is the longitudinal single-width flow, in m²/s; Ex is the transverse diffusion coefficient, in m²/s; Ey is the longitudinal diffusion coefficient, in m²/s; S is the source–sink term, in g/m²·s; and F(C) is the biochemical reflection term.

2.3.2. Rainfall Generation Model

This study is based on the influence of rainfall runoff on the water quality of paddy fields and ditches; that is, the rainfall intensity is the main factor affecting different results. Therefore, different return periods (different rainfall intensities) were selected to simulate situations of 1, 2, 5, 10, 20, and 50 years. The runoff and pollutant concentrations generated in the simulated area are different under different rainfall conditions. Therefore, before carrying out simulations, it is necessary to consider the runoff and pollutant discharge in ditches under different return periods.

Choosing a suitable design rain type and rainstorm intensity formula is the key to designing rainfall drainage flow and pollutant discharge models. Currently, the most commonly used design rain type is Chicago rain type, which adopts the average rainfall intensity; the appropriate rainfall peak can be chosen, and the design is widely used in various drainage engineering design studies. Therefore, in this study, the Chicago rain pattern generator was used to generate rain patterns with different rainfall intensities. Meanwhile, according to the rainfall conditions in Jiangsu, the duration of rain patterns in different return periods was set to 8 h. The current urban rainstorm intensity formula of Jiangsu can be queried in Urban Drainage, Volume 5 of the Manual of Water Supply and Drainage Design, which was compiled by Tongji University using the following analytical method:

$$I=\frac{14.103\left(1+0.62\ln P\right)}{{\left(t+16.763\right)}^{0.671}}$$

(7)

where I is the rainfall intensity, in mm/min; P is the return period, in a; and t is the rainfall duration, in min.

To determine the rain peak coefficient, the rain peak coefficients of annual maximum rainfall process samples with the same duration are arithmetically averaged. Finally, the rain peak coefficient for each duration is obtained via the weighted average according to the corresponding duration length. In this paper, the peak rainfall coefficient for each return period was set to 0.5 based on related research at home and abroad, and local rainfall conditions in the study area. The rainfall conditions for each return period were calculated using Formula (7), as shown in Figure 5.

2.3.3. SWMM Model

SWMM 5.1 is a simulation software that can use rainfall to generate runoff and pollutant discharge in catchment areas. Through its hydrological model, SWMM 5.1 can deal with various hydrological processes when urban runoff occurs. The hydraulic module is used to simulate the unsteady flow of external water flow in water transmission systems such as pipelines, channels, storage and treatment facilities, and water distribution buildings. At the same time, the water quality module is used as an extension module to simulate the water pollution load generated in the process of runoff production and confluence.

SWMM 5.1 provides the classical Houghton equation, Green–Ampter equation, numerical method of runoff curve, and other computational equations to simulate infiltration. Among them, the classical Horton equation is more accurate in simulating rainwater infiltration in a small urban area, so this study chose the classical Horton equation as the calculation equation of infiltration. Its calculation equation is as follows:

$$f=f_1+\left(f_0-f_1\right)e^{-kt}$$

(8)

where f is the infiltration rate of soil, in mm/h; f₀ is the initial infiltration rate, in mm/h; f₁ is the stable infiltration rate; k is the infiltration attenuation coefficient, in h⁻¹; and t is the infiltration duration, in h.

Surface confluence refers to the process in which the surface runoff generated by each subcatchment area during the course of rainfall is collected to the outlet of the catchment area. The nonlinear reservoir method is used to calculate the surface confluence, and the equations are solved using the simultaneous continuity equation and the Manning equation. The continuity equation is as follows:

$$\frac{dV}{dt}=A\frac{dd}{dt}=Ai*-Q$$

(9)

where V is the total water storage, in m³; A is the subcatchment area, in m²; t is time, in s; d is the water depth, in m; i* is the net rainfall, in m; and Q is runoff discharge, in m³.

The SWMM model provides three calculation methods for water flow in a pipeline: the steady flow method, the moving wave method, and the dynamic wave method. Among them, the dynamic wave method is theoretically the most accurate in solving the complete Saint-Venant equations, which include the continuity and momentum equations in the conduit and the mass conservation equation at the node. The basic equation is as follows:

$$\frac{\partial A}{\partial t}+\frac{\partial Q}{\partial x}=q_L$$

(10)

$$\frac{1}{g}·\frac{\partial v}{\partial t}+\frac{y}{g}·\frac{\partial v}{\partial x}+\frac{\partial h}{\partial x}=S_0-S_f$$

(11)

where Q is the flow rate, in m³/s; A is the water crossing section area, in m²; q_L indicates the incoming flow per unit length, in m³/s. v is the flow rate, in m/s; h stands for hydrostatic head, in m; t is time, in s; x is the distance, in m; and S₀ is the slope of the bottom of the pipe and is the gravity term. S_f is the friction gradient and the friction term.

In the SWMM model, the accumulation of pollutants can be represented by the exponential scour function, proportional scour function, and the average concentration function. The exponential scour function was used in this study, and the formula is as follows:

$$W=A_1{q}^{A_2}B$$

(12)

where W is the scouring load, in mg/h; A₁ is the erosion coefficient; A₂ is the erosion index, d; q is the runoff rate per unit area, in mm/h; and B is the pollutant concentration, in mg/L.

2.4. Model Building and Validation

2.4.1. Ditch Grid Division

The CAD map was drawn according to the boundary shape of the rice field ditch in the study area, and the boundary map was imported into the SMS (surface-water modeling system). The SMS 10.1 software editing tool was used to divide the grid units in the area, and a triangular grid with high simulation accuracy was selected to divide the ditch into 740 grid units. As shown in Figure 6, the grid quality was checked to ensure the accuracy of the simulation results. Then, SMS and MIKE mesh conversion programs were used to convert the SMS output file into the corresponding mesh file. Finally, according to the actual geographical elevation of the research area, the elevation of the ditch area in the mesh file was interpolated using the MIKE editing function to complete the generation of the terrain file.

2.4.2. MIKE 21 Model Modeling and Parameter Calibration

The upstream boundary was selected as the flow boundary of the model. Considering that the flow of the upstream boundary changes due to the rainfall process, the flow was constructed as a time series file and imported into the upstream boundary. Similarly, the change in pollutant concentration along the upstream boundary with rainfall was also constructed as a time series file and imported into the upstream boundary. The downstream boundary was taken as the water level boundary of the model, and the water depth of the ditch was set at 1m according to the actual water depth of the ditch. The remaining borders served as land borders.

According to the actual situation of paddy field drainage outlets in the simulated area, four drainage outlets were generalized into the point source input model at the corresponding positions of ditches as source items affecting the simulated area, and the changes in the discharge and pollutant concentrations of each drainage outlet with rainfall were constructed into a time series file to import the model. The locations of drainage outlets are shown in Figure 7. The total time step was set to 8 h, and the time step to 10 min.

Dry and wet water depth: In order to ensure stable operation of the model, the default values recommended by the model were used for the dry and wet boundary conditions; that is, the dry water depth was 0.005 m, the submerged water depth was 0.05 m, and the wet water depth was 0.1 m.

Eddy viscosity: Eddy viscosity can reflect the phenomenon and degree of river turbulence. The eddy viscosity in this model was calculated using the Smagorinsky formula, and the eddy viscosity can be automatically adjusted according to the flow gradient.

Manning number: The Manning number can indicate the roughness of the river bed.

Degradation rate: The degradation rate reflects the rate of degradation of the pollutant in the water column.

Considering the influence of rainfall, fertilization, and other factors, water quality models of the paddy field ditches were constructed for June and July–September. The nitrification rate and denitrification rate were also set in the model. The calibration results of parameters such as eddy viscosity, Manning number, degradation rate, nitrification rate, and denitrification rate in the two water quality models from June and July–September are shown in

Table 1. Parameter calibration results.

Months	Eddy Viscosity	Manning Number	Degradation Rate	Nitrification Rate	Denitrification Rate
June	0.28	32 m1/3·s−1	0.03 d−1	0.05 d−1	0.05 d−1
July–September	0.28	32 m1/3·s−1	0.03 d−1	0.05 d−1	0.05 d−1

. After the model parameters were set, the model was calculated according to Formulas (1)–(6), and the results were obtained.

2.4.3. SWMM Model Building and Parameter Calibration

In this study, the SWMM model was used to simulate the drainage runoff and pollutant discharge of paddy fields in the simulated area under different rainfall conditions. The main time step was set at 10 min, and the distribution of catchment areas and outfalls was set according to the actual paddy fields in the study area, as shown in Figure 8. Catchment areas 2 and 3 are consistent due to the actual area and planting conditions, and the parameter settings in the model are also consistent. The parameter calibration results of each catchment area and land use pollutant are shown in

Table 2. Parameter calibration results of the catchment area.

Catchment Area	Area (m2)	Land Channel Width (m)	Slope (%)	Impervious Rate (%)	Land Use Area Percentage (%)
1	850	0.2	0.03	10	100
2	1400	0.2	0.03	10	100
3	1400	0.2	0.03	10	100
4	1900	0.2	0.03	10	100

,

Table 3. The calibration results of land use pollutant parameters in the SWMM model for June.

Parameters	Buildup	Max. Buildup	Rate Constant	Washoff	Coefficient	Exponent
TN	SAT	50	0.8	EXP	0.01	1.3
TP	SAT	10	0.4	EXP	0.01	1.2

and

Table 4. The calibration results of land use pollutant parameters in the SWMM model for July–September.

Parameters	Buildup	Max. Buildup	Rate Constant	Washoff	Coefficient	Exponent
TN	SAT	10	0.4	EXP	0.01	1.3
TP	SAT	3	0.2	EXP	0.01	1.2

. During the calibration process, trial and error settings were carried out according to the empirical values of model parameters [21,22,23]. The pollutant growth and erosion functions in the model were adjusted and determined by referring to farmland standards in the relevant literature. After the model parameters were set, the model was calculated according to Formulas (8)–(12), and the results were obtained.

2.4.4. Model Validation

The measured drainage velocity, pollutant concentration, and sediment release data of the study area during the rice growth period in 2013 and 2014 were selected to verify the water quality model of the ditch, the rainfall runoff model, and the ditch water quality model, considering rainfall and sediment release. The error formula was used to calculate the relative error, and the Nash efficiency coefficient (NSE) was used to calculate the model accuracy. The root mean square error (RMSE) was used to evaluate the performance of the model. The calculation formulas are as follows:

$$\delta =\frac{\left|{\tilde{y}}_i-y_i\right|}{y_i}$$

(13)

$$NSE=1-\left(\sum {\left(y_i-{\tilde{y}}_i\right)}^2/{\sum \left(y_i-\overline{y}\right)}^2\right)$$

(14)

$$RMSE=\sqrt{{\sum }_1^n{\left(y_i-{\tilde{y}}_i\right)}^2/n}$$

(15)

where δ is the relative error; NSE is the Nash efficiency coefficient; RMSE is the root mean square error; y_i is the measured value; ${\tilde{y}}_i$ is the simulated value of the model; $\overline{y}$ is the average value of the measured value; and n is the number of measurements.

The comparison of simulated and measured values for the water quality model of the ditch, the rainfall runoff model, and the ditch water quality model considering rainfall and sediment release is shown in Figure 9.

It can be seen from the figure that under different growth periods for rice, the simulated data and the measured data show high accuracy. According to the calculations from Formulas (13)–(15), the relative errors of simulated pollutant concentrations and measured pollutant concentrations, as well as simulated runoff and measured runoff, are all less than 15%, and the NSE is above 0.9. The RMSE values are all less than half of the standard deviation of the measured data. Therefore, the above models are representative and reliable, and can better simulate the hydrodynamic and water quality of the actual region.

3. Results and Discussion

3.1. Rainfall Runoff and Water Quality in Paddy Fields during Different Return Periods

The drainage runoff and water quality of paddy fields with return periods of 1, 2, 5, 10, 20, and 50 years were simulated, and the total water volume, total nitrogen, and total phosphorus emissions from paddy fields in 8 h under different return periods in the study area were obtained. The changes are shown in Figure 10 and Figure 11.

It can be seen from the figure that with the increase in return period, the total runoff also increases gradually. The emissions of total nitrogen and total phosphorus increased with the increase in return period, and there was a significant gap between June and July–September. During the return period, total nitrogen emissions ranged from 10.26 to 13.78 kg/ha in June, and from 0.85 to 4.17 kg/ha in July–September. Total phosphorus emissions ranged from 0.26 to 0.51 kg/ha in June, and from 0.16 to 0.40 kg/ha in July–September. The rainfall intensity of each return period in June was greater than that in July–September, and the erosion intensity of rainfall runoff was also greater, which made the pollutant concentrations in the ditches in June larger than those in July–September. This is because during the fertilization stage in June, the urea applied had not been transformed or the transformation was incomplete, and the nitrogen absorption from rice was also small. At this time, rainfall led to a large amount of nitrogen loss, along with surface runoff [24,25]. Moreover, the rainfall intensity in this period was also large, and led to serious nitrogen pollution in the ditch water in June.

3.2. Changes in Water Quality in Paddy Ditches during Different Return Periods

Changes in ditch water quality during six sets of different return periods were simulated. To analyze the differences in ditch water quality under different rainfall scenarios, the locations of sampling points 1, 2, and 3 in the ditch were selected, as shown in Figure 12, to analyze the changes in total nitrogen and total phosphorus concentrations over time. Sampling points 1, 2, and 3 can be regarded as the entrance of the ditch, the middle of the ditch, and the outlet of the ditch, which reflect the overall condition of the ditch more comprehensively. The simulation results are shown in Figure 13 and Figure 14. The pollutant concentrations at ditch sampling points 1, 2, and 3 were correlated with each return period, and the results are shown in Figure 15.

As can be seen from Figure 13 and Figure 14, in June and July–September, the concentrations of total nitrogen and total phosphorus at each sampling point in the paddy field canal first increased and then decreased with time, while the concentration of pollutants remained basically stable after the end of rainfall. The reason for this changing trend may be that at the beginning of rainfall, rainwater has a certain initial flushing effect on the paddy field, and more pollutants are washed and discharged into the ditches. With the progress of rainfall, the pollutants carried in the runoff gradually decrease, and the concentration of pollutants decreases rapidly with the dilution of cleaner rain, and then gradually stabilizes. Due to the influence of fertilization and rainfall, the total nitrogen concentration and total phosphorus concentration at all sampling points in June were higher than those in July–September. This is because the rain occurred after fertilization in June, and the water quality of the ditch was bound to be seriously affected by the effect of fertilizer nutrients that were not yet absorbed, and rain erosion. In the case of different return periods, with the increase in return periods, the concentration values of total nitrogen and total phosphorus at each sampling point in the channel at different times decreased gradually; that is, the peak concentration of pollutants was higher when the return period was small. This may be because when the return period increases, the runoff volume increases, and the pollutants cannot be completely and rapidly dissolved in the rainwater runoff and washed away [26,27]. In addition, when the rainfall runoff volume is large, the dilution effect may be more obvious, so the pollutant concentration decreases with the increase in return period. As shown in Figure 15, the larger the sphere and the brighter the color, the stronger the correlation. It can be seen that the concentration of nitrogen and phosphorus pollutants at each sampling point presents a strong positive correlation with the low return period, indicating that the low return period has a greater impact on the water quality of the ditch. The correlation between the concentration of nitrogen and phosphorus pollutants at sampling point 2 and each return period is stronger than those for sampling points 1 and 3. This is because sampling point 2 was not only affected by rainfall, but also by sewage discharge at nearby drainage outlets. During rainfall, rainwater flushes out pollutants in the paddy field from these drainage outlets and passes through sampling point 2 first. As a result, the concentration of pollutants at sampling point 2 is more obvious than that at sampling points 1 and 3.

3.3. Changes in Water Quality in Paddy Ditches Considering Rainfall and Sediment Release

Six sets of water quality simulations for ditches with different return periods were completed using the water quality model of paddy ditches considering rainfall and sediment release, and it was found that the impacts of sediment releases on the water quality of paddy ditches were very similar in different return periods. The results are shown in Figure 16, where changes in the water quality of paddy ditches with or without considering released sediment were compared using a 1-year return period as an example. The pollutant concentrations at sampling points 1, 2, and 3 were correlated with whether or not released sediment was considered in the ditches, and the results are shown in Figure 17.

It can be seen from Figure 16 and Figure 17 that when released sediment is considered, the total nitrogen concentration and total phosphorus concentration at each sampling point in the paddy field ditch still show a trend of first increasing, then decreasing, and finally stabilizing over time, which is consistent with the situation without sediment release being considered. The concentrations of total nitrogen and total phosphorus at each sampling point were significantly higher than those without considering sediment release. The concentrations of total nitrogen and total phosphorus at the ditch outlet in June and July–September were 21.63 mg/L and 0.88 mg/L, and 6.67 mg/L and 0.69 mg/L, respectively. It can be seen that sediment release has a negative impact on the water quality of ditches, which is caused by the diffusion of nutrient salts accumulated in the sediment. Since the interface between soil and water in ditches is permeable, the pore water carrying nutrient salts continuously discharges upward, and correspondingly increases the concentrations of nitrogen and phosphorus in the overlying ditch water. There was a positive correlation between all sampling points and whether sediment release was considered, but there was a stronger correlation between sediment release considered and sediment release not considered, indicating that sediment release had a greater impact on the water quality of ditches. Many studies have also proven that the pollution load is high when sediment is released, which will have a greater impact on the water body [28,29].

4. Conclusions

There are many factors affecting the water quality of paddy field ditches. In this research, the water quality of a typical paddy field drainage ditch in the Taihu Lake Basin was modeled under the influence of different periods, different return periods, and sediment release conditions from the ditch, and the following conclusions were obtained:

(1)

Under different return periods, the total nitrogen concentration and total phosphorus concentrations in the water bodies of paddy ditches in June and July–September both increased and then decreased with time, while the pollutant concentrations remained stable and unchanged by the end of the rainfall period, with the nitrogen and phosphorus concentrations in June being larger than those in July–September, resulting in more serious pollution. As the return period increased, the concentration values of both total nitrogen and total phosphorus in ditch drainage gradually decreased.

(2)

The sediment released from the ditch resulted in an increase in both the total nitrogen concentration and total phosphorus concentration in ditch drainage, and an increase in the pollution load.

(3)

From the sediment release results, considering sediment release was important for the actual water quality conditions, and made the simulation more consistent with the actual situation. From the results of the correlation between pollutant concentration and return period and whether or not sediment release at each sampling point should be considered, the pollutant concentration showed a strong positive correlation with a low return period, considering the sediment released.