Study and Application of Estimation Method for Environmental Background Pollution Fluxes into the River in Source Areas

Abstract

The background values of pollutants are becoming increasingly prominent in evaluating water quality and management, and their load and output characteristics are of great significance to the study of changes in the background values. Thus, this paper studies the estimation method for background pollutant fluxes into a river in a forested river source in northeastern China with deficient data or without data. Based on the results of a soak experiment and leaching experiment, as well as natural rainfall runoff experiments, the conversion relationship of the pollutant concentration in the medium was established, the transformation relationship of the pollutant concentration in the medium was established, its load into the river in a small watershed was estimated, and the estimation method was verified according to the measurements (average relative error: 20.5%). Then, the background pollutant loading into the river was calculated on a large scale by improving the pollutant output coefficient model and the universal soil loss equation (average relative error: 18.7%, 24.7%). The results show that the method mentioned above can better reflect the background pollutant loading into the river. This study provides an effective strategy to estimate the background pollutant loading into rivers in areas without data or with missing data, and also provides a theoretical basis for the study of zoning and the formulation of a water environment evaluation system based on background values.

1. Introduction

The surface water background value, highly associated with water quality assessment, has gained increasing attention from local governments and ecology and environment departments. The China National Environmental Monitoring Centre has repeatedly emphasized its significance at national and regional conferences on water quality and ecology. Due to the lack of targeted and systematic research, its formation mechanism in different regions still remains unclear, and the method used to apply it to evaluate water quality is still very vague. Technical regulations on the environmental background determination of surface water and groundwater (provisional) [1] were issued by the Ministry of Ecology and Environment of the People’s Republic of China in 2019 to define an environmental background value and explain how to determine and apply it. The technical scheme for the water quality standard evaluation of surface water functional areas [2] issued by the Ministry of Water Resources of the People’s Republic of China in 2014 mentions that the water quality standard of water functional zones affected by background values shall be evaluated based on the revised data. Nevertheless, considering the differences in the region, index, and its mechanism, the quantitative method is not given in the two regulations.

The investigation on environmental background values in China was first conducted in the 1990s. During this period, a large number of relative studies with a certain scale emerged, involving the Lijiang River, Songhua River, Xiangjiang River, Dongting Lake, Jialing River, etc. Most of the studies focused on heavy metals and soil sediments in water and yielded fruitful achievements [3,4,5,6,7]. Recently, researchers have been paying more attention to the effects of human activities on the environment, while few report on the environmental background pollutant values in China. In 2018, an author enrolled in research on the surface water background values at the sources of rivers in northeastern China. Taking Heilongjiang Province as an example, it was found that the concentrations of chemical oxygen demand (COD), permanganate index (COD_Mn), and a high level of ammonia nitrogen (NH₄-N) in the surface water affected the environmental background. These have significantly disturbed the water quality evaluation results in this river and its downstream [8].

Combining previous experience and the regional characteristics, our team carried out unprecedent research on environmental background values in the Tangwang River Basin in Heilongjiang Province in northeastern China [9]. We studied the impacts of pollutants in the background area on the water quality of downstream rivers. The problem with the surface water background value in Heilongjiang Province was demonstrated [10]. Relying on the water function zones, the background and non-background areas in the river source reserves were identified and divided [11], thus providing a research basis for studying the environmental background value. A background value database for the river source was established through continuous and extensive data monitoring. Then, a new method of water quality evaluation under the influence of the background value was proposed [12,13].

Through several years of monitoring and research, we proposed a technical method for identifying the background area in river sources in northeastern China, identified the types and sources of background pollutants, studied the methods for eliminating the influence of background values, established a background value database through water quality monitoring in the past 3 years, determined the characterization method for the background value range, and proposed a water quality standard that meets the requirements of the standard evaluation method. A degradation model was used to simulate the degree and scope of the impact of pollutants on the downstream river. The above actions were all conducted to eliminate the impact of background values on water quality management and evaluation. However, how are background pollutants formed in a river? What is the pollutant load in the background area? How is it lost from the source? What is the migration and transformation from the land to rivers? What is the output process of pollutants? These issues were not clarified in the preliminary work; however, they are necessary for us to understand the background values resulting from our systematic research. Therefore, the typical areas were selected to perform in situ experiments to investigate the formation mechanism of background pollutant values [14].

It must be acknowledged that the lack of long-term measured data, such as hydrological and water quality data, limits the application of the non-point source mechanism model in a background area. The output coefficient model has been applied to estimate non-point source pollutants due to its simple calculation method, fewer required parameters, and its ability to generalize the complex process of non-point source pollution occurrence. However, there are significant differences in the pollutant output coefficients among different regions. Therefore, we combined empirical models (also known as statistical models) to determine the pollutant output coefficients through in situ-measured data from a small watershed. This method is based on a statistical analysis of hydrological, water quality, and rainfall data, establishing a correlation between non-point source pollution load and various factors in the watershed, and constructing empirical formulas to calculate non-point source pollution load through a regression analysis. It is generally applicable to small watersheds with relatively simple internal structures, as the relationship between rainfall, runoff, and pollution load in small watersheds is relatively simple, mostly linear or nonlinear.

Based on the above methods, the typical areas were selected to perform in situ experiments to investigate the formation mechanism of background value [14]. Based on those results, this paper puts forward a set of estimation methods for background pollutants loading into source rivers in northeastern China. The purpose is to promote the research division in China to formulate water environment evaluation methods under the influence of background pollutant values.

2. Materials and Methods

2.1. Study Area

The Kamalan River is the second largest tributary in the river source water reserve of Huma River, located in the hinterland of Greater Khingan Range. It flows into Huma River from west to east with a flow of about 1/10th that of Huma River (Figure 1), with the catchment range of 1264.7 km² and length of about 94.3 km². The watershed is dominated by forest and grassland, accounting for 96.4%; the vegetation type is mainly Larix gmelinii, accounting for 75.2%; and the soil type is mainly brown coniferous forest soil, accounting for 71.4%. During the experiment, the sampling and experimental sites were A1~A3. A1 was the outlet section of the Kamalan River Basin, A2 was the sampling point under the forest, and A3 was the outlet of the No. 2 Bridge sub-watershed.

2.2. Study Methods

2.2.1. In Situ Experiment

The strength of the source was analyzed by conducting an experiment with a soaking solution in the forest soil humic layer and litter layer. The concentration of background pollutants in the soaking solution was measured to determine the maximum possible loss in these layers. The litter (soil humus layer) was soaked in distilled water for 48 h with a mass ratio of 1:4. It was then vibrated for 3 min and filtered using two layers of filter paper with a Brinell funnel until the filtration became transparent. The COD, COD_Mn and NH₄-N were subsequently detected following the Technical Code for Monitoring of Surface Water and Sewage [HJ/T91-2002] [15]. Sample details can be found in

Table 1. The sample information for experiments involving source immersion and leaching analysis.

Experiment	Type	Quantity	Remarks
			Sampling Sites	Sampling Time
Source strength experiment	Litter layer	9	A2	20190330; 20190808; 20191220; 20200320; 20200624; 20200701; 20200719; 20200723; 20200819
	Soil humus layer	4	A2	20190808; 20190320; 20190624; 20200819
Leaching experiment	Litter layer	5	A2	20200613; 20200617; 20200624
	Soil humus layer	5	A2	20200719; 20200723; 20200730; 20200801

.

The device for the litter leaching solution is depicted in Figure 2 (left, patent authorized). It consists of a funnel filter with an upper cylinder (diameter 0.5 m, height 0.2 m) and a lower cone (height 0.3 m). The filter screen, which collaborates with the inner wall of the filter, forms a receiving cavity. The opening at the first end of the filter serves as the receiving cavity, and the bottom of the open end is connected to a guide pipe. The other end of the guide pipe is connected to a 2 L water collector, which collects the litter solution.

The device for the soil humic layer leaching solution is illustrated in Figure 2 (left, patent authorized). It consists of a cattle tendon box (3 × 2 × 1.5 m), diversion pipe, rainfall equipment, elevator and diversion pipe. A water outlet is provided on the side wall of the box, which is connected to the water collector through a diversion pipe. The soil humic layer is laid inside the box following the original soil structure. The A2 arrangement ensures uniform distribution of the device. The water sample collected after natural rainfall in the collector represents the loss of natural background pollutant source strength.

2.2.2. Estimation Method

The estimation method for background pollutants in a small watershed is based on the characteristics of pollution loss, migration, and transformation during the process of rainfall runoff. The pollutant load characteristics of the background area are estimated using on-site measured data from soak experiments. The pollutant concentration into the river coefficient is estimated based on the leaching experiments with the litter and soil humus layers, and the pollutant load output coefficient is estimated based on the measured runoff coefficient. The annual output of pollutants in the background area is then estimated.

The estimation of pollutant flux into a river in a large-scale watershed is based on experimental results, combined with an improved output coefficient model and an improved universal soil loss equation.

Using the Johnes classical output coefficient model as a basis [16,17], the flux of characteristic background pollutants in the No. 2 bridge sub-watershed was estimated by improving the parameters of the statistical model. The Kamalan River Basin was divided into 33 sub-basins, which were used as control units to extract the rainfall and slope factors.

$$L=\sum _{i=1}^n\alpha \beta E_{ij}{A}_i+P$$

$$\alpha =\frac{{Pre}_i}{\overline{Pre}}$$

$$\beta =\frac{{Slo}_i}{\overline{Slo}}$$

The equation above was used in this study to estimate the total amount of background pollutants into the river (L), where E_ij represents the output coefficient of j pollutants of class i land use type (kg·km⁻²·a⁻¹), A_i represents the area of class i land use (km²), p is the amount of pollutants input by precipitation (kg). α denotes the rainfall factor, β is the slope factor, ${Pre}_i$ denotes the rainfall in the sub-watershed i, $\overline{Pre}$ denotes the average rainfall in the whole basin, ${Slo}_i$ is the slope of the i sub-watershed, $\overline{Slo}$ is the average slope of the whole basin.

The non-point source pollutants in the Kamalan River Basin are primarily caused by rainfall-runoff erosion of the underlying surface. Soil erosion acts as the main carrier of non-point source pollution into the river. However, due to the high forest coverage in the basin, background pollutants primarily enter the river through organic erosion on the underlying surface rather than soil erosion. Therefore, in this study, the model used to estimate the total amount of background pollutants takes into account the universal soil loss equation and the characteristics of land-based organic erosion [18,19].

$$A=R·K·LS·C·F$$

$$R=\sum _{i=1}^{12}\left(1.735\times {10}^{\left(1.5lg\frac{P_i}{P}-0.8188\right)}\right)$$

$$K=7.594\left(0.0034+0.0405exp\left(-\frac{1}{2}{\left(\frac{log\left(D_g+1.659\right)}{0.7101}\right)}^2\right)\right)$$

$$D_g\left(mm\right)=exp\left(0.01\sum f_i{lnm}_i\right)$$

$$LS={\left(0.045L\right)}^a\left(65.41sin\theta +4.56sin\theta +0.065\right)$$

$$C=0.6508-0.3436lgc$$

The variables used in the equation are as follows: A represents the erosion amount of non-point source organic pollutants t/(km²·a), R is the rainfall-runoff factor, K is the organic pollutants erodibility factor t/(km²·a), LS represents the slope length and slope factor, C is the vegetation and management factor, F is the soil and water conservation measure factor, p is the total annual rainfall, P_i is the average monthly rainfall, D_g is the geometric average particle size, f_i is the percentage of particle-size composition in the original soil, m_i is the arithmetic mean value less than the particle size, and θ represents the slope. The value ‘a’ is determined based on the slope angle θ, where if $\theta >2$.86, a = 0.5, 2.86 ≥ $\theta$ ≥ 1.72, a = 0.4, for 1.72 ≥ $\theta$ ≥ 0.57, a = 0.3, and for θ ≤ 0.57, a = 0.2.

2.2.3. Validation Method

The estimation method for pollutant flux into rivers was validated through in situ monitoring at the outlet of a small watershed. The main objective was to continuously monitor water quality and flow during rainfall and determine the annual output by comparing it with the estimated results.

In the case of large-scale pollutant inflows into a watershed, the estimation method utilized measured data on water quality and quantity. In months without data, the relationship between runoff and pollutant inflows was interpolated to obtain the annual output of large-scale pollutants in the watershed. These results were then compared with the improved method.

2.3. Data Sources

From 2019 to 2020, samples of litter layer and soil humus layer were collected at A2 for a source strength experiment. A total of 13 samples were collected. At A2, an experimental device was set up for a leaching experiment, and 10 valid data points were collected, as shown in Table 1.

To conduct the experiment on natural rainfall runoff, the No. 2 Bridge sub-watershed with an area of 7 km² was the sampling point. The observation period lasted from June to August 2020. Water quality and quantity at the sub-watershed outlet were simultaneously monitored, with COD, COD_Mn and NH₄-N being detected. The water samples were collected in 1 L polyethylene bottles, dried, labeled and incubated. The flow rate was determined by multiplying the rectangular cross-sectional area by the flow rate. Rainfall data were obtained from the records of the outdoor environmental meteorological station.

During the experiment, the water quality and quantity at the outlet of the Kamalan River Basin were monitored simultaneously. The water samples were collected in 1 L polyethylene bottles, dried, labeled and incubated. The flow velocity was measured using portable current meter (ls1206b/ls300-A, XinYe, Qingdao, China). The water crossing section was calculated based on the measured data of the Kamalan River and the water level. The location and environment of the NO. 2 Bridge sub-watershed can be seen in Figure 3.

The land type was determined using Landsat 8 OLI satellite remote sensing image interpretation. The DEM data were obtained from GDEMDEM 30 m resolution digital elevation data, and vegetation type was based on China’s 1 million vegetation type spatial distribution data. The soil type was determined by the second national land survey.

3. Results and Discussion

3.1. Source Experiment

3.1.1. Soak Experiment

Surface water background values in the source water reserve of Heilongjiang Province are primarily influenced by the loss and leaching of organic matter in forest soil and litter [10].

Table 2. Characteristic value of background pollutant concentrations in source leaching experiment (mg/L).

Soak Experiment	Eigenvalue	COD	CODMn	NH4-N
Litter layer	Max	1043	380	3.64
	Average	736.5	275	6.31
	Min	628	192	8.26
Soil humus layer	Max	217	158	4.38
	Average	247.5	123.2	2.10
	Min	284	108.6	1.31

presents the detection results for background pollutant concentrations in the natural source strength, specifically the soaking solution of litter and soil humus layer. The soaking solution of litter exhibited a COD concentration ranging from 628 to ~1043 mg/L, with the maximum concentration being approximately twice that of the minimum. The COD_Mn concentration ranged from 192 to ~380 mg/L, while the concentration of NH₄-N ranged from 3.64 to 8.26 mg/L. On the other hand, the leaching solution of soil humus layer showed lower concentrations of background pollutants compared to the litter layer. The COD concentration ranged from 217 to ~284 mg/L, with an average of 247.5 mg/L. Similarly, the COD_Mn concentration ranged from 108.6 to 158 mg/L, with an average of 123.2 mg/L. The average value of NH₄-N was 2.1 mg/L.

It is worth noting that significant variations in surface water background values were observed across different regions. In the river source reserve of Heilongjiang Province, the primary source of surface water background values is the loss and leaching of organic matter in forest soil and litter. The source strength soak experiment revealed that the concentration of background pollutants in the litter soaking solution was significantly higher than that in forest soil. On average, the concentrations of COD and NH₄-N in the litter soaking solution were approximately three times those of forest soil, while the concentration of CODMn was approximately two times higher.

3.1.2. Pollution Load Characteristics

Litter serves as a crucial energy source within the ecosystem, as it can be decomposed and reused by organisms, thereby maintaining the nutrient balance in the forest ecosystem. In a study conducted by Shi Zhong Jie and Zhang Xin Ping, the annual amount in fallen leaves of coniferous forests was determined to be 21.7 t/(km²·a) [20,21,22]. Additionally, through the source strength experiment, it was observed that the content of background pollutants in the litter soaking solution represented the maximum possible loss from litter. This implies that the maximum possible loss per kilogram of litter was approximately COD: 2.95 g, COD_Mn: 1.1 g and NH₄-N: 0.025 g. Furthermore, for every square kilometer of litter, the respective values were 6.39 t, 2.39 t and 0.055 t.

Soil, on the other hand, serves as a nutrient reserve medium within the forest ecosystem, providing essential nutrients and energy for biological growth. It also plays a crucial role in material circulation [23,24]. In the Kamalan River Basin, the soil is predominantly covered by brown coniferous forests. The leaching layer (layer a) of the soil has a thickness of approximately 0.1 M, with a soil density of about 1.25 g/cm³. Similarly to the litter, the content of background pollutants in the soaking solution of the soil humic layer represented the maximum possible loss from the soil. In this case, the maximum loss per kilogram of soil was estimated to be COD: 0.99 g, COD_Mn: 0.49 g, and NH₄-N: 0.008 g. Moreover, for every square kilometer of soil, the respective values were 123.7 t, 61.1 t and 1.05 t (

Table 3. Maximum possible dissolution of chemical indicators in litter/soil per square kilometer.

Natural Source		COD	CODMn	NH4-N
Litter layer	Soaking solution (mg/L)	736.5	275	6.31
	Dilution multiple	4	4	4
	The maximum possible loss per kilogram litter (g)	2.95	1.10	0.025
	Per square kilometer (t)	6.39	2.39	0.055
Soil humus layer	Soaking solution (mg/L)	247.5	123.2	2.1
	Dilution multiple	4	4	4
	The maximum possible loss per kilogram soil (g)	0.99	0.49	0.008
	Per square kilometer (t)	123.70	61.10	1.050

). These estimation results indicate that the soil humus in the Kamalan River Basin serves as the primary reserve medium for background pollutants, and is approximately 20 times greater than that of litter.

3.2. Leaching Experiment

The rain intensity and rainfall in the leaching experiment were recorded and are presented in

Table 4. Estimation results for background pollutant concentration loss coefficient and river inflow coefficient.

Experimental Process	Sample Type	Monitoring Data	COD	CODMn	NH4-N	Rain Intensity (mm/h)	Rainfall (mm)
Litter layer	Soaking solution (mg/L)	/	736.5	275	6.31	\	\
	Leaching solution (mg/L)	24/06/2020	138	52.1	0.66	30.6	5.1
		30/07/2020	291	148	3.96	28.1	24.6
		01/08/2020	139	80.3	2.33	7.4	11.1
Leaching coefficient in litter			0.26	0.34	0.38	\	\
Soil humus layer	Soaking solution (mg/L)	/	247.5	123.2	2.1	\	\
	Leaching solution (mg/L)	24/06/2020	173	72.8	1.09	30.6	5.1
		30/07/2020	224	124.8	2.23	28.1	24.6
		01/08/2020	121	52.7	1.21	7.4	11.1
Leaching coefficient in soil			0.70	0.68	0.72	\	\
Sub-watershed outlet	No. 2 Bridge sub-watershed (mg/L)	/	35.1	13.6	0.21	\	\
Pollutant concentration inflow coefficient			0.2	0.16	0.14	\	\

. The data show that as the rainfall increases, the concentration of background pollutants in the leaching solution also increases. For instance, when the rainfall was the highest at 24.1 mm, the corresponding concentrations of COD, COD_Mn, and NH₄-N in the litter leaching solution were 291 mg/L, 148 mg/L, and 3.96 mg/L, respectively. In comparison, the concentrations in the soil humic leaching solution were 224 mg/L, 124.8 mg/L, and 2.23 mg/L, respectively. From the leaching experiments, it can be concluded that rain intensity does not play a major role in the leaching of background pollutants. The forest canopy’s interception effect reduces the erosion caused by dripping, especially in areas with dense and extensive canopy coverage [25]. Therefore, the primary mechanism of rainwater penetrating the underlying surface is erosion rather than splash erosion.

The leaching coefficient of background pollutants in the litter and soil humic layer could be estimated based on the source strength and leaching experiments. The leaching coefficients of COD, COD_Mn, and NH₄-N were 0.25, 0.32, and 0.23, respectively, in the litter layer, and 0.7, 0.72, and 0.54, respectively, in the soil humus layer. It was observed that the leaching coefficient of the soil humus layer was higher than that of the litter layer. The study area has high forest vegetation coverage and strong soil infiltration capacity. When precipitation leaches the litter, it carries a significant amount of nutrients into the soil ecosystem. Initially, the runoff continues to leach the nutrients during the early stage of infiltration, but as the infiltration depth increases, the leaching effect transforms into an adsorption effect. In the experiment, the average concentrations of COD, COD_Mn, and NH₄-N at the outlet of the No. 2 Bridge sub-watershed were 35.1 mg/L, 13.6 mg/L, and 0.21 mg/L, respectively. These results allowed us to estimate the background pollutants’ concentration inflow coefficient as 0.2, 0.16, and 0.14, respectively.

3.3. Background Pollutant Flux in Sub-Watershed

3.3.1. Estimation Method Based on Inflow Coefficient

The runoff in the Kamalan River Basin primarily supplies the river channel through soil flow or underground runoff. During this process, the runoff first washes out the background pollutants in the litter and enters the soil layer, where its concentration increases significantly. It then moves towards the outlet of the sub-watershed, with an inflow coefficient ranging from about 0.14 to 0.2 (Table 4). Based on the source strength experiment, the concentration of the runoff varies from the source to the outlet of the sub-watershed, with an inflow coefficient ranging from about 0.1 to 0.14. The average runoff coefficient of the watershed is 0.12 (section rainfall runoff experiment). When estimating the value using c × Q, the loss coefficient is approximately 0.012 to 0.017. Considering the background pollutants loaded per unit area (Table 3), it can be estimated that the annual fluxes of COD, COD_Mn, and NH₄-N per unit area in the sub-watershed are 2.1 t/(km²·a), 0.81 t/(km²·a), and 0.013 t/(km²·a), respectively.

3.3.2. Method Verification Based on Measured Data

Four natural rainfall-runoff processes were observed in the No. 2 Bridge sub-watershed from June to August 2020. The outlet flow process and the concentration of COD are presented in Figure 4. The total rainfall amounts were 3.3 mm, 8.6 mm, 11.7 mm, and 24 mm, respectively. The calculated runoff coefficient ranged from 0.08 to 0.23, with an average of 0.12. According to ∫cqdt, the COD loads in the four processes of the No. 2 Bridge sub-watershed were 62.16 kg, 234.68 kg, 368.18 kg, and 1918.71 kg, respectively. The discharge time, peak time, and concentration curve exhibited variations under different rainfall conditions.

Table 5. Background pollutant load output of four rainfall flow processes in No. 2 Bridge sub-watershed (kg).

Process	Rainfall (mm)	COD	CODMn	NH4-N
1	3.3	62.16	15.85	0.51
2	8.6	234.68	82.19	1.72
3	11.7	368.18	137.52	3.12
4	24	1918.71	852.93	15.39
R2		0.91	0.92	0.94

presents the values of COD_Mn and NH₄-N. It is evident that the correlation coefficient between rainfall and the output of background pollutant load was greater than 0.9, indicating a strong response relationship with rainfall.

A strong correlation was observed between the background value and flow in the No. 2 Bridge sub-watershed (Figure 5). The main driving factor for the output of background pollutant load from a single non-point pollution source is rainfall runoff. There is a strong linear relationship between the concentration and flow of these pollutants. The source strength does not change significantly, indicating that the loss process of background pollutants is determined by rainfall runoff.

There was no significant impact from human activities on the underlying surface conditions in a short period of time. A positive correlation was found between runoff and non-point source pollution output [26]. The experiments conducted in the No. 2 Bridge sub-watershed showed a correlation coefficient greater than 0.9 between rainfall and background pollutant load output. By considering the rainfall data from the Bishui hydrological station in 2019 (shown in Figure 3), the estimated annual output of background pollutant load in the No. 2 Bridge sub-basin for 2019 is presented in

Table 6. Estimated river inflow flux of background pollutant load per unit area in the study area.

Estimation Method	COD	CODMn	NH4-N
Verification Method Based on Measured Data t/(km2·a)	1.84	0.69	0.01
Estimation Method Based on Inflow Coefficient t/(km2·a)	2.10	0.81	0.013
Relative error %	14.1	17.4	30.0

and Figure 6. The background pollutant outputs for COD, COD_Mn, and NH₄-N in the No. 2 Bridge sub-watershed were 18.2 t, 7.7 t, and 0.11 t, respectively.

The annual load output of background pollutants per unit area in the study area, estimated from the measured data in 2019, was higher than that from the concentration inflow coefficient. The relative errors of COD, COD_Mn, and NH₄-N were 14.1%, 17.4%, and 30.0%, respectively. This level of accuracy meets the estimation requirements for remote areas without data. The estimation method based on the inflow coefficient in small watersheds is suitable for this study area.

3.4. Estimation Method for Large Scale Watershed

The estimation of large-scale pollutant flux can often be achieved through mechanism models, but this approach requires a significant amount of measured data and parameters [26]. However, in this study, the lack of hydrological and water quality data in the natural remote mountainous study area restricted the application of mechanism models. On the other hand, statistical models are simpler in structure and usually provide a certain level of accuracy with only a few parameters. These models are commonly used for estimating non-point source pollution load in large-scale basins where data are available [27,28]. In the case of the Kamalan River Basin, the improved output coefficient model and universal soil loss equation are typically used to estimate background pollutant output.

3.4.1. Improved Output Coefficient Model

In terms of pollutant output coefficient, domestic studies often rely on existing literature to estimate the inflow of non-point source pollutants in areas without data using the output coefficient model [29]. However, the model’s applicability and accuracy in certain regions are not high due to environmental differences, which significantly impacts its feasibility. While many studies have improved the classical output coefficient model by considering factors such as rainfall and terrain, the applicability of the pollutant output coefficient is rarely discussed. In this paper, we calculated the output coefficients for COD, COD_Mn, and NH₄-N based on a rainfall-runoff experiment conducted at the No. 2 Bridge (Table 6, background pollutant fluxes per unit area). Building upon this, we estimated the fluxes of background pollutants in the Kamalan River Basin, taking into account the spatial distribution differences in rainfall and slope on the output characteristics (Figure 7).

According to the improved pollutant output coefficient model, the annual average fluxes of COD, COD_Mn and NH₄-N in the Kamalan River Basin are 2338.7 t, 877 t, and 12.7 t, respectively. These distribution characteristics are depicted in Figure 8.

3.4.2. Improved Universal Soil Loss Equation

The USLE model was utilized to estimate the inflow flux of background pollutants in the Kamalan River Basin. Initially, the Kamalan River Basin was divided into 33 sub-basins according to their catchment characteristics. The model was then enhanced based on the specific features of the study area.

$$F=B_i\sum _{i=1}^{33}{\lambda }_{i}c(R_i·K·{LS}_i·C·F)$$

$$F=B_i\sum _{i=1}^{33}{\lambda }_{i}cK(R_i·{LS}_i·C·F)$$

The annual flux of pollutants into the river (F) was measured in t/(km²·a), while the background pollutant average concentration (c) was measured in g/kg. The area of each sub-watershed is denoted as Bi, and the pollutant transport ratio of sub-basin i is represented by λi. This ratio is influenced by factors such as the size of the sub-watershed (Bi), slope, and rainfall. However, the variation in Bi is much greater than that of slope and rainfall spatial distribution. Therefore, this study focused on the relationship between λi and Bi. Wu [30] proposed a research formula for sediment transport ratio and watershed area to determine the pollutant transport ratio of each sub-watershed. In this formula, Bi represents the area of each sub-watershed, and μ determines the correction coefficient for the equation of the sediment transport ratio.

$$F=\sum _{i=1}^{33}\mu \frac{0.1406B_i^2-0.1253B_i+0.4065}{0.3995B_i^2+0.8735B_i+0.7176}cK(R_i·{LS}_i·C·F)$$

Rainfall-runoff factor (R): The rainfall reanalysis data from 2008 to 2019 (National Centers for Environmental Prediction, CFSR) was used to replace the input rainfall data. The spatial difference of rainfall was utilized to extract the multi-year monthly and annual average rainfall of each sub-basin, which was then centered within the sub-watershed. The R value for each sub-watershed was calculated using the rainfall runoff factor formula. Slope length and slope factor (LS): Topographic indicators such as slope length and slope were obtained through the extraction of 30 × 30 m elevation data from the Kamalan River. Vegetation and management factor (C): The value of the biological measure factor B, as specified in China’s universal soil loss equation, was mainly considered. The forest land in the study area had a vegetation coverage of 75.2%, resulting in a C value of 0.02. Water and soil conservation measure factor (F): Since no water and soil conservation measures have been applied in the Kamalan River Basin, the value of F was set to 1. Organic pollutant erodibility factor of t/(km²·a): The predominant soil type in the Kamalan River Basin is brown coniferous forest soil. The soil erodibility factor for the 33 sub-watersheds was assumed to be constant in this model. Background material concentration (c, g/kg) in unit mass organic pollutants: It was assumed that the background material concentration (mass) carried by unit mass organic pollutants remained constant in this model. Equation correction coefficient (μ): The pollutant transport ratio of each sub-watershed was determined by the relationship between the sediment transport ratio and watershed area. Therefore, the transfer of this formula was considered as a correction coefficient μ to eliminate migration deviation. μ is a constant. In summary, the model can be generalized into the following forms:

$$F=0.02\sum _{i=1}^{33}\phi \frac{0.1406B_i^2-0.1253B_i+0.4065}{0.3995B_i^2+0.8735B_i+0.7176}(R_i·{LS}_i)$$

$$\phi =\mu cK$$

Therefore, φ was the key factor used to estimate the large-scale fluxes into the river. In the section titled “Method Verification Based on Measured Data”, the fluxes of background pollutants into the river in the No. 2 Bridge sub-watershed in 2019 were measured as COD: 18.2 t, COD_Mn: 7.7 t, and NH₄-N: 0.11 t, respectively. The generalized model was then applied to the No. 2 Bridge sub-watershed in 2019 to calculate φ. R was calculated based on the rainfall data from the Bishui hydrological station in 2019, and the calculated φ values for COD, COD_Mn, and NH₄-N were 34.3 t/(km²·a), 13.1 t/(km²·a), and 0.19 t/(km²·a), respectively. Subsequently, the estimation equation for background pollutant flux into the river in the Kamalan River Basin was as follows. The annual average fluxes of COD, COD_Mn, and NH₄-N were 2072.1 t, 789.9 t, and 11.5 t, respectively, which are slightly lower than those estimated with the improved output coefficient model. The relative errors were 12.9%, 11%, and 10.4%, respectively.

$$F_{COD}=0.686\sum _{i=1}^{33}\frac{0.1406B_i^2-0.1253B_i+0.4065}{0.3995B_i^2+0.8735B_i+0.7176}(R_i·{LS}_i)$$

$$F_{CODMn}=0.262\sum _{i=1}^{33}\frac{0.1406B_i^2-0.1253B_i+0.4065}{0.3995B_i^2+0.8735B_i+0.7176}(R_i·{LS}_i)$$

$$F_{\mathrm{N}\mathrm{H}4-\mathrm{N}}=0.0038\sum _{i=1}^{33}\frac{0.1406B_i^2-0.1253B_i+0.4065}{0.3995B_i^2+0.8735B_i+0.7176}(R_i·{LS}_i)$$

3.5. Rationality and Applicability Analysis

In this experiment, we monitored the water quality and quantity of the outlet section in the Kamalan River Basin. The monitoring periods were from 9 May to 28 May 2019, 16 June to 1 July 2019, and 19 July to 9 August 2019. For the dates when monitoring was not conducted, we interpolated the data. From 9 May to 9 August 2019, we observed the background pollutant output process at the control section of the Kamalan River, as shown in Figure 9. The total output of COD, COD_Mn, and NH₄-N was 1412.7 t, 421.5 t, and 9.8 t, respectively.

Despite the lack of synchronous monitoring data for the Kamalan River in other time periods of 2019, a clear correlation was shown between the runoff and the inflow of non-point source pollution into the river in the Kamalan River Basin. Therefore, based on the annual distribution characteristics of rainfall and the inflow of pollutants from 9 May to 9 August 2019, it was possible to estimate the amount of background pollutants in other times of the year. Figure 10 illustrates the annual distribution of multi-year average rainfall in the Kamalan River Basin, where the rainfall from 9 May to 9 August accounts for 49.9% of the total annual rainfall. Assuming a unique runoff coefficient for the entire Kamalan River Basin, and considering the relationship between runoff and non-point source pollution load, the estimated amounts of COD, COD_Mn, and NH₄-N were 2825.5 t, 842.9 t, and 19.6 t, respectively.

The estimated COD and NH₄-N inflow in the Kamalan River Basin were significantly higher than those obtained using method (1) and method (2), while the COD_Mn values were similar to those reported in 2019 (

Table 7. Results of estimates of annual fluxes of background pollutants into the Kamalan River Basin by three methods (t/a).

Estimation Method	COD	CODMn	NH4-N
(1) Improved Output Coefficient Model	2338.7	877.0	12.7
(2) Improved Universal Soil Loss Equation	2072.1	789.9	11.5
(3) Estimated Inflow in 2021 Based on Measured Data	2825.5	842.9	19.6

). The relative errors for COD, COD_Mn, and NH₄-N inflows using method 1 were 17.2%, 4%, and 35%, respectively, while for method 2, they were 26.6%, 6.3%, and 41.3%. The estimated results closely matched the calculated values. Given the limited availability of data for the remote mountainous forest land that makes up the background area, it is feasible to estimate large-scale background pollutant fluxes into the river using the improved output coefficient model and the improved universal soil loss equation.

4. Conclusions

The impact of environmental background pollutant values on environmental management and evaluation in specific areas is increasingly significant. To clarify the characteristics of pollutant loss, migration, and transfer in background areas, and to understand the output characteristics, a series of results were obtained based on previous research. This included field sampling, indoor analysis, and in situ monitoring experiments. For the No. 2 Bridge sub-watershed, the characteristics of pollutant load were understood through source intensity experiments. The pollutant concentration inflow coefficient was determined to range from 0.1 to 0.14. Additionally, the pollutant load inflow coefficient was estimated to be between 0.012 and 0.017, based on statistical analysis of hydrological, water quality, and rainfall data. The annual inflow flux was estimated, and its rationality was verified.

In the Kamalan River Basin, the background pollutants’ annual flux into the river was estimated by improving the output coefficient model and universal soil loss equation, combined with hydrological and meteorological data. The rationality of the estimation was verified based on the measured data at the basin’s outlet.

The study applied statistical models to address the issue of the mechanism model’s limited applicability in areas with insufficient data. While the accuracy of these models may be lower than that of the mechanism model, they offer a solution for studying the loss, migration, and transformation of pollution logistics in the background area of the source river. Additionally, they provide a method for estimating pollutant load output flux in data-scarce areas. Although numerous indoor and field in situ experiments have been conducted, our research has certain limitations. Some experiments cannot be conducted continuously due to constraints such as the location of the study area in a reserve. In the future, our team plans to conduct further environmental background research, considering multiple variables, in different regions of northeastern China.