Next Article in Journal
Hot-Point Ice Thermal Drills: Design Parameters, Recommendations, and Examples
Previous Article in Journal
Comparative Study of RANS Models for Simulating Turbulent Flow and Heat Transfer in Corrugated Pipes
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Water
Volume 17
Issue 17
10.3390/w17172651
water-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Spatial Targeting and Budget-Adaptive Optimization of Best Management Practices for Cost-Effective Nitrogen Reduction

by
Yunkai Fan
1,
Huazhi Zhang
2,
Bing Yu
1,3,*,
Ming Cong
1 and
Zhuohang Xin
1,3
1
School of Infrastructure Engineering, Dalian University of Technology, Dalian 116024, China
2
Changjiang Schinta Software Technology Co., Ltd., Wuhan 430010, China
3
Major Infrastructure Construction Technology Innovation Center, Ningbo Institute of Dalian University of Technology, Ningbo 315000, China
*
Author to whom correspondence should be addressed.
Water 2025, 17(17), 2651; https://doi.org/10.3390/w17172651
Submission received: 28 July 2025 / Revised: 26 August 2025 / Accepted: 4 September 2025 / Published: 8 September 2025
(This article belongs to the Section Water Quality and Contamination)
Browse Figures
Versions Notes

Abstract

This study developed a Soil and Water Assessment Tool (SWAT) model for the Fuzhou River Basin in China to quantify the spatial distribution, sources, and reduction potential of total nitrogen (TN) load. We comprehensively evaluated the effectiveness of eight Best Management Practices (BMPs) and 186 combinations thereof in reducing TN load. Our analysis demonstrated that adding more BMPs did not yield proportionally additive benefits but instead led to reduced cost-effectiveness (CE) once the number of BMPs exceeded three. Targeting BMPs to Critical Source Areas (CSAs) increased CE by an average of 15.6% compared to watershed-wide application, although the environmental benefit (EB) was lower (22.0% versus 32.8% on average). We identified a critical budget threshold of 70 million CNY. Below this threshold, CSA-targeting optimized BMPs delivered the most cost-effective TN reductions (123.0 kg/104 CNY per year). However, with a sufficient budget exceeding this threshold, our findings support implementing BMPs throughout the entire watershed, which maximized the TN reduction rate to over 40%. Overall, our findings highlight that spatial targeting and budget-adaptive implementation of BMPs are essential for maximizing both economic efficiency and environmental benefits, providing a practical decision approach for nutrient management in river basins.

1. Introduction

In recent years, urbanization and population growth have intensified industrial and agricultural activities within watersheds, leading to substantial nitrogen releases into aquatic ecosystems [1,2]. Elevated levels of total nitrogen (TN) in water can cause a range of ecological problems, such as harmful algal blooms and reduced dissolved oxygen levels [3,4,5]. To mitigate the environmental impacts of nitrogen loading, it is essential to understand the nitrogen sources and implement appropriate management interventions.
Mathematical modeling is an efficient and accurate method for understanding nitrogen transport processes [6,7]. Among various models, the Soil and Water Assessment Tool model (SWAT) is particularly effective in simulating runoff and pollutant transport across diverse spatial and temporal scales and has been widely applied worldwide [8]. Based on the quantification of pollution sources and impacting mechanisms, Best Management Practices (BMPs) can be implemented to assess the potential effectiveness in reducing TN concentrations and loads in water bodies. BMPs constitute a comprehensive system for protecting water resources from pollution, which has been proven effective in controlling the transport of TN into receiving water bodies [9,10,11]. Given the diversity of BMPs, as they differ in placement costs and nitrogen removal efficiency [12,13,14], selecting the most suitable BMPs requires appropriate evaluating models. Xie et al. conducted a comparative analysis of various hydrological models for BMP evaluation and found that the SWAT model is nearly universally applicable, demonstrating its comprehensiveness and accuracy [15]. In recent years, the use of the SWAT model to evaluate and select BMPs has gained popularity [16]. For instance, Haas et al. used the SWAT model to simulate TN load in the Treene watershed and determined that a 30% reduction in fertilizer application was the most effective BMP [17]. Leta et al. employed SWAT to identify the most effective BMPs for reducing annual sediment yield in the Nashe watershed, finding that stone/soil bunds and terracing achieved reduction rates of 57.98% and 54.77%, respectively [18].
The distribution of pollutant loads may vary considerably across different areas within a watershed, with a small proportion of runoff often carrying the majority of the pollutants [19]. For example, Winchell et al. simulated phosphorus distribution in the Champlain watershed of the Mississippi River and found that a sub-watershed covering only 20% of the area contributed 74% of the total phosphorus (TP) load [20]. Similarly, Niraula et al. found that a sub-watershed within the Saugahatchee Creek watershed covered only 6.5% of the total area, while it contributed 26.5% of the sediment load, 23.1% of the total phosphorus (TP) load, and 13.9% of the total nitrogen (TN) load [21]. Because of the high impacts of these regions, studies have recommended concentrating management efforts on Critical Source Areas (CSAs) within watersheds to optimize resource allocation and save costs in reducing pollutant loads [22,23,24]. Liu et al. evaluated BMPs in the CSAs of the Zhuxi watershed in China and found that a 5-m vegetative buffer strip was the most effective for nutrient removal [25]. Similarly, Babaei et al. identified the optimal BMPs for nitrogen and phosphorus removal in the CSAs of the Dez River in Iran, concluding that a 10-m buffer strip was the best choice [26]. Importantly, targeting interventions to CSAs is not only environmentally effective but also highly cost-effective, as it concentrates limited resources on the areas that yield the highest pollutant reduction per unit of investment [27]. Mazdak Arabi et al. demonstrated that optimizing the selection and placement of BMPs in two small watersheds in the United States improved cost-effectiveness (CE) by nearly threefold [28].
Building upon the significant research progress in identifying CSAs and evaluating the effectiveness of individual BMPs, there still remains a lack of systematic comparative research on evaluating CE of numerous BMP combinations across different spatial scales [29,30,31]. Furthermore, in practical decision-making contexts, the feasibility of BMPs is heavily influenced by available budget [32]. Therefore, it is imperative to focus on the trade-offs between CE and environmental benefit (EB) under varying budget constraints to support the selection of optimal solutions under different budgetary conditions.
Therefore, to bridge these knowledge gaps and address local environmental priorities, this study takes the Fuzhou River Basin (FRB) in northeastern China as a case study, with a specific focus on TN load reduction. The study region suffers from severe TN pollution issues, where intensive agricultural and industrial activities have resulted in persistent exceedances of regulatory water quality standards [33,34]. The main objectives are (1) to simulate streamflow and TN transport processes in the FRB using a calibrated and validated SWAT model; (2) to analyze the spatiotemporal variations of TN loads and identify CSAs within the basin; (3) to quantitatively evaluate the TN reduction rates and CE of a wide range of BMP combinations (single and multiple BMPs) implemented across the entire watershed and within the identified CSAs; and ultimately (4) to provide optimal BMP combinations tailored to different budget constraints, explicitly considering the trade-offs between CE and EB.

2. Materials and Methods

2.1. Study Area

The Fuzhou River, located in Dalian City, China (Figure 1a), stretches 129.4 km and flows into the Bohai Sea. The FRB covers an area of 1648 km2, with elevations ranging from 3 m to 510 m. The average annual temperature in the watershed ranges from 8.6 °C to 10.5 °C, and the average annual precipitation is 654 mm, with 75% of the rainfall occurring between July and September.
The FRB features multiple monitoring sites (Figure 1b), including the Wafangdian meteorological station (WFD), Guanjiatun hydrological station (GJT), and Santaizi water quality station (STZ). Data from these monitoring sites were used in the SWAT model. Monitoring data from the STZ showed that the TN concentration in FRB significantly surpassed the regulatory limit. This elevated TN concentration can be primarily attributed to the presence of multiple wastewater treatment plants and direct industrial emissions within FRB, along with non-point source pollution from rural domestic sewage, livestock farming, and agricultural planting.
To understand the nitrogen transport processes and identify effective BMPs for TN reduction, the following steps were undertaken, as shown in Figure 2: (1) development of a SWAT model to simulate runoff and TN dynamics; (2) analysis of pollution source contribution rates using the model; (3) identification of CSAs within the sub-watersheds; and (4) evaluation of BMP combinations applied sequentially across the watershed and CSAs, with optimization for different budget constraints.

2.2. Water Quality Model

The Soil and Water Assessment Tool (SWAT) model, a physically based semi-distributed hydrological model developed by the United States Department of Agriculture [35], was employed to simulate runoff and nitrogen transport processes in this study. The SWAT model operates on a daily time step and divides watersheds into interconnected sub-watersheds based on stream networks. These sub-watersheds are further discretized into Hydrological Response Units (HRUs)—homogeneous areas defined by unique combinations of land use, soil type, and slope. This structure enables robust simulation of land management impacts on water quality across complex landscapes [36].
In the FRB, sub-watersheds were delineated using a 30 m DEM and stream network data. The data were from the Geospatial Data Cloud (https://www.gscloud.cn/ (accessed on 14 March 2022)) and the Ecological and Environmental Bureau of Wafangdian City, and all data have undergone standard hydrological processing procedures [37]. A sub-watershed threshold of 2000 hm2 was applied, resulting in 30 sub-watersheds (Figure S1). HRUs were defined based on six land use classes, four soil types, and slope categories. Land use data (30-m resolution) were sourced from the Resource and Environmental Science Data Center of the Chinese Academy of Sciences (https://www.resdc.cn/ (accessed on 14 March 2022)), while soil data (1-km resolution) were obtained from the World Soil Database (HWSD) China Soil Dataset. Although the 1-km soil data resolution is coarser than the 30 m DEM and LULC data, it represents the best available consistent dataset for the FRB. The uncertainties were mitigated through majority resampling to 30 m grids aligned with dominant land use during HRU creation [38]. The model ultimately identified 496 HRUs, as detailed in Table S1. Meteorological variables, including daily maximum and minimum air temperatures, precipitation, wind speed, and relative humidity, from 2008 to 2020, were obtained from meteorological stations (Figure 1b).
Through field investigations, the nitrogen sources in the FRB have been identified as comprising five main contributors. Point sources include wastewater treatment plant effluents and industrial discharges, while non-point sources consist of rural domestic sewage, livestock farming, and agricultural planting. Discharge volumes and pollutant concentrations from wastewater treatment plants and industries from 2017 to 2020 were obtained from the official monitoring records of the Wafangdian City Ecological Environment Sub-bureau. The specific discharge locations and volumes were detailed in Figure S2 and Table S2. Non-point source pollution from rural domestic sewage was estimated at the township level, using population data and per capita TN generation rates from the Dalian Statistical Yearbook and the Handbook of Pollutant Emission Coefficients for Non-point Sources. These data were aggregated at the township level, as listed in Table S3. Pollution from livestock farming was also calculated at the township level, with livestock numbers including meat chickens, pigs, dairy cattle, and beef cattle. The data were obtained from the Dalian Statistical Yearbook, and pollutant emission coefficients were from the Handbook of Pollutant Emission Coefficients for Agricultural Pollution Sources. Livestock-related pollution at each township was detailed in Table S4. Non-point source pollution from agricultural farming was assessed using data on cultivated areas and applied nitrogen fertilizer amounts from the Dalian Statistical Yearbook. The specific agricultural pollution data are provided in Table S5.
In this study, the SUFI2 algorithm within SWAT-CUP was used to rank the sensitivity of eleven runoff parameters and twelve TN parameters. The sensitivity ranking results and the values of each parameter are shown in Table S6. Then, calibration and validation of both runoff and TN were conducted. The runoff was observed at the GJT hydrological station (Figure 1b), which is situated on the mainstem and is the only gauging station within the FRB. The runoff data were divided into a warm-up period (2008–2009), a calibration period (2014–2017), and a validation period (2018–2020). The observed TN concentrations were collected from the STZ monitoring site (Figure 1b), which is situated at the downstream outlet of the basin and can effectively capture the TN export from the whole basin. These data spanned from 2017 to 2020, with the calibration period covering 2017–2019 and the validation period covering 2019–2020. Model performance was evaluated using the coefficient of determination (R2) and the Nash–Sutcliffe Efficiency (ENS), where values closer to 1 indicated better model performance.

2.3. Identifying Critical Source Area of TN

In practice, due to constraints in land space and funding, decision-makers often prioritize management measures in regions with higher pollution contribution rates, aiming to achieve significant pollution reductions with minimal investment. For non-point source pollution, we applied the cumulative pollution load curve method to classify sub-watersheds based on TN load contributions, as quantified by the SWAT model (Section 3.1) [39,40]. Sub-watersheds were categorized into five risk levels based on cumulative load thresholds: extremely high-risk (80–100%), high-risk (60–80%), moderate-risk (40–60%), low-risk (20–40%), and extremely low-risk (<20%). The first three risk levels (extremely high, high, and moderate) were identified as non-point CSAs in this study. The classification criteria were based on the work of Wei et al. [39,40]. Identifying point CSAs was more straightforward, focusing on the discharge volumes from wastewater treatment plants and industries within their respective sub-watersheds.

2.4. The Simulation and Evaluation of BMPs

This study selected eight BMPs, comprising both engineering and non-engineering measures, for the mitigation and control of pollutants from point and non-point sources. The BMPs include Contour Tillage (CT), Strip Cropping (SC), Nitrogen Fertilizer Reduction of 10% (FR10), 25% (FR25), 50% (FR50), Point-source Reduction (PR), Terracing (TR), and Filter Strips (FS). These BMPs were selected because they have been proven to achieve effective pollution reduction at feasible costs and have been extensively studied and adopted, particularly in watersheds where non-point source pollution is prominent [23,41,42,43,44]. BMP parameters were established based on the SWAT model’s input–output manual [45], the relevant literature [46,47,48,49,50,51], and local conditions (Table 1). This study incorporated the research of Du et al. by using constructed wetlands in sub-watersheds with point-source inputs to reduce point-source pollutants [52]. Previous studies on constructed wetlands in the FRB informed the strategy, with a target of 30% TN reduction for point-source pollution [53,54]. The values of the USLE_P parameter, which accounts for soil and water conservation in ST, TC, and TR, were listed in Table S7, while slope-dependent TR width values were provided in Table S8. The baseline scenario (BAS) was defined as the calibrated SWAT model without any BMP implementation.
The effectiveness of BMPs was evaluated based on environmental benefits (EBs) and cost-effectiveness (CE) [55]. The SWAT model was run with adjusted parameters for each BMP scheme (either individual or combined), and the TN load reduction at the outlet was calculated by comparing each scenario to the baseline (Equation (1)). Cost-effectiveness was expressed as TN load reduction per unit cost (Equation (2)). BMP implementation costs were derived from field surveys, market research, and the relevant literature [46,47,49,51,56,57,58,59], as detailed in Table 1. To evaluate the impact of cost uncertainties on cost-effectiveness results, we performed a Monte Carlo sensitivity analysis with 1000 iterations. Unit costs for each BMP were independently randomized within ±20% of baseline values using a normal distribution, and CE values for all combinations were recalculated per iteration. The analysis confirmed the robustness of our results, showing that mean CE values deviated by <5% from baseline, while optimal BMP rankings and spatial strategies remained largely unchanged (Table S9).
R ( % ) = P B A S − P B M P P B A S × 100 %
R —TN pollution load reduction rate, %; P B A S —The TN load under baseline scenario, kg/year; P B M P —The TN load under BMPs practices, kg/year.
C E = p B A S − p B M P Cos t
CE—cost-effectiveness value, kg/104 CNY·year; Cost: implementation cost of BMPs, 104 CNY; P B A S —TN load under baseline scenario, kg/year; P B M P —TN load under BMPs practices, kg/year.

3. Results and Discussion

3.1. SWAT Calibration and Validation

The streamflow simulations (Figure 3a) showed good accuracy, with an R2 of 0.83 and an ENS of 0.94 during the calibration period and an R2 of 0.81 and ENS of 0.89 during the validation period, indicating that the model reliably captured the key hydrological processes and temporal dynamics within the FRB. For TN load simulation (Figure 3b), the model also performed satisfactorily, with an R2 and ENS of 0.76 and 0.62, respectively, during the calibration period, and 0.74 and 0.61 during the validation period. The ENS values exceeded the acceptable threshold of 0.5, confirming the model’s validity [60], despite some limitations in precisely simulating peak events, e.g., during the summer of 2020. Such discrepancies have also been reported by previous studies, probably arising from model structural assumptions and the complexity of nitrogen processes [61]. While the primary objective of employing SWAT in this study is to evaluate the relative effectiveness of BMPs, the biases in the model simulation are likely to have a minimal impact on this comparative analysis.

3.2. Spatial and Temporal Analysis of Source Contribution

Figure 4 shows the annual average TN load entering the sea from five pollution sources at STZ between 2017 and 2020. The total annual TN load was 1370.34 tons, with non-point source pollution accounting for over 72% of this total. Among non-point sources, agricultural planting made the largest contribution, consistent with findings from other river basins in China [62,63]. Point sources accounted for 27.9% of the TN load, with wastewater treatment plants being the largest contributor, significantly outweighing direct industrial emissions. The contribution rates of various pollution sources to TN at STZ were ranked as follows: agricultural planting (46.7%) > wastewater treatment plants (26.7%) > livestock farming (22.2%) > rural domestic sewage (3.2%) > direct industrial emission (1.2%).
The monthly average contributions of various pollution sources from 2017 to 2020 are shown in Figure 4. The agricultural planting contribution was particularly high in May and between August and October, mainly due to crop planting and fertilization schedules. In the FRB, apples and maize were the main crops, with maize seeding concentrated in May and secondary fertilization of the orchard in late July/early August. Furthermore, a larger portion of the annual rainfall occurred in July and August, which increased surface runoff and transported large quantities of agricultural fertilizers into the river. During subsequent dry months, the legacy nutrients in the soil from earlier fertilization could continue to contribute to the pollutant load in river water [64]. The contributions from livestock farming and wastewater treatment plants were more significant in August and September. Livestock activities were more active during the summer months, driven by the warmer weather, which encourages higher feeding rates and greater waste output [65]. Similarly, the higher per capita water usage in urban areas during this period contributed to the increased nutrient load from wastewater treatment plants [66]. Rural domestic sewage contributions were higher in August, which was consistent with the seasonal pattern of increased human activities during summer. The rise in population density and water usage in rural areas during warmer months contributed to greater sewage volumes. Direct industrial emissions, however, showed minimal monthly variation, indicating a relatively consistent contribution throughout the year.
Figure 5 shows the proportional contributions of TN sources in each sub-watershed. Agricultural planting was the dominant contributor, accounting for 12.2% to 69.0% of the TN load. Specifically, sub-watersheds Nos. 1, 2, 5, 6, 29, and 30 had agricultural planting as the primary source, contributing over 60%. This spatial heterogeneity was primarily attributed to land use composition and soil characteristics. Sub-watersheds 16, 20, 22, 24, 28, 29, and 30 contain extensive croplands (>60% land area) where fertilizer-intensive cultivation directly drives high TN exports. In contrast, sub-watersheds 1, 2, 3, and 5 were dominated by woodlands (>70% coverage) with minimal point source impacts; consequently, agriculture and livestock accounted for a relatively high proportion of pollution contributions. Sub-watersheds 6 and 18 are characterized by intensive orchard coverage (45–60% of the area) and disturbed soils of backfilled loams, resulting in highly permeable substrates with low nutrient retention capacity. This condition accelerated fertilizer leaching during rainfall, a process consistent with the findings of Feng et al. [67]. Livestock farming contributed between 5.7% and 35.3%, with sub-watersheds Nos. 3, 6, 16, 22, 28, 29, and 30 showing livestock contributions over 30%. Rural domestic sewage and direct industrial plants contributed relatively minor, with a maximum proportion of no more than 6.2% and 2.7% in any single sub-watershed, respectively. Notably, a large wastewater treatment plant in sub-watershed No. 12 contributed 80% of its TN load. This significantly impacted downstream sub-watersheds (e.g., Nos. 4, 7, and 11), where pollution from the wastewater treatment plant contributed over 40% of their TN load.

3.3. Critical Source Areas Identification

Figure 6 illustrates the classification of non-point source pollution risk levels across sub-watersheds. The non-point TN load per unit area ranged from 4.28 to 18.43 kg/ha. Sub-watersheds Nos. 4, 6, 9, 15, 16, 17, and 18 were categorized as high-risk, with unit TN loads ranging from 14.23 to 18.43 kg/ha, covering 14.17% of the total basin area. Sub-watersheds Nos. 8, 10, 21, 23, and 29 were classified as medium-high risk, with TN loads ranging from 13.45 to 13.97 kg/ha, covering 15.14% of the basin. Sub-watersheds Nos. 3, 19, 24, and 26 fell into the medium-risk category, with unit TN loads ranging from 11.02 to 13.10 kg/ha, covering 20.30% of the basin area. Sub-watersheds Nos. 2, 12, and 30 were classified as medium-low risk, with unit TN loads ranging from 10.44 to 10.88 kg/ha, covering 22.21% of the area. Finally, sub-watersheds Nos. 1, 5, 7, 11, 13, 14, 20, 22, 25, 27, and 28 were categorized as low risk, with TN loads ranging from 4.28 to 10.38 kg/ha, covering 28.18% of the total area. Sub-watersheds classified as medium, medium-high, and high risk were identified as non-point CSAs, covering 49.6% of the total basin area and contributing 59.9% of the non-point TN load. The proportion of pollution contributed by CSAs is similar to the study of Chen et al. [68]. For point CSAs, the Longshan Wastewater Treatment Plant in sub-watershed No. 12 accounted for over 90% of the point-source TN (Table S2, Figure S2), making sub-watershed No. 12 the designated point CSA. The final CSAs classification results are shown in Figure 7. It is important to note that only PR (Table 1) was implemented in point CSA, while other BMPs were targeted at non-point CSAs.

3.4. Environmental Benefits of Implementing BMPs

Table 2 presents the simulated TN load reductions following the implementation of BMPs across the entire watershed and within CSAs. When applied to the whole watershed, TN load reduction rates ranged from 4.9% to 24.3%. The BMPs ranked by reduction rate were FR50 (24.3%) > TR (15.5%) > FS (12.4%) > SC (12.2%) > FR25 (12.0%) > PR (7.8%) > CT (7.6%) > FR10 (4.9%). Among these, FR50 achieved the highest reduction, primarily due to agricultural planting being the main contributor to TN loads. This finding aligned well with Jamshidi et al., which showed that reducing emissions at the source was more effective than subsequent remediation due to nitrogen transport mechanisms [69]. TR and FS were the most effective engineering BMPs, consistent with findings in the Lam Takong Basin in Thailand [70]. When BMPs were concentratedly applied in CSAs, TN load reduction rates ranged from 3.1% to 15.2%. The BMPs ranked by reduction rate were FR50 (15.2%) > TR (8.3%) > FR25 (7.4%) > FS (6.6%) > SC (6.5%) > PR (5.9%) > CT (4.0%) > FR10 (3.1%). FR50 and TR were still the two most effective measures among eight BMPs. Notably, FR25 was prioritized ahead in the ranking compared to the watershed scale, indicating that reducing nitrogen fertilizer application became more effective in CSAs.
Eight individual BMPs were combined into dual to sextuple combinations, with the constraint that only one of the three FR measures (FR10, FR25, or FR50) could be included in each combination, resulting in a total of 93 unique combinations. Box plots illustrating the TN reduction rates from dual to sextuple are presented in Figure 8a for both the watershed and CSA scales. At the watershed scale, the average TN reduction rates for dual to sextuple combinations were 24.6%, 30.3%, 33.4%, 35.4%, and 36.8%, respectively. At the CSA scale, the averages were 14.9%, 18.6%, 20.7%, 22.3%, and 23.6%. In both the watershed and CSA scales, BMP combinations significantly outperformed individual BMPs (Table 2), with TN reduction rates increasing as more BMPs were implemented. This trend aligns with findings of Uniyal et al. in the Baitarani river basin, India [71]. However, the rate of TN reduction exhibited diminishing returns as the number of BMPs increased, suggesting that the effects of BMP combinations were not merely additive. Therefore, simply increasing the number of BMPs did not consistently lead to proportional improvements in TN reduction. Similar conclusions were supported by Liu et al. in the Poyang Lake watershed in China [25]. Comparing the EBs between the watershed and CSA scales, the average TN reduction rates at the watershed scale were 39.4% to 35.9% higher than at the CSA scale for dual to sextuple combinations, despite the watershed area being approximately twice the size of the CSAs. This suggested that concentrating BMPs in CSAs can enhance pollutant reduction effectiveness, which aligns with the conclusions drawn by Shen et al. [72].

3.5. Cost-Effectiveness of Implementing BMPs

Environmental policymakers must consider the costs associated with implementing BMP combinations, with a particular focus on maximizing pollutant reduction per unit cost. Therefore, in addition to the TN reduction rate, CE is also a key criterion for evaluating BMP combinations. Table 3 presents the CE values for each individual BMP across the watershed and within the CSAs. At the watershed scale, the CE ranking for the eight BMPs was as follows: FS > PR > FR10 > FR50 > FR25 > TR > SC > CT. At the CSA scale, the ranking was FS > PR > SC > FR10 > FR50 > FR25 > TR > CT. FS exhibited the highest CE value at both scales. As a rectangular strip placed downstream of pollution sources, FS relies on soil’s ability to filter, dilute, infiltrate, and absorb pollutants, as well as microbial transformations, making it highly effective in treating contaminants with relatively small land area requirements [73]. Research showed that a 3-m-wide FS can significantly reduce surface runoff sediment, while 5–8 m was sufficient to filter some nitrate nitrogen [74]. Although the unit area installation cost for FS was relatively high (1150 CNY/hm2), its small required area resulted in a lower total cost. FS also achieved a relatively high TN reduction rate (12.4% at watershed and 6.6% at CSAs, Section 3.3), contributing to its high CE. PR ranked second in CE, largely due to its low implementation cost (0.23 CNY/t point source sewage). When comparing the performance between the watershed and CSA scales, the CE values of PR, CT, and TR exhibited notably higher values at the watershed scale, suggesting that these BMPs were more suited for widespread deployment across the entire watershed. In contrast, SC, FR, and FS were more appropriate for targeted implementation within CSAs.
Figure 8b illustrates box plots of CEs for different BMP combinations at both watershed and CSAs. At watershed, the average CE values for dual to sextuple combinations were 60.5, 49.6, 39.6, 32.8, and 28.1 kg/104 CNY·year, while at CSAs, the values were 66.5, 55.9, 46.2, 39.2, and 34.3 kg/104 CNY·year, respectively. Regardless of the implementation scale, increasing the number of BMPs resulted in a decrease in CE values. This decrease can be attributed to the cumulative costs of implementing multiple BMPs, while the combined TN reduction rates of the combinations were less than the sum of the individual BMPs. This aligned with the findings of Liu et al. in an agricultural watershed in northeastern Indiana [75]. A comparative analysis revealed that the CE values in CSAs outperformed those of the entire watershed, indicating that targeted interventions with a smaller coverage area were more cost-effective than broader watershed-scale implementations. Similar results have been reported in the Arachtos River Basin in western Greece [76].
The observed trade-off—where CSA implementations achieve higher CE but lower EB compared to watershed-scale implementations—may be attributed to several reasons. First, although CSAs concentrate high-yield pollution sources, enabling cost-effective reduction of the major source, they inherently miss diffuse background loads from non-CSA areas. This is particularly critical for soluble nitrogen transported via groundwater or distant runoff, which accumulates at the watershed outlet [77,78]. Second, CSA-focused BMPs primarily intercept near-field pollutants. In contrast, watershed-wide deployment reduces nitrogen all along the flow path—from headwater legacy nutrient leaching to mid-reach inter-sub-watershed fluxes and downstream dissolved residuals—generating higher cumulative TN reduction but incurring redundant coverage costs [79]. Finally, the CE advantage of CSAs is most apparent under limited budgets, as initial investments in the most critical areas yield the largest and most immediate reduction gains. Beyond a certain budget threshold, achieving additional proportions of reduction requires targeting more dispersed and lower-concentration sources, making further improvements exponentially more costly [80].

3.6. Optimal Management Schemes Under Different Cost Ranges

In determining optimal BMP combinations, environmental policymakers need to simultaneously take into account the EBs (TN reduction rates, Section 3.4), cost-effectiveness (CEs, Section 3.5), and the total available funding. The selected combinations should primarily meet the environmental standards and have an acceptable CE. For this, we screened the combinations with a TN reduction rate greater than 15% and a CE value below 30 kg/104 CNY·year. Ultimately, there were 75 watershed-wide combinations and 69 CSA combinations retained. To address varying financial capacities, we categorized total investment costs into six distinct ranges (below 30, 30–50, 50–70, 70–90, 90–110, and above 110 million CNY). We identified optimal BMP combinations for both CSA and watershed-wide implementation within each cost range by selecting those with either the highest CE values or highest TN reduction rates (Figure 9). Table 4 provides detailed specifications of these optimal combinations, including their constituent BMPs and performance metrics. All combinations were illustrated in Figures S3 and S4, and Table S9.
The highest and lowest costs of the watershed combinations were 225.3 million CNY and 21.6 million CNY, respectively, with 84% of the combinations having costs between 60 and 170 million CNY. For the CSAs combinations, the highest and lowest costs were 119.4 million CNY and 19.2 million CNY, respectively, with 80% of the combinations having costs between 30 and 90 million CNY.
In the cost range below 30 million CNY, the CE values were generally higher than the other cost ranges. The highest CE value occurred at the FR10+FS+PR scheme at CSAs (point 1), with a CE of 123.0 kg/104 CNY·year. Though not having the highest CE value, the combination FR10+FS+PR at the watershed (point 2) had the highest environmental benefits within the cost range, with a TN reduction rate of 23.3%. In the range of 30 to 50 million CNY, it was obvious that CSAs scale had more retained combinations than the entire watershed (12 versus 3). The highest CE value came from the combination FR25+FS+PR at CSAs, which was 98.6 kg/104 CNY·year (point 3). The second and third highest CE values were also from the combinations at the CSA scale. This demonstrated that at relatively lower cost ranges, applying the control measures at the concentrated CSAs was more cost-effective than at the watershed-wide areas. The FR25+FS+PR at the watershed (point 4) had the highest TN reduction rate of 30.2%. In the range of 50 to 70 million CNY, the CSAs’ combinations showed obviously superior performances than those at the watershed. The top CE value was 71.1 kg/104 CNY·year, from FR50+FS+PR at CSAs (point 5). At the same time, FR50+FS+PR+SC at CSAs (point 6) achieved the highest TN reduction rate of 29.8%, at a cost of 63.4 million CNY.
After reaching 70 million CNY, a shift occurred, as all optimal BMP combinations beyond this threshold were derived from the watershed-wide application. In the 70 to 90 million CNY cost range, the combination of FR50+FS+PR at the watershed (point 7) achieved the highest TN removal rate and CE, making it an irreplaceable option. This combination reached an impressive TN reduction rate of 42.5%, largely outperforming other combinations. The reason for this could be that it minimizes nitrogen fertilizer use across the entire watershed while integrating point-source reductions, which can reduce TN generation at the source (addressing both point and non-point sources). Additionally, FS was the most effective engineering measure within this cost range. And the CE for this combination reached 79.9 kg/104 CNY·year, far surpassing that of other combinations in the same cost range. In the 90 to 110 million CNY range, the highest TN reduction rate and CE were also achieved by FR25+FS+PR+TR at the watershed (point 8). However, this combination resulted in a TN reduction rate of 34.8% and a CE of 54.6 kg/104 CNY·year, both of which were lower than those achieved by FR50+FS+PR in the previous cost range. Therefore, it is recommended that if the budget falls within this range, the FR50+FS+PR combination at the watershed should still be the preferred choice. In the cost range above 110 million CNY, it was clear that the number of CSA combinations was significantly lower than that of watershed combinations. Implementing FR50+PR+TR across the watershed (point 9) yielded the highest CE in this range (58.0 kg/104 CNY·year). Unsurprisingly, deploying all measures across the watershed (point 10) achieved the highest TN reduction rate (47.5%). However, unless environmental policymakers aimed for the ultimate environmental benefits, this would not be a wise choice due to the prohibitively high costs. On the frontier curve of the relationship, it can be observed that the TN reduction rate no longer increased a lot after the cost exceeded 130 million CNY. This indicated that the total cost should be kept below 130 million CNY when implementing measures across the entire watershed.
These results establish a dynamic decision framework for policymakers, anchored by the critical 70 million CNY budget threshold. Below this level, prioritizing cost-effective CSA-targeted combinations achieves superior CE by concentrating on high-yield pollution sources—delivering 19.2% TN reduction at just 8.5% of the peak implementation costs. When exceeding 70 million CNY, shifting to watershed-wide implementation (e.g., FR50+FS+PR) addresses residual pollution from non-CSA areas, elevating reduction rates beyond 40% despite lower CE. To accommodate practical budgetary constraints, during unexpected funding reductions, core CSA measures (FR10, FS, PR) should be maintained to prevent TN rebound. Conversely, when more funding is available, the implementation should be expanded progressively—first to sub-watersheds adjacent to CSAs, before covering the entire watershed. This adaptive approach makes the plan flexible and cost-effective.

4. Conclusions

This study analyzed the spatiotemporal variation of total nitrogen (TN) load contributions in the Fuzhou River Basin and proposed optimal strategies for controlling TN nutrients, considering the environmental benefits, cost-effectiveness, and total cost. The key conclusions are as follows: (1) 46.7% of TN loads originated from agricultural planting, peaking between August and December due to seasonal fertilization and monsoon rainfall, while wastewater treatment plants dominated contributions in other months. (2) Non-point source CSAs clustered in central sub-watersheds with high cropland density, whereas point source CSAs localized in Sub-watershed 12 (containing >90% industrial emissions). (3) Increasing the number of BMPs consistently enhanced environmental benefits (EB) but diminished cost-effectiveness (CE), revealing an inherent conflict between efficiency and cost in the implementation of multiple measures. (4) CSA-targeted combinations (e.g., FR25+FS+PR) provided higher CE (98.6 kg/104 CNY· year) while maintaining moderate EB (20.0% TN reduction); watershed-wide deployment (e.g., FR50+FS+PR) achieved greater EB (42.5% TN reduction) with acceptable CE (79.9 kg/104 CNY year). This paper designed a BMPs assessment approach for TN reduction based on cost budget ranges, providing a generalizable theoretical method for dynamic optimization and budget allocation in multi-pollutant source watershed management decision-making.
It is important to note that this study assumed uniform application of BMPs across the entire watershed and within CSAs, without fully accounting for spatial heterogeneity. Furthermore, the effectiveness of BMPs is influenced by local variations in land use, soil type, and pollution sources, which remain a source of uncertainty—especially since current understanding relies mainly on modeling rather than empirical implementation. Future studies should therefore prioritize developing spatially adaptive BMP strategies tailored to individual sub-watershed conditions, while also increasing field validation to better assess and reduce TN exceedance in practice.

Supplementary Materials

The following supporting information can be downloaded at https://www.mdpi.com/article/10.3390/w17172651/s1. Table S1: The HRU delineation result; Table S2: The locations and discharge volumes of wastewater treatment plants and industries; Table S3: The population numbers and per capita TN emission rates at each township; Table S4: The livestock number and pollution of each township; Table S5: The livestock number and pollution of each township; Table S6: The parameters sensitivities; Table S7: The values of the USLE_P parameter; Table S8: The parameters of TR at different slopes; Table S9: The BMPs combinations; Figure S1: The sub-watersheds delineation; Figure S2: The locations of wastewater treatment plants and industries; Figure S3: The watershed combinations within the cost range; Figure S4: The CSAs combinations within the cost range.

Author Contributions

Conceptualization, Z.X. and B.Y.; methodology, Y.F. and H.Z.; software, H.Z.; validation, B.Y., Z.X., and Y.F.; formal analysis, Y.F.; investigation, M.C.; resources, Z.X.; data curation, M.C.; writing—original draft preparation, H.Z.; writing—review and editing, Y.F.; visualization, Y.F.; supervision, B.Y.; project administration, B.Y.; funding acquisition, Z.X. All authors have read and agreed to the published version of the manuscript.

Funding

This work is supported by the National Key Research and Development Program of China (grant number 2023YFC3208402) and the National Natural Science Foundation of China (grant numbers 52479002 and U21A20155).

Data Availability Statement

The original contributions presented in this study are included in the article/Supplementary Materials. Further inquiries can be directed to the corresponding author.

Conflicts of Interest

Author Huazhi Zhang was employed by the company Changjiang Schinta Software Technology Co., Ltd. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as potential conflicts of interest.

References

  1. Giglou, A.N.; Nazari, R.; Karimi, M.; Lawrence Museru, M.; Ntow Opare, K.; Nikoo, M.R. Future eco-hydrological dynamics: Urbanization and climate change effects in a changing landscape: A case study of Birmingham’s river basin. J. Clean. Prod. 2024, 447, 141320. [Google Scholar] [CrossRef]
  2. Whitall, D.; Castro, M.; Driscoll, C. Evaluation of management strategies for reducing nitrogen loadings to four US estuaries. Sci. Total Environ. 2004, 333, 25–36. [Google Scholar] [CrossRef]
  3. Carpenter, S.; Caraco, N.F.; Correll, D.L.; Howarth, R.; Sharpley, A.N.; Smith, V. Non-Point pollution of surface waters with phosphorus and nitrogen. Ecol. Appl. 1998, 8, 559–568. [Google Scholar] [CrossRef]
  4. Feng, L.; Wang, Y.; Hou, X.; Qin, B.; Kuster, T.; Qu, F.; Chen, N.; Paerl, H.W.; Zheng, C. Harmful algal blooms in inland waters. Nat. Rev. Earth Environ. 2024, 5, 631–644. [Google Scholar] [CrossRef] [PubMed]
  5. Zhang, W.; Rong, N.; Jin, X.; Meng, X.; Han, S.; Zhang, D.; Shan, B. Dissolved oxygen variation in the North China Plain river network region over 2011–2020 and the influencing factors. Chemosphere 2022, 287, 132354. [Google Scholar] [CrossRef]
  6. Chu, T.-W.; Shirmohammadi, A.; Montas, H.; Sadeghi, A. Evaluation of the SWAT model’s sediment and nutrient components in the piedmont physiographic region of Maryland. Trans. ASABE 2004, 47, 1523–1538. [Google Scholar] [CrossRef]
  7. Yuan, L.; Sinshaw, T.; Forshay, K.J. Review of Watershed-Scale Water Quality and Nonpoint Source Pollution Models. Geosciences 2020, 10, 25. [Google Scholar] [CrossRef] [PubMed]
  8. Devia, G.K.; Ganasri, B.P.; Dwarakish, G.S. A review on hydrological models. Aquat. Procedia 2015, 4, 1001–1007. [Google Scholar] [CrossRef]
  9. Chen, D.; Shen, H.; Hu, M.; Wang, J.; Zhang, Y.; Dahlgren, R.A. Legacy nutrient dynamics at the watershed scale: Principles, modeling, and implications. Adv. Agron. 2018, 149, 237–313. [Google Scholar] [CrossRef]
  10. Ongley, E.D.; Xiaolan, Z.; Tao, Y. Current status of agricultural and rural non-point source Pollution assessment in China. Environ. Pollut. 2010, 158, 1159–1168. [Google Scholar] [CrossRef]
  11. Zheng, Y.; Han, F. Markov Chain Monte Carlo (MCMC) uncertainty analysis for watershed water quality modeling and management. Stoch. Environ. Res. Risk Assess. 2016, 30, 293–308. [Google Scholar] [CrossRef]
  12. Cano, O.; Barkdoll, B. Multiobjective, socioeconomic, boundary-emanating, nearest distance algorithm for stormwater low-impact BMP selection and placement. J. Water Resour. Plan. Manag. 2016, 143, 05016013. [Google Scholar] [CrossRef]
  13. Maringanti, C.; Chaubey, I.; Popp, J. Development of a multiobjective optimization tool for the selection and placement of BMPs for nonpoint source pollution control. Water Resour. Res. 2009, 45, W06406. [Google Scholar] [CrossRef]
  14. McDowell, R.; Cosgrove, G.; Orchiston, T.; Chrystal, J. A cost-effective management practice to decrease phosphorus loss from dairy farms. J. Environ. Qual. 2014, 43, 2044–2052. [Google Scholar] [CrossRef]
  15. Xie, H.; Chen, L.; Shen, Z. Assessment of agricultural Best Management Practices using Models: Current issues and future perspectives. Water 2015, 7, 1088–1108. [Google Scholar] [CrossRef]
  16. Krysanova, V.; White, M. Advances in water resources assessment with SWAT—An overview. Hydrol. Sci. J. 2015, 60, 771–783. [Google Scholar] [CrossRef]
  17. Haas, M.B.; Guse, B.; Fohrer, N. Assessing the impacts of Best Management Practices on nitrate pollution in an agricultural dominated lowland catchment considering environmental protection versus economic development. J. Environ. Manag. 2017, 196, 347–364. [Google Scholar] [CrossRef]
  18. Leta, M.K.; Waseem, M.; Rehman, K.; Tränckner, J. Sediment yield estimation and evaluating the best management practices in Nashe watershed, Blue Nile Basin, Ethiopia. Environ. Monit. Assess. 2023, 195, 716–736. [Google Scholar] [CrossRef]
  19. Zhang, X.; Chen, P.; Dai, S.; Han, Y. Analysis of non-point source nitrogen pollution in watersheds based on SWAT model. Ecol. Indic. 2022, 138, 108881. [Google Scholar] [CrossRef]
  20. Winchell, M.F.; Folle, S.; Meals, D.; Moore, J.; Srinivasan, R.; Howe, E.A. Using SWAT for sub-field identification of phosphorus critical source areas in a saturation excess runoff region. Hydrol. Sci. J. 2015, 60, 844–862. [Google Scholar] [CrossRef]
  21. Niraula, R.; Kalin, L.; Srivastava, P.; Anderson, C.J. Identifying critical source areas of nonpoint source pollution with SWAT and GWLF. Ecol. Modell. 2013, 268, 123–133. [Google Scholar] [CrossRef]
  22. Ahsan, A.; Das, S.K.; Khan, M.H.R.B.; Ng, A.W.M.; Al-Ansari, N.; Ahmed, S.; Imteaz, M.; Tariq, M.A.U.R.; Shafiquzzaman, M. Modeling the impacts of best management practices (BMPs) on pollution reduction in the Yarra River catchment, Australia. Appl. Water Sci. 2023, 13, 98. [Google Scholar] [CrossRef]
  23. Wu, L.; Liu, X.; Chen, J.; Li, J.; Yu, Y.; Ma, X. Efficiency assessment of best management practices in sediment reduction by investigating cost-effective tradeoffs. Agric. Water Manag. 2022, 265, 107546. [Google Scholar] [CrossRef]
  24. Sharpley, A.; Jarvie, H.P.; Buda, A.; May, L.; Spears, B.; Kleinman, P. Phosphorus legacy: Overcoming the effects of past management practices to mitigate future water quality impairment. J. Environ. Qual. 2013, 42, 1308–1326. [Google Scholar] [CrossRef]
  25. Liu, W.; Zhang, L.; Wu, H.; Wang, Y.; Zhang, Y.; Xu, J.; Wei, D.; Zhang, R.; Yu, Y.; Wu, D.; et al. Strategy for cost-effective BMPs of non-point source pollution in the small agricultural watershed of Poyang Lake: A case study of the Zhuxi River. Chemosphere 2023, 333, 138949. [Google Scholar] [CrossRef]
  26. Babaei, H.; Zakermoshfegh, M.; Delavar, M. Identifying critical sources and evaluate the Best Management Practices to control nutrient load of the Dez River Basin using the SWAT model. J. Watershed Manag. Res. 2020, 11, 142–154. [Google Scholar] [CrossRef]
  27. Liu, Y.; Guo, T.; Wang, R.; Engel, B.A.; Flanagan, D.C.; Li, S.; Pijanowski, B.C.; Collingsworth, P.D.; Lee, J.G.; Wallace, C.W. A SWAT-based optimization tool for obtaining cost-effective strategies for agricultural conservation practice implementation at watershed scales. Sci. Total Environ. 2019, 691, 685–696. [Google Scholar] [CrossRef] [PubMed]
  28. Arabi, M.; Govindaraju, R.S.; Hantush, M.M. Cost-effective allocation of watershed management practices using a genetic algorithm. Water Resour. Res. 2006, 42, 2405–2411. [Google Scholar] [CrossRef]
  29. Babaei, H.; Nazari-Sharabian, M.; Karakouzian, M.; Ahmad, S. Identification of Critical Source Areas (CSAs) and Evaluation of Best Management Practices (BMPs) in Controlling Eutrophication in the Dez River Basin. Environments 2019, 6, 20. [Google Scholar] [CrossRef]
  30. Rudra, R.P.; Mekonnen, B.A.; Shukla, R.; Shrestha, N.K.; Goel, P.K.; Daggupati, P.; Biswas, A. Currents Status, Challenges, and Future Directions in Identifying Critical Source Areas for Non-Point Source Pollution in Canadian Conditions. Agriculture 2020, 10, 468. [Google Scholar] [CrossRef]
  31. Wang, M.; Chen, L.; Wu, L.; Zhang, L.; Xie, H.; Shen, Z. Review of Nonpoint Source Pollution Models: Current Status and Future Direction. Water 2022, 14, 3217. [Google Scholar] [CrossRef]
  32. Razavi, H.S.; Toscano, G.; Nejadhashemi, A.P.; Deb, K.; Linker, L. Next-generation techniques for parameter reduction for BMP multiobjective optimization in watershed planning. Environ. Model. Softw. 2025, 193, 106651. [Google Scholar] [CrossRef]
  33. Dong, L.; Xin, Z.; Song, C.; Zhang, C.; Bai, X. Applicability of statistical and modeling methods in assessing the basin-scale grey water footprint. Ecol. Indic. 2022, 143, 109345. [Google Scholar] [CrossRef]
  34. Luo, X.; Tang, A.; Shi, K.; Wu, L.; Li, W.; Shi, W.; Shi, X.; Erisman, J.W.; Zhang, F.; Liu, X. Chinese coastal seas are facing heavy atmospheric nitrogen deposition. Environ. Res. Lett. 2014, 9, 095007. [Google Scholar] [CrossRef]
  35. Arnold, J.G.; Kiniry, J.R.; Srinivasan, R.; Williams, J.R.; Haney, E.B.; Neitsch, S.L. SWAT 2012 Input/Output Documentation; Texas Water Resources Institute: College Station, TX, USA, 2013; pp. 1–650. [Google Scholar]
  36. Gassman, P.; Reyes, M.; Green, C.; Arnold, J. The Soil and Water Assessment Tool: Historical Development, Applications, and Future Research Directions. Trans. ASABE 2007, 50, 1211–1250. [Google Scholar] [CrossRef]
  37. Raman, V. Determination of parameters for watershed delineation using various satellite derived digital elevation models and its accuracy Graphical abstract. Glob. NEST J. 2023, 25, 124–137. [Google Scholar] [CrossRef]
  38. Her, Y.; Frankenberger, J.; Chaubey, I.; Srinivasan, R. Threshold Effects in HRU Definition of the Soil and Water Assessment Tool. Trans. ASABE 2015, 58, 367–378. [Google Scholar] [CrossRef]
  39. Li, H.; Zhang, J.; Zhang, S.; Zhang, W.; Zhang, S.; Yu, P.; Song, Z. A framework to assess spatio-temporal variations of potential non-point source pollution risk for future land-use planning. Ecol. Indic. 2022, 137, 108751. [Google Scholar] [CrossRef]
  40. Wei, P.; Ouyang, W.; Gao, X.; Hao, F.; Hao, Z.; Liu, H. Modified control strategies for critical source area of nitrogen (CSAN) in a typical freeze-thaw watershed. J. Hydrol. 2017, 551, 518–531. [Google Scholar] [CrossRef]
  41. Wu, Z.; Liang, H.; Wu, H.; Xie, Z.; Liu, H.; Zhang, J.; Zhu, J.; Zheng, B.; Wan, W. Application of improved risk assessment methods to best management practices research for non-point source pollution: From a hilly mountainous-dominated region of Southern China. Ecol. Indic. 2024, 165, 112223. [Google Scholar] [CrossRef]
  42. Dibaba, W.T.; Demissie, T.A.; Miegel, K. Prioritization of Sub-Watersheds to Sediment Yield and Evaluation of Best Management Practices in Highland Ethiopia, Finchaa Catchment. Land 2021, 10, 650. [Google Scholar] [CrossRef]
  43. Li, H.; Zhu, N.; Qiao, J.; Tang, J. Evaluating the long-term effects of best management practices on pollution reduction and soil quality improvement in sloping farmland of the Three Gorges Reservoir area. Agric. Water Manag. 2024, 297, 108844. [Google Scholar] [CrossRef]
  44. Zettam, A.; Briak, H.; Kebede, F.; Ouallali, A.; Hallouz, F.; Taleb, A. Efficiencies of best management practices in reducing nitrate pollution of the Sebdou River, a semi-arid Mediterranean agricultural catchment (North Africa). River Res. Appl. 2022, 38, 613–624. [Google Scholar] [CrossRef]
  45. Arnold, J.G.; Srinivasan, R.; Muttiah, R.S.; Williams, J.R. Large area hydrologic modeling and assessment part I: Model development. JAWRA J. Am. Water Resour. Assoc. 1998, 34, 73–89. [Google Scholar] [CrossRef]
  46. Huang, K. Research on Non-Point Source Pollution Management Measures of Danjiang River Basin Based on SWAT Mdoel. Master’s Thesis, Xi’an University of Science and Technology, Xi’an, China, 2020. [Google Scholar]
  47. Huang, W. Research on Characteristics and Control of Non-Point Source Pollution in Zhupi River Basin Based on SWAT Model. Master’s Thesis, Huazhong University of Science and Technology, Wuhan, China, 2022. [Google Scholar]
  48. Kaini, P.; Artita, K.; Nicklow, J.W. Optimizing structural Best Management Practices using SWAT and genetic algorithm to improve water quality goals. Water Resour. Manag. 2012, 26, 1827–1845. [Google Scholar] [CrossRef]
  49. Ruan, B.; Wu, L.; Guo, J.; Du, B. Assessment and selection of the best management practices for the Jinghe River Basin considering the comprehensive benefits of environmental costs. China Environ. Sci. 2024, 44, 319–328. [Google Scholar] [CrossRef]
  50. Teshager, A.D.; Gassman, P.W.; Secchi, S.; Schoof, J.T. Simulation of targeted pollutant-mitigation-strategies to reduce nitrate and sediment hotspots in agricultural watershed. Sci. Total Environ. 2017, 607–608, 1188–1200. [Google Scholar] [CrossRef]
  51. Xin, W.; Fengyan, W.; Tiesong, H.; Xiang, Z.; Xiang, L.; Jie, H. Modeling competitive-cooperative relationship between flood control and hydropower generation within hierarchical gaming structure of reservoir operation. J. Hydrol. 2023, 620, 129429. [Google Scholar] [CrossRef]
  52. Du, J.; Li, H.; Zhao, X.; Li, J. Simulation of pollution control effect in the Weihe River Valley with SWAT model. J. Ecol. Rural. Environ. 2016, 32, 563–569. [Google Scholar] [CrossRef]
  53. Jia, A.; Wang, H.; Xu, S. Purification Effect of Composite Constructed Wetland on Polluted River Water in Dalian Area. Water Resour. Power 2017, 35, 26–29. [Google Scholar]
  54. Zheng, R. Nitrogen Pollution Investigation and Study on Water Purification Technology in the Upper Fuzhou River. Master’s Thesis, Dalian Univeristy of Technology, Dalian, China, 2018. [Google Scholar]
  55. Li, Z.; Xue, W.; Sereewatthanawut, I. Evaluation of best management practices (BMPs) for mitigating non-point source pollution in the Wang River Basin of Thailand based on the SWAT Model. Environ. Technol. Innov. 2025, 39, 104327. [Google Scholar] [CrossRef]
  56. Chen, J. National-Scale Environmental and Economic Comprehensiv Eassessment of Tailwater Constructed Wetlands for Wastewater Treatment. Master’s Thesis, Chinese Academy of Environmental Sciences, Beijing, China, 2023. [Google Scholar]
  57. Chen, Z.M.; Chen, B.; Zhou, J.B.; Li, Z.; Zhou, Y.; Xi, X.R.; Lin, C.; Chen, G.Q. A vertical subsurface-flow constructed wetland in Beijing. Commun. Nonlinear Sci. Numer. Simul. 2008, 13, 1986–1997. [Google Scholar] [CrossRef]
  58. Wang, X.; Zhang, Y.; Ou, Y.; Duan, S. Optimization and economic evaluation on cost-benefit of Best Management Practices in nonpoint source pollution control. Ecol. Environ. 2009, 18, 540–548. [Google Scholar] [CrossRef]
  59. Zhang, Y. Observation and Simulation for Soil Fertility and Crop Yield Effects of Long-Term Rotational Conservation Tillage Andfertilization on Winter Wheat-Spring Maize Rotation Field of Loess Plateau. Ph.D. Thesis, North West Agriculture and Forestry University, Yangling, China, 2018. [Google Scholar]
  60. Moriasi, D.; Gitau, M.; Pai, N.; Daggupati, P. Hydrologic and Water Quality Models: Performance Measures and Evaluation Criteria. Trans. ASABE 2015, 58, 1763–1785. [Google Scholar] [CrossRef]
  61. Raaj, S.; Gupta, V.; Singh, V.; Shukla, D. A novel framework for peak flow estimation in the himalayan river basin by integrating SWAT model with machine learning based approach. Earth Sci. Inform. 2024, 17, 211–226. [Google Scholar] [CrossRef]
  62. Fan, L.; Yuan, Y.; Ying, Z.; Lam, S.K.; Liu, L.; Zhang, X.; Liu, H.; Gu, B. Decreasing farm number benefits the mitigation of agricultural non-point source pollution in China. Environ. Sci. Pollut. Res. 2019, 26, 464–472. [Google Scholar] [CrossRef]
  63. Zou, L.; Liu, Y.; Wang, Y.; Hu, X. Assessment and analysis of agricultural non-point source pollution loads in China: 1978–2017. J. Environ. Manag. 2020, 263, 110400. [Google Scholar] [CrossRef] [PubMed]
  64. Han, J.; Xin, Z.; Han, F.; Xu, B.; Wang, L.; Zhang, C.; Zheng, Y. Source contribution analysis of nutrient pollution in a P-rich watershed: Implications for integrated water quality management. Environ. Pollut. 2021, 279, 116885. [Google Scholar] [CrossRef]
  65. Fu, D.; Gong, Y.; Chen, C.; Gui, X.; Liu, H.; Chen, S.; Ren, J.; Hou, B. Nitrogen and Phosphorus Loading Characteristics of Agricultural Non-Point Sources in the Tuojiang River Basin. Water 2023, 15, 3503. [Google Scholar] [CrossRef]
  66. Franco, A.A.; Iglesias-Arroyo, D.; Egea-Corbacho, Á.; Martín-García, A.P.; Quiroga, J.M.; Coello, M.D. Influence of tourism on microplastic contamination at wastewater treatment plants in the coastal municipality of Chiclana de la Frontera. Sci. Total Environ. 2023, 900, 165573. [Google Scholar] [CrossRef]
  67. Feng, W.; Wang, S.; Tan, K.; Ma, L.; Hu, C. Simulation of spatial and temporal variation of nitrate leaching in the vadose zone of alluvial regions on a large regional scale. Sci. Total Environ. 2024, 916, 170114. [Google Scholar] [CrossRef]
  68. Chen, M.; Lei, Q.; Chen, L.; Du, X.; Liu, H.; Luo, J. Best management practices for reducing nutrient loads under diverge hydrological conditions: A case study in an agricultural watershed in Southwest China. Ecol. Indic. 2025, 176, 113631. [Google Scholar] [CrossRef]
  69. Jamshidi, S.; Imani, S.; Delavar, M. Impact assessment of Best Management Practices on the water footprint of agricultural productions. Int. J. Environ. Res. 2020, 14, 641–652. [Google Scholar] [CrossRef]
  70. Pongpetch, N.; Suwanwaree, P.; Yossapol, C.; Dasananda, S.; Kongjun, T. Using SWAT to assess the critical areas and nonpoint source pollution reduction Best Management Practices in Lam Takong River Basin, Thailand. EnvironmentAsia 2015, 8, 41–52. [Google Scholar] [CrossRef]
  71. Uniyal, B.; Jha, M.K.; Verma, A.K.; Anebagilu, P.K. Identification of critical areas and evaluation of best management practices using SWAT for sustainable watershed management. Sci. Total Environ. 2020, 744, 140737. [Google Scholar] [CrossRef]
  72. Shen, W.; Chen, R.; Zhao, X.; Lu, X.; Yan, H.; Wang, L. Implementing Best Management Practices in Complex Agricultural Watersheds: Insights from High-Resolution Nitrogen Load Dynamics Analysis. Water 2025, 17, 821. [Google Scholar] [CrossRef]
  73. Wigington Jr, P.; Griffith, S.; Field, J.; Baham, J.; Horwath, W.; Owen, J.; Davis, J.; Rain, S.; Steiner, J. Nitrate removal effectiveness of a riparian buffer along a small agricultural stream in western Oregon. J. Environ. Qual. 2003, 32, 162–170. [Google Scholar] [CrossRef]
  74. Haycock, N.E.; Pinay, G. Groundwater nitrate dynamics in grass and poplar vegetated riparian buffer strips during the winter. J. Environ. Qual. 1993, 22, 273–278. [Google Scholar] [CrossRef]
  75. Liu, Y.; Wang, R.; Guo, T.; Engel, B.A.; Flanagan, D.C.; Lee, J.G.; Li, S.; Pijanowski, B.C.; Collingsworth, P.D.; Wallace, C.W. Evaluating efficiencies and cost-effectiveness of best management practices in improving agricultural water quality using integrated SWAT and cost evaluation tool. J. Hydrol. 2019, 577, 123965. [Google Scholar] [CrossRef]
  76. Panagopoulos, Y.; Makropoulos, C.; Mimikou, M. Reducing surface water pollution through the assessment of the cost-effectiveness of BMPs at different spatial scales. J. Environ. Manag. 2011, 92, 2823–2835. [Google Scholar] [CrossRef]
  77. Pionke, H.B.; Gburek, W.J.; Sharpley, A.N. Critical source area controls on water quality in an agricultural watershed located in the Chesapeake Basin. Ecol. Eng. 2000, 14, 325–335. [Google Scholar] [CrossRef]
  78. Chang, D.; Lai, Z.; Li, S.; Li, D.; Zhou, J. Critical source areas’ identification for non-point source pollution related to nitrogen and phosphorus in an agricultural watershed based on SWAT model. Environ. Sci. Pollut. Res. 2021, 28, 47162–47181. [Google Scholar] [CrossRef] [PubMed]
  79. Li, Q.; Yu, Q.; Wang, F.; Yan, W.; Wang, J. Nitrogen removal in the Chaohu Lake, China: Implication in estimating lake N uptake velocity and modelling N removal efficiency of large lakes and reservoirs in the Changjiang River network. Ecol. Indic. 2021, 124, 107353. [Google Scholar] [CrossRef]
  80. Rabotyagov, S.S.; Campbell, T.D.; White, M.; Arnold, J.G.; Atwood, J.; Norfleet, M.L.; Kling, C.L.; Gassman, P.W.; Valcu, A.; Richardson, J.; et al. Cost-effective targeting of conservation investments to reduce the northern Gulf of Mexico hypoxic zone. Proc. Natl. Acad. Sci. USA 2014, 111, 18530–18535. [Google Scholar] [CrossRef] [PubMed]
Water 17 02651 g001
Figure 1. The Fuzhou River Basin (FRB).
Figure 1. The Fuzhou River Basin (FRB).
Water 17 02651 g001
Water 17 02651 g002
Figure 2. Overall workflow.
Figure 2. Overall workflow.
Water 17 02651 g002
Water 17 02651 g003
Figure 3. Comparison of simulated and observed (a) streamflow at Guanjiatun (GJT) hydrological station, and (b) total nitrogen (TN) load at Santaizi (STZ) water quality station.
Figure 3. Comparison of simulated and observed (a) streamflow at Guanjiatun (GJT) hydrological station, and (b) total nitrogen (TN) load at Santaizi (STZ) water quality station.
Water 17 02651 g003
Water 17 02651 g004
Figure 4. Average monthly and annual TN load of pollution sources.
Figure 4. Average monthly and annual TN load of pollution sources.
Water 17 02651 g004
Water 17 02651 g005
Figure 5. TN source proportions in each sub-watershed of FRB.
Figure 5. TN source proportions in each sub-watershed of FRB.
Water 17 02651 g005
Water 17 02651 g006
Figure 6. Non-point source pollution load risk levels.
Figure 6. Non-point source pollution load risk levels.
Water 17 02651 g006
Water 17 02651 g007
Figure 7. CSAs identification.
Figure 7. CSAs identification.
Water 17 02651 g007
Water 17 02651 g008
Figure 8. The TN reduction rates (a) and CEs (b) box plots of BMPs combinations.
Figure 8. The TN reduction rates (a) and CEs (b) box plots of BMPs combinations.
Water 17 02651 g008
Water 17 02651 g009
Figure 9. Optimal BMPs combinations within different cost ranges.
Figure 9. Optimal BMPs combinations within different cost ranges.
Water 17 02651 g009
Table 1. Best Management Practices (BMPs) setting parameters and unit costs.
Table 1. Best Management Practices (BMPs) setting parameters and unit costs.
BMPsParameters SettingUnit Costs
PRAdjust point source emissions (wastewater treatment plants and direct industrial emissions) to 70% of original emissions.0.23 CYN/t point source sewage
FR (10, 25, 50)Modify parameters in the fertilizer database: nitrogen content in fertilizer reduced to 90%, 75%, and 50% of original value, respectively.204.1, 510.2, 1020.2 CYN/hm2
SCAdjust STRIP_N to 0.19, STRIP_CN to original CN2 value-3, STRIP_C to 0.2.375 CYN/hm2
FSSet FILTER_RATIO to 5, FILTER_CON to 0, FILTER_CH to 0, and filterw parameters to default values.1150 CYN/hm2
TRSet TERR_CN to original CN2 value-5.760 CYN/hm2
CTAdjust CONT_CN to original CN2 value-2.750 CYN/hm2
Table 2. TN reduction rate of each BMP.
Table 2. TN reduction rate of each BMP.
WatershedCSAs
Types of BMPsBMPsTN Load (t)Reduction Rate (%)TN Load (t)Reduction Rate (%)
BAS1523.40.01523.40.0
Non-engineering measuresCT1407.37.61462.34.0
SC1337.612.21424.36.5
FR101448.24.91475.93.1
FR251341.412.01410.77.4
FR501153.824.31292.315.2
Engineering measuresPR1405.37.81433.65.9
TR1287.015.51397.38.3
FS1334.412.41423.56.6
Table 3. CE of each single BMP (kg/104 CNY·year).
Table 3. CE of each single BMP (kg/104 CNY·year).
Types of BMPsBMPsWatershedCSAs
Non-engineering measuresCT24.322.0
SC38.871.3
FR1057.862.8
FR2555.959.6
FR5056.861.1
Engineering measuresPR129.9121.4
TR48.844.7
FS219.5234.2
Table 4. Selected optimal BMP combinations at watershed and CSAs.
Table 4. Selected optimal BMP combinations at watershed and CSAs.
Total Cost Range
(Million CNY)		Selection CriteriaPoint
Number		BMPs CombinationsImplementation AreaTN Reduction
Rate (%)		Total Cost
(Million CNY)		CE
(kg/104 CNY·Year)
<30The highest CE1FR10+FS+PRCSAs15.619.2123.0
The highest EB2FR10+FS+PRWatershed23.329.0122.1
30–50The highest CE3FR25+FS+PRCSAs20.030.698.6
The highest EB4FR25+FS+PRWatershed30.248.694.9
50–70The highest CE5FR50+FS+PRCSAs27.859.271.1
The highest EB6FR50+FS+PR+SCCSAs29.863.471.1
70–90The highest CE7FR50+FS+PRWatershed42.581.179.9
The highest EB7FR50+FS+PRWatershed42.581.179.9
90–110The highest CE8FR25+FS+PR+TRWatershed34.897.054.6
The highest EB8FR25+FS+PR+TRWatershed34.897.054.6
>110The highest CE9FR50+PR+TRWatershed46.1121.058.0
The highest EB10FR50+FS+PR+TR+SC+CTWatershed47.5225.332.1
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Fan, Y.; Zhang, H.; Yu, B.; Cong, M.; Xin, Z. Spatial Targeting and Budget-Adaptive Optimization of Best Management Practices for Cost-Effective Nitrogen Reduction. Water 2025, 17, 2651. https://doi.org/10.3390/w17172651

AMA Style

Fan Y, Zhang H, Yu B, Cong M, Xin Z. Spatial Targeting and Budget-Adaptive Optimization of Best Management Practices for Cost-Effective Nitrogen Reduction. Water. 2025; 17(17):2651. https://doi.org/10.3390/w17172651

Chicago/Turabian Style

Fan, Yunkai, Huazhi Zhang, Bing Yu, Ming Cong, and Zhuohang Xin. 2025. "Spatial Targeting and Budget-Adaptive Optimization of Best Management Practices for Cost-Effective Nitrogen Reduction" Water 17, no. 17: 2651. https://doi.org/10.3390/w17172651

APA Style

Fan, Y., Zhang, H., Yu, B., Cong, M., & Xin, Z. (2025). Spatial Targeting and Budget-Adaptive Optimization of Best Management Practices for Cost-Effective Nitrogen Reduction. Water, 17(17), 2651. https://doi.org/10.3390/w17172651

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop