Greater Sustainability in the Future of Hanjiang River Under Climate Change: The Case of Nitrogen

Abstract

Water resources are essential for human survival and sustainable development. However, the global freshwater scarcity, exacerbated by climate change, presents significant sustainability challenges. Using the SWAT model, we simulated the spatiotemporal distribution of total nitrogen (TN) in the Hangjiang River Basin from 2005 to 2020. The average TN concentration was 2.16 mg/L, with the soil nitrogen pool contributing 92.78% of emissions, highlighting the need to address the soil nitrogen legacy. Sampling showed average concentrations of TN, nitrate, ammonium, nitrite, and dissolved organic nitrogen at 3.01 mg/L, 1.66 mg/L, 0.21 mg/L, 0.02 mg/L, and 1.11 mg/L, respectively. Precipitation accounted for 61.4% of nitrogen emission variability, indicating that water resource sustainability will be significantly influenced by climate change. Projections indicated that from 2020 to 2050, climate change will increase runoff by 6.19 m³/s and reduce TN concentration by 0.004 mg/L annually, potentially enhancing the overall sustainability of water resources. It’s necessary to continue strengthening the prevention and control of agricultural non-point source pollution and reduce nitrogen discharge to further enhance water resource security for the Beijing–Tianjin–Hebei development. The findings provide critical insights to inform policies aimed at protecting water sources and ensuring public water safety.

1. Introduction

Adequate and high-quality freshwater resources are essential for the sustainable development of human society [1,2]. Currently, approximately half of the global population faces unsustainable water supply, with nearly one-quarter residing in regions facing water scarcity [3]. In 2022, per capita water resources in eight provinces, including Tianjin, Ningxia, and Beijing, fell below the international threshold for extreme water scarcity of 500 m³ [4]. Additionally, global river pollution, particularly from agricultural non-point sources of nitrogen and phosphorus, remains a critical issue [5,6]. Huang et al. [7] pointed out that despite significant improvements in water quality in China, nitrogen contributes to approximately 48% of river pollution. Therefore, controlling non-point source pollution of nitrogen and phosphorus is vital for sustainable development.

The core of sustainable development is the preservation of development potential and opportunity in the future [8]. While current monitoring techniques could effectively identify and respond to pollutants in runoff, climate change introduces considerable uncertainty in water quality management. For example, the influence of precipitation variability on pollution load exhibits a dual nature. When precipitation increases significantly, or the frequency of extreme precipitation events rises, intensified surface erosion transports a greater quantity of pollutants into rivers [9,10]. Concurrently, increased runoff generation on land surfaces and direct precipitation inputs into water bodies augment the overall streamflow, which in turn dilutes the pollution load and mitigates its environmental impact [11]. A study predicts a roughly 50% increase in the concentrations of ammonia nitrogen, organic nitrogen, and total nitrogen in the Ganges River within the time frame of 2040 to 2060 [12]. In contrast, the ammonia nitrogen concentration in the upper reaches of the Huai River basin is predicted to decrease by 0.2% to 42.6% between 2020 and 2050 [10]. Therefore, whether climate change exerts positive or negative impacts on surface water pollution and to what extent remains a crucial issue for future strategic planning and management of water resources.

The Hanjiang River, recognized as the largest tributary of the Yangtze River with a total length of 1577 km, is endowed with abundant water resources and superior water quality [13,14]. It serves as a vital water source and critical conservation zone for the Central Line Project of South-to-North Water Diversion as well as the Yangtze River Diversion to Wei River Project [15]. Numerous studies assess the runoff of the Hanjiang River to determine its capacity for spatiotemporal water resource allocation. Zhao et al. [16] predicted that the runoff of the Hanjiang River would gradually increase at an annual rate of 4.747 m³/s from 2020 to 2035 under natural development scenarios. Yue et al. [17], utilizing multiple CMIP6 climate models, found an upward trend in the runoff of the upper and middle reaches of the Hanjiang River in the future, with a slightly higher increase under a low-carbon emission pathway compared to a high-carbon emission pathway.

Despite the relatively optimistic water volume forecasts for the Hanjiang River basin in the future, current research on nitrogen pollution is predominantly centered on status analysis [18]. Zhang et al. [19] observed the total nitrogen (TN) concentration fluctuating between 0.22 and 2.91 mg/L from 2017 to 2019, with an ascending trend that signals potential water quality risks. Research mainly emphasizes basin planning and management, quantifying annual nitrogen loads, and proposing management measures for polluted areas [20]. Based on the HYPE model, Yang and Li [21] calculated the TN pollution intensity to be within the range of 0.354 to 6.139 kg/ha, highlighting a notable load in the southern part of the basin. Utilizing the SWAT model, Li et al. [22] estimated the annual average TN load at the Ankang hydrological station, advocating for a comprehensive approach including stubble coverage, grassed waterway, and returning farmland to forest land as effective control measures. However, whether future nitrogen concentrations in the Hanjiang River basin will meet standards and continue to provide high-quality water remains an open question. The quality of water is directly linked to the health of the people within the water catchment areas. Therefore, the investigation of prospective nitrogen pollution dynamics is of paramount importance for the establishment of ecological conservation strategies for water resources and for safeguarding the sustainable development of the Beijing–Tianjin–Hebei region.

The Soil and Water Assessment Tool (SWAT) model has demonstrated superior efficacy in modeling hydrological processes and water quality at the basin scale [16,17,20,23]. In this study, the SWAT model was used to simulate the spatiotemporal distribution of TN concentration in the Hanjiang River basin from 2005 to 2020 and to identify key pollution sources and critical emission drivers. This analysis aims to provide insights for the development and implementation of targeted non-point source pollution management strategies. Additionally, the study examined the distribution of nitrogen across different chemical forms within the basin, offering a comprehensive overview of the current nitrogen pollution situation. Beyond focusing on water supply and demand through water diversion projects, the study placed emphasis on the impacts of climate change on nitrogen pollution trends. Projections of TN emissions under different shared socioeconomic pathways (SSPs) from 2020 to 2050 were also made, providing valuable information for the sustainability of water resources in the region. The findings aim to support scientific decision-making for water resource management and the formulation of sustainable development strategies, ensuring access to sufficient and clean water for local residents.

2. Materials and Methods

2.1. Study Area

The Hanjiang River is one of the principal tributaries of the Yangtze River. Its upper reaches are delineated by the Qinling Mountains to the north and the Dabashan Mountains to the south, with a geographical location spanning 110.5–116° E and 31.5–34.5° N (Figure 1). The basin primarily encompasses the cities of Hanzhong, Ankang, and Shiyan, covering an area of approximately 74,000 km². Characterized by a subtropical monsoon climate, the region experiences an average annual temperature ranging from 12 to 15 °C, accompanied by average annual precipitation between 750 and 1100 mm, predominantly occurring from June to October. This ample rainfall has given rise to an extensive network of rivers within the basin, including notable tributaries such as the Du River and the Jia River, which contribute to the abundant water resources in the region. The hydrological richness has also facilitated a high vegetation cover, reaching up to 76%, thus supporting a diverse and robust ecological environment. However, the region’s economic development remains relatively limited, creating a complex tension between the urgent need for economic growth and the critical demands of environmental conservation.

2.2. Data Collection

Data utilized for the simulation within the SWAT model include: (1) The Digital Elevation Model (DEM), a 90 m resolution DEM, serving as the foundational topographical input, was sourced from the Resource and Environmental Science Data Platform (https://www.resdc.cn/, accessed on 15 July 2024). (2) Land use and land cover data: The 2020 Land Use Dataset [24], with a resolution of 1 km, categorizes into arable land, grassland, woodland, water area, construction land, and unused land, also obtained from the Resource and Environmental Science Data Platform. (3) Soil data: The Harmonized World Soil Database (https://www.fao.org/soils-portal/data-hub/soil-maps-and-databases/, accessed on 15 July 2024) provided a soil type dataset with a resolution of 1 km. (4) Current climate data: Daily parameters from 2000 to 2022, including precipitation, temperature, wind speed, solar radiation, and relative humidity, were obtained from the China Meteorological Data Network (https://data.cma.cn/, accessed on 15 July 2024). (5) Future climate scenarios: Projections for future climate change in terms of precipitation and temperature were derived from general circulation models (GCM) in the CMIP6 framework, including FGOALS-g3, GFDL-ESM4, MIROC6, MRI-ESM2-0, NESM3, NorESM2-LM, and NorESM2-MM, with data accessed through the Earth System Grid Federation (https://aims2.llnl.gov/search, accessed on 15 July 2024). (6) Runoff and TN concentration: Monthly runoff data from January 2014 to September 2022 were extracted from the Shaanxi Hydrological and Water Resources Information Network (http://www.shxsw.com.cn/, accessed on 15 July 2024), while monthly water quality data from January 2021 to December 2022 were sourced from the National Surface Water Quality Data Release System (https://waterpub.cnemc.cn:10001/, accessed on 15 July 2024). (7) TN emission: The nitrogen emissions from non-point sources, such as livestock farming, rural life, and the application of chemical fertilizers, were quantified by integrating the statistical survey of emission sources and the calculation methods and coefficients manual (http://www.mee.gov.cn, accessed on 15 July 2024) for pollution discharge.

Field sampling was conducted at four intervals: September 2021, March 2022, October 2023, and March 2024. At each sampling site (Figure 1), 1 L of water sample was collected using a rinsed polytetrafluoroethylene plastic bottle, with concentrated sulfuric acid added to adjust the pH to less than 2 for preservation. The samples were stored at 4 °C and analyzed within 7 days. After filtration through a 0.45 μm microporous membrane, the water samples were analyzed for various chemical forms of nitrogen using the Smartchem200 automated discrete analyzer. The analysis and determination of nitrate nitrogen (${NO}_3^{-}$–N), nitrite nitrogen (${NO}_2^{-}$–N), ammonia nitrogen (${NH}_4^{+}$–N), and TN were all conducted using the spectrophotometric method, following the national standards for surface water environmental quality. Standard curves were established using gradient dilution methods, with the requirement that the coefficient of determination for each standard curve be greater than 0.995. Each sample was measured twice, and the average of the two measurements was used as the final concentration. The concentration of dissolved organic nitrogen (DON) was calculated by subtracting the aforementioned inorganic nitrogen components from the TN.

2.3. Geographical Detector

The geographical detector is an effective analytical tool for assessing spatial heterogeneity and the influence of various driving factors, providing valuable insights for understanding the influence of socioeconomic and natural environmental factors on spatial attributes [25,26]. The explanatory power of individual factors is quantified using the q statistic, calculated according to Equation (1). The q-value ranges from 0 to 1, indicating that the factor explains 100 × q% of the spatial heterogeneity of the attribute.

$$q=1-{\displaystyle \frac{{\sum }_{n=1}^L{N}_h{\sigma }_h^2}{N{\sigma }^2}}$$

(1)

where h, ranging from 1 to L, represents the strata of the factor. N_h and N denote the number of spatial units in the stratum and the entire area, respectively. σ_h² and σ² are the variances of the spatial attribute in stratum and the entire area, respectively.

The intensity of land use serves as an indicator of the degree of human exploitation and utilization of land resources. In this study, land use types were classified into four hierarchical levels: unused land (Level 1), woodland, grassland, and water area (Level 2), arable land (Level 3), and construction land (Level 4). This classification framework was employed to quantify the overall intensity of land use across the study area (Equation (2)) [27].

$$intensity of land use=100\times \sum _1^n{A}_i·S_i$$

(2)

where A_i represents the level of grade i land use intensity classification, and S_i represents the proportion of grade i land use type area.

Nighttime light can reflect human activity footprints and is characterized by high spatiotemporal resolution, making them widely used in studies of human impacts. For example, Huang et al. [7] found that in China’s Liao River, Hai River, Yellow River, Yangtze River, Huai River, and Pearl River, nighttime light had contribution values of 0.73 and 0.62 to the variation of total phosphorus and ammonia nitrogen, respectively, making it the most powerful predictor among the factors studied.

In this study, six driving factors for the year 2020 were selected, including the intensity of land use, annual precipitation, nighttime light, average elevation, population density, and annual mean temperature. Employing the ArcGIS, each factor was delineated in the five distributional strata using natural breaks [28,29]. This approach enabled a comprehensive analysis of how these factors influenced the TN load intensity (the nitrogen load per unit area) within the Hanjiang River subbasins for the year 2020.

2.4. Downscaling

Given the relatively low spatial resolution of GCM, the application of downscaling methodologies is imperative for the correction and optimization of rational climatic information to a local scale [30,31]. The delta bias correction technique is a straightforward and widely employed method for downscaling, as indicated by Equations (3) and (4) [30,32]. This method is favored for its simplicity and effectiveness in adjusting the bias present in large-scale model outputs, thereby aligning them more closely with observed data at local scales [33]. Additionally, the Taylor diagram, a graphical representation, serves to effectively illustrate the differences in the performance metrics of various models, including the correlation coefficient (r) and the normalized root mean square error (NRMSE) Equation (5). This study adopts the period from 2000 to 2013 as the baseline historical period for the delta downscaling correction of daily precipitation and the temperatures of maximum and minimum at 25 individual stations. Subsequently, an evaluation of the performance suitability of the GCMs, after downscaling, is conducted.

$$T_d^{*}=T_{d,GCM}+\left(T_{mi,obs}-T_{mi,GCM}\right)$$

(3)

$$P_d^{*}=P_{d,GCM}·\left({\displaystyle \frac{P_{mi,obs}}{P_{mi,GCM}}}\right)$$

(4)

where T_d^* and P_d^* denote the daily temperature and precipitation values post the application of delta bias correction, respectively. The variables T_d,GCM and P_d,GCM correspond to the daily temperature and precipitation as projected by the GCM. T_mi,GCM and P_mi,GCM signify the monthly mean temperature and precipitation, respectively, as simulated by the GCM for month i within the historical periods. The observed counterparts, T_mi,obs and P_mi,obs, represent the monthly mean temperature and precipitation recorded at the meteorological station for month i during the same historical timeframe.

$$NRMSE={\displaystyle \frac{\sqrt{\frac{{\sum }_{i=1}^n{\left(S_i-O_i\right)}^2}{n}}}{\overline{O}}}$$

(5)

where n represents the sample size. S_i and O_i stand for simulated and observed values, respectively. $\overline{O}$ denotes the mean values of the observed data.

2.5. SWAT Model

SWAT, an advanced semi-distributed hydrological model developed by the United States Department of Agriculture, is based on the principles of physical hydrological processes and temporal continuity [34,35]. It is highly regarded for its effectiveness in assessing runoff, sediment transport, and nutrient load in large and complex watersheds. The SWAT model delineates the watershed into several subbasins based on elevation, which are subsequently stratified into hydrological response units (HRU) according to land use, soil classification, and slope characteristics. By simulating the hydrological processes and nutrient cycles of each HRU in detail, the model integrates these assessments to provide a comprehensive evaluation of runoff and pollution conditions across the entire basin. In this study, the SWAT model (SWAT2012 rev. 692) was utilized to divide the Hanjiang River basin into 62 subbasins, including 1239 HRUs, facilitating an extensive analysis of the runoff and nitrogen pollution. Additionally, scenario analysis was conducted to simulate conditions devoid of specific nitrogen inputs, thereby quantifying the contributions of various nitrogen sources to TN discharge. These sources encompass livestock farming, rural domestic, fertilizer application, atmospheric deposition, point sources, and soil nitrogen pools.

SWAT-CUP, an effective automated calibration program, is specifically designed to adjust the parameters of the SWAT model within recommended ranges, enhancing its accuracy in reflecting actual hydrological and water quality conditions [35]. This study calibrated and validated the model using monthly observed data from four hydrological stations and six water quality monitoring stations to ensure a high degree of consistency between the simulated outcomes and the observed hydrological fluxes and TN emissions within the Hanjiang River basin. The accuracy of the model is assessed by two key indicators [34] (Equations (6) and (7)): the coefficient of determination (R²) and the Nash efficiency coefficient (NSE). This study adheres to the principle of first calibrating runoff and then calibrating pollutants. It repeatedly simulates the given parameters and ranges to obtain the optimal results’ statistics and recommended ranges for the next parameter adjustment, iterating until the evaluation parameters meet the expected criteria.

$$R^2={\left[{\displaystyle \frac{{\sum }_{i=1}^n\left(O_i-O_{ave}\right)·\left(S_i-S_{ave}\right)}{\sqrt{{\sum }_{i=1}^n{\left(O_i-O_{ave}\right)}^2}·\sqrt{{\sum }_{i=1}^n{\left(S_i-S_{ave}\right)}^2}}}\right]}^2$$

(6)

$$NSE=1-{\displaystyle \frac{{\sum }_{i=1}^n{\left(O_i-S_i\right)}^2}{{\sum }_{i=1}^n{\left(O_i-O_{ave}\right)}^2}}$$

(7)

where n represents the total number of observations. O_i and S_i denote the observed and simulated values, respectively. O_ave and S_ave represent the mean values of the observed and simulated data, respectively.

Non-point source pollution is the result of the combined effects of numerous anthropogenic and natural factors. This study primarily considers the soil nitrogen pool, rural domestic, atmospheric deposition, fertilizer application, livestock farming, and large-scale point sources. By reducing specific nitrogen sources, the contributions of various nitrogen sources are analyzed (Equations (8) and (9)) [36].

$${TN}_i={TN}_{all}-{TN}_{all,i}$$

(8)

$${TN}_{soil}={TN}_{all}-\sum {TN}_{all,i}$$

(9)

where TN_i represents the TN quantities originating from nitrogen source i. TN_all and TN_all,i represent the total and lack of specific nitrogen source i, respectively, and i is rural domestic, atmospheric deposition, fertilizer application, livestock farming, and large-scale point source.

3. Results and Discussion

3.1. The Performance of SWAT Model

When employing the SWAT model, it is standard practice to perform calibration and validation [37,38]. To enhance predictive accuracy, this study selected 13 runoff and 15 nitrogen cycle parameters as key variables for model correction (Table S1). Precipitation data from Hanzhong, Yangxian, Ankang, and Baihe stations revealed a notable increase in rainfall during the autumn months (Figure S1). The precipitation trends in Hanzhong and Yangxian were similar, as were those in Ankang and Baihe. Over the period 2005 to 2022, Hanzhong station recorded an average annual precipitation of 1011.06 mm, with above-average rainfall in 2010, 2011, 2013, 2014, and 2021 (Figure S1a). Yangxian station had an average of 864.73 mm, with above-average years in 2011, 2017, 2019, and 2021 (Figure S1b). Ankang station reported an average of 792.37 mm, with above-average precipitation in 2005, 2007, 2008, 2010, 2019, and 2021 (Figure S1c). Baihe station’s average was 790.01 mm, with above-average years in 2005, 2009, 2010, 2011, 2017, and 2021 (Figure S1d). The peak precipitation across all stations occurred in 2021, with a decreasing trend in average precipitation observed from the upper reaches to the lower reaches of the Hanjiang River.

Runoff and precipitation exhibited similar variation trends, with peak runoff observed in 2021 at all stations (Figure 2). Analysis of simulated data for Hanzhong (Figure 2a), Yangxian (Figure 2b), Ankang (Figure 2c), and Baihe (Figure 2d) hydrological stations indicated high model accuracy during both calibration and validation periods, with R² values greater than 0.75 and NSE values exceeding 0.7. This aligns with the superior performance of the SWAT model in simulating runoff across various global basins, such as the Ganges basin [39] (R² > 0.85, NSE > 0.75), the Mississippi Big Sunflower River basin [40] (R²: 0.61–0.70, NSE: 0.55–0.63), and the Yellow River basin [41] (R²: 0.65–0.73, NSE: 0.5–0.78). Moriasi et al. [42] suggest that satisfactory SWAT model performance is indicated when R² exceeds 0.6 and NSE surpasses 0.5 in runoff simulations. Therefore, the calibrated SWAT model effectively captured runoff dynamics in the Hanjiang River basin.

Employing the SWAT model to simulate the TN pollution in the Hanjiang River basin, the accuracy of the simulation results was as excellent as that of the runoff simulation (Figure 3). The simulation outcomes at six water quality monitoring stations (Chi River, Tian River, Jia River, Yue River, Ren River, and Lan River) demonstrated R² values exceeding 0.65 and NSE values greater than 0.5, indicating a strong consistency between the simulated and observed TN values. Regarding nitrogen pollution, Moriasi et al. [42] suggest that satisfactory model performance is indicated when R² and NSE for the simulated versus observed data reach 0.3 and 0.35, respectively; values above 0.6 denote good performance, while those above 0.7 signify excellent performance. Consequently, the calibrated SWAT model accurately simulated both the hydrological cycle and nitrogen pollution in the Hanjiang River basin, proving to be a reliable tool for simulating and predicting hydrological processes and TN emissions in the basin.

3.2. Characteristics of TN Emissions

Using the SWAT model, we analyzed interannual and monthly TN emissions in the Hanjiang River basin. The results revealed that the TN concentration in the Hanjiang River exhibited two gentle “V” shaped trends over the 16-year period from 2005 to 2020 (Figure 4a). The first “V” shape occurred from 2005 to 2012, with the lowest point recorded in 2007 at an average concentration of 1.93 mg/L. The second “V” shape, spanning 2013 to 2020, reached its minimum in 2017, with an average concentration of 1.67 mg/L. Overall, the average TN concentration in the Hanjiang River was 2.16 mg/L during the 16 years, with a peak concentration of 2.56 mg/L recorded in 2013.

The Hanjiang River basin exhibited relatively mild nitrogen pollution, with TN concentrations lower than those of many rivers worldwide (

Table 1. Comparison of TN concentration.

Study Area	Study Period	Description	TN Concentration(mg/L)
This research	2005~2020	The basin area is approximately 74,000 km2, with a vegetation cover of 76%	2.161 ± 1.087
Amazon basin [46]	2011~2013	Basin area exceeds 6 million km2, mainly consisting of tropical forests, snow-capped mountains, and savannas, with an annual precipitation of about 2000 mm.	0.363 ± 0.085
Mississippi River, United States [47]	2013~2017	An important agricultural area, the basin area is about 278,000 km², with an average annual precipitation of 950 mm.	3.438 ± 2.299
Lobo Stream, Brazil [48]	2018	The watershed, spanning 220 km2, receives about 1500 mm of annual rainfall and is a water source conservation area.	0.55~1.45
Sąpólna River, Poland [44]	2021–2022	87.7 km2, a biological reserve with an annual precipitation of about 625 mm.	3.194
Athabasca River, Canada [45]	1983~2013	160,000 km2, dominated by forests, with an annual precipitation of about 510 mm.	0.502 ± 0.228
Nanfei River, Tangxi River, and Pai River in China [49]	2014~2017	Subtropical monsoon climate, with an annual average rainfall of 1003.4 mm, 70% of the land is used for agriculture.	9.137 ± 3.550
Zijiang River, China [50]	2020~2021	28,000 km2, an important rice-growing area with an average annual precipitation of 1200 to 1800 mm.	1.604~3.574
Anjiagou watershed, China [51]	2006~2014	Agricultural basin	4.17 ± 1.027
Downstream of the Yellow River, China [43]	2018	North China Plain agricultural irrigation area.	7.889 ± 0.795

). For instance, the TN concentration in the lower reaches of the Yellow River in China exceeded 7 mg/L [43], while the average TN concentration of the Sąpólna River in Poland was 3.194 mg/L [44]. In contrast, the TN concentration of the Athabasca River in Canada with more green spaces was notably lower, at less than 1 mg/L [45]. Thus, while the nitrogen pollution in the Hanjiang River basin is comparatively low, further control measures are necessary.

Monthly TN concentrations exhibited an upward trend from January to March, followed by a decline from April to August and a subsequent rise from September to December (Figure 4b). Overall, TN concentrations were higher during the dry season compared to the wet season. Notably, TN peaked at 2.929 ± 0.672 mg/L in March during the dry season, while the minimum concentration of 1.669 ± 0.244 mg/L was recorded in August during the wet season. It is worth noting that there was a peak of TN concentration in June, up to 2.960 ± 1.201 mg/L, likely linked to the intensive application of base fertilizers for autumn crops, such as maize and rice. In accordance with the fertilization guidelines from the Ministry of Agriculture and Rural Affairs of the People’s Republic of China, rice cultivation areas are recommended to apply about 150 kg N/ha, with the rice transplanting typically scheduled for the end of May. Meanwhile, maize cultivation areas are advised to apply about 180 kg N/ha, and the sowing period is mainly concentrated in early June. During this time, the water demand for vegetation growth within the river basin increases, while the enhanced precipitation intensity further promotes nitrogen leaching, thereby leading to a rise in the TN concentration.

Further analysis of regional differences in TN concentration within the Hanjiang River basin revealed that upstream regions experienced significantly lower TN pollution than downstream regions (Figure 4c, p < 0.05), with concentrations approximately 64% lower. When the accumulation of pollutants exceeds the self-purification capacity of a river, the concentration of pollutants in the river tends to increase. The Hanjiang River originates from the Qinling Mountains, which are characterized by good forest and grassland cover. As the altitude decreases, the terrain gradually becomes gentler. Along the banks of the Hanjiang River, human activities have become more concentrated, leading to the formation of urban areas such as Ankang and Shiyan. These urban areas have increased nitrogen inputs from sources such as fertilizers, livestock manure, domestic sewage, and industrial wastewater. Consequently, the TN concentration in the downstream regions may be relatively higher. Among tributaries along the north and south banks of the Hanjiang River, TN concentration on the north bank was only 8.59% higher than those on the south bank, showing no significant difference. Moreover, TN concentration in the main stream of the Hanjiang River was 61% higher than in the tributaries (p < 0.05). This was associated with the concentration of farmland and urban areas along the gently sloping banks of the main stream.

Further research into diverse nitrogen emission scenarios and a quantitative assessment of the contributions of various pollution sources to TN pollution in the Hanjiang River basin revealed that the soil nitrogen pool was the primary source, accounting for 92.78% of the total (Figure 4d). Other contributors included atmospheric deposition (4.38%), fertilizer application (2.53%), livestock farming (0.16%), rural living (0.15%), and point sources (0%). Chen et al. [52] researched the Wen-Rui Tang River basin in Zhejiang Province, China, and found that nitrogen emissions from the soil nitrogen pool ranked second only to those from urban wastewater. The proportion of construction land in the Hanjiang River basin is only about 0.4% [24], in stark contrast to about 40% in the Wen-Rui Tang River basin. This indicated that the contribution of urban wastewater to nitrogen emissions was relatively lower, thereby highlighting the significance of the soil nitrogen pool. Ye et al. [53] conducted an analysis of nitrogen runoff from the Han, Rong, and Lian rivers in Guangdong Province, China, which revealed that in rural areas, the soil nitrogen pool was the principal source of nitrogen emissions, accounting for approximately 68% of the total emissions. Moreover, the Hanjiang River basin received approximately 40% less annual rainfall (800–1100 mm) compared to the Han, Rong, and Lian Rivers basin (1670–1720 mm), reducing the risk of wet deposition of nitrogen and fertilizer leaching [54]. Consequently, this situation increased the contribution of nitrogen emissions from the soil nitrogen pool.

The excessive and irrational application of chemical fertilizers in agricultural production led to a substantial surplus of nitrogen in the soil, which has been persistently degrading water quality [55]. Liu et al. [56] found that between 1900 and 2015, soil nitrogen surpluses in the Yangtze and Pearl River basins exhibited an increasing trend, which in turn drove an upward trend in groundwater nitrogen loads. The accumulation of excess nitrogen has made it extremely challenging to achieve short-term improvements in water quality. Meter et al. [55] found that it could take nearly 35 years to mitigate the biogeochemical impacts of 142 Tg of residual soil nitrogen in the Mississippi River Basin. Despite China’s commitment to achieving zero net growth in chemical fertilizer use by 2020 and optimizing wastewater treatment facilities, the residual nitrogen issue may have become a significant source of water pollution [57,58]. Thus, the key to controlling nitrogen pollution in the Hanjiang River Basin lies in the scientific management of the soil nitrogen pool to mitigate the effects of nitrogen legacy.

3.3. Distribution of Nitrogen in Different Forms

Nitrogen typically exists in various chemical forms. The inorganic nitrogen composition was similar to that of Indonesia’s Batang Arau River, Mexico’s El Fuerte River, and Turkey’s Kucuk Menderes River (

Table 2. Comparison of inorganic nitrogen concentration.

Study Area	Study Period	${NH}_4^{+}$–N	${NO}_2^{\mathit{-}}$–N	${NO}_3^{\mathit{-}}$–N	Description
Indonesia’s Batang Arau River [60]	2014	0.182–0.510	0.0–0.148	0.739–1.942	172 km2, the upstream area is dominated by forest cover, while the downstream area is densely populated with cities.
Mexico’s El Fuerte River [45]	2017	0.012–0.34	0–0.012	0.002–0.875	36,000 km2, the upstream land is mostly forested, the downstream area is agricultural, and the annual precipitation ranges from 311 to 1200 mm.
Turkey’s Küçük Menderes River [61]	2017~2018	0.43	0.09	1.73	7 km2, mainly characterized by wetlands.
This study	2023~2024	0.21	0.02	1.66	74,000 km², with about 76% covered by forests and grasslands and about 22.7% by arable land.

). This research analyzed the content of nitrogen in various chemical forms within the Hanjiang River basin during both wet and dry seasons, finding average concentrations of TN, ${NO}_3^{-}$–N, ${NH}_4^{+}$–N, ${NO}_2^{-}$–N, and DON (Figure 5) to be 3.01 mg/L, 1.66 mg/L, 0.21 mg/L, 0.02 mg/L, and 1.11 mg/L, respectively. Due to the use of sulfuric acid for acidification and preservation during the experimental design, nitrite nitrogen was converted to nitrate nitrogen, resulting in an underestimation of nitrite nitrogen concentrations [59].

Compared to the same period in 2021 to 2022, the concentrations changed by +6.4%, −17.1%, +157%, −77.4%, and +67.8%, respectively (Figure S2). Additionally, ${NO}_3^{-}$–N was the predominant form of nitrogen in the runoff, comprising approximately 55.23% of the TN content, consistent with findings from many river systems, such as the Nala watershed of the Pearl River basin [62] and the East Tiaoxi River basin [44], where ${NO}_3^{-}$–N constituted 80% and 47% of the TN respectively. The variation of nitrogen concentrations in different chemical forms between the north and south banks of the Hanjiang River was generally not significant, except for ${NO}_2^{–}$–N (Figure 5g), which was significantly higher on the north bank at 0.018 ± 0.013 mg/L compared to 0.012 ± 0.006 mg/L on the south bank. Furthermore, the concentrations of TN and ${NO}_3^{-}$–N (Figure 5a,c) on the north bank of the Hanjiang River were slightly elevated compared to the south bank, by 6.8% and 7.6%, respectively.

Nitrogen concentrations exhibit notable fluctuations between dry and wet seasons. For instance, Bu et al. [63] observed that ${NH}_4^{+}$–N levels were significantly elevated during the wet season in the Taizi River basin, while TN, ${NO}_3^{-}$–N, and ${NO}_2^{–}$–N were higher during the dry season. Shi’s sampling and analysis in the Dan River basin [64] revealed that both ${NH}_4^{+}$–N and ${NO}_3^{-}$–N levels were higher during the wet season. However, this study identified that the concentrations of TN, ${NO}_3^{-}$–N, ${NH}_4^{+}$–N, and ${NO}_2^{–}$–N during the wet season were significantly (p < 0.05) lower than those observed in the dry season, with reductions of 13%, 15%, 79%, and 49%, respectively. This discrepancy may be attributed to the stronger dilution effect of precipitation on nitrogen levels in the Hanjiang River basin, despite rainwater introducing nitrogen pollutants into the rivers. Moreover, the concentrations of DON were recorded at 1.179 ± 0.666 mg/L in the wet season and 1.046 ± 0.744 mg/L in the dry season, with no significant differences detected between the two periods. The concentration of organic nitrogen and inorganic nitrogen frequently exhibit a negative correlation, attributable to the dynamic transformation processes between these two forms of nitrogen. These transformations may include the assimilation of inorganic nitrogen by aquatic organisms to synthesize essential organic compounds and bacterial ammonification of DON in aquatic environments [65,66]. During the wet season, as temperatures rise and biological activities become more vigorous, the conversion of inorganic nitrogen into organic nitrogen may be enhanced. This process likely results in a less pronounced decrease in organic nitrogen concentration during the flood season despite the dilution effect.

Li et al. [13] conducted a study in 2008 that highlighted increased nitrogen pollution levels during the wet season in the Hanjiang River basin, attributing this phenomenon to heightened agricultural activities. According to FAO statistics, the nitrogen fertilizer application intensity in China was 213.92 kg N/ha in 2008, which decreased to 191.55 kg N/ha in 2022. In fact, since the peak nitrogen fertilizer application intensity of 235.38 kg N/ha in 2014, there has been a year-on-year decline in nitrogen fertilizer application intensity in China, reflecting the increasing stringency of China’s management of non-point source pollution. It is possible that after nearly 15 years of pollution management, non-point source agricultural pollution in the Hanjiang River has been relatively well controlled, resulting in lower nitrogen concentrations during the wet season. In addition, the precipitation in 2008 was about 861 mm, while the precipitation in 2022 was about 1098 mm, which may make the nitrogen load in the wet season more diluted.

3.4. Driving Factors of TN Spatial Heterogeneity

Both SWAT model simulation and field sampling analysis revealed notable spatial distribution differences in nitrogen emissions. Therefore, the relationship between spatial heterogeneity of TN and various factors such as climatic conditions (annual precipitation and average annual temperature), physical environment (average elevation), and human activities (night light, population density, and land use intensity) was further studied. Our findings indicated a significant linear correlation between TN emission intensity and five of these factors, except for land use intensity, in the Hanjiang River basin (Figure 6, p < 0.05). Among them, the correlation coefficient between TN emission intensity and annual precipitation was the highest, reaching 0.53, indicating a significant positive correlation. Both increased precipitation and rainfall intensity enhance water erosion [67], which accelerates nitrogen migration [68], contributing to a strong linear relationship with TN emission intensity. Additionally, there was a significant correlation between TN emission intensity and average elevation, with a correlation coefficient of 0.33. This suggests that the higher altitude areas, which are predominantly mountainous and characterized by steeper slopes, facilitate the migration of nitrogen pollutants, leading to increased TN emission intensity. Furthermore, since land use intensity, night light, and population density all reflect the influence of human activities, there was a very significant (p < 0.001) linear correlation among these factors.

Geographical detectors are commonly employed to assess the influence of various factors on pollution. In the Three Gorges Reservoir area, pig breeding volume and rural population significantly affected the intensity of TN emissions [69]. In the Wu River basin, anthropogenic factors such as cultivated land, livestock, fertilization, and population were the primary drivers of TN spatial variation [70]. This study quantitatively evaluated the explanatory power of these driving factors on the spatial heterogeneity of TN emission intensity, finding that annual precipitation had the greatest impact (61.4%,

Table 3. The driving factors of spatial heterogeneity about TN emission intensity.

Factor	Explanatory Power (q-Value)	Significance (p-Value)
Precipitation	0.614	<0.001
Temperature	0.301	0.071
DEM	0.297	0.003
Nighttime lights	0.230	0.005
Population density	0.205	0.100
Land use intensity	0.142	0.096

), followed by average annual temperature (30.1%), average elevation (29.7%), night lights (23.0%), population density (20.5%), and land use intensity (14.2%). Notably, annual precipitation, average elevation, and night lights were statistically significant (p < 0.05). The sources of TN likely contribute to these findings: the Three Gorges Reservoir area was primarily affected by rural living, livestock and poultry, and fertilizer application [69], while approximately 45% of TN emission in the Wu River basin originated from fertilizer use [70]. In contrast, 90% of nitrogen emissions in the Hanjiang River basin came from the natural soil nitrogen pool, with spatial variation in TN emissions being more significantly influenced by climatic and natural factors. In summary, climatic factors, particularly precipitation, were the key determinants of spatial heterogeneity in TN distribution in the Hanjiang River basin.

3.5. Sustainability Change Under Climate Change

Climate change poses a significant global challenge, affecting sustainable development across various sectors, including agriculture, ecological environments, and hydrological cycles. Numerical simulation methods are frequently employed to assess the effects of climate change on water resources within river basins. In the Ganges River basin in India [71], CanESM5 exhibited the highest correlation of 0.169, and BCC-CSM2-MR showed the lowest NRMSE of 0.301, indicating their relative efficacy in simulating precipitation patterns in the region. Similarly, Jia et al. [72] evaluated the applicability of eight climate models in the source region of the Yellow River, finding that both temperature and precipitation correlations exceeded 0.8, reflecting satisfactory simulation accuracy. To investigate the trends in runoff and nitrogen emission changes in the Hanjiang River basin influenced by climatic factors, the latest CMIP6 climate models were employed. This study identified that the NorESM2-MM model demonstrated the highest correlation coefficient (0.70) and the lowest NRMSE (0.80) in precipitation simulation, indicating a robust representation of precipitation conditions in the Hanjiang River basin (Figure S3a). Furthermore, all climate models demonstrated high correlation coefficients (0.92–0.95) and low NRMSE (0.25–0.39) for simulating maximum and minimum temperatures, showcasing their strong capability in modeling temperature conditions in the basin (Figure S3b,c). Overall, the NorESM2-MM climate model displayed superior applicability for predicting future climate changes in the Hanjiang River basin.

By 2050, many water sources are projected to exhibit a similar trend of increasing runoff. Due to the impact of glacier melting and precipitation changes, the total runoff in the source region of the Yangtze River is expected to increase at an annual rate of 4.2 m³/s [73]. The annual runoff in the Changbai Mountain basin is projected to increase by 1.5 to 1.7 m³/s [74]. Additionally, Ethiopia’s Lake Tana basin may experience an unprecedented runoff increase of up to 220% [75]. After integrating four distinct climate scenarios, it was predicted that the average annual runoff in the Hanjiang River basin would increase by 6.19 m³/s (Figure 7a), indicating that the sustainability of water resources in the region would be improved in the future and water resources supply would be more abundant.

In contrast to the anticipated runoff trends, TN concentration in the Hanjiang River basin exhibited a slight downward trend under global climate change (Figure 7b), with an average annual decrease of approximately 0.004 mg/L. Mann–Kendall trend test (Figure S4) found that the UF statistics (p = 0.12) were predominantly below zero, except for the period from 2012 to 2014, indicating a decreasing trend of TN concentration throughout the study period. Based on the SSP126, SSP245, SSP370, and SSP585 climate scenarios, the average TN concentrations in the basin from 2021 to 2050 were predicted to be 2.045 ± 1.275 mg/L, 2.018 ± 1.198 mg/L, 1.995 ± 1.237 mg/L and 1.881 ± 1.044 mg/L, respectively (Figure 8). These projections were lower than the historical average of 2.161 ± 1.087 mg/L from 2005 to 2020, suggesting that TN levels in water quality are expected to remain favorable across various climate scenarios. Similarly, Mack et al. [76] found that under SSP245 and SSP585 scenarios, total nitrogen concentrations in European surface waters are projected to decrease by 2060 but emphasize that more aggressive measures are needed to alleviate the impacts of human activities. Overall, under future climate change projections, nitrogen sustainability in the Hanjiang River basin is anticipated to improve. Moreover, since 2014, the intensity of nitrogen fertilizer use in China has been declining year by year. China has been making unremitting efforts to reduce fertilizer application, steadily promote precision agriculture, and continue to follow the path of green development. People can control agricultural non-point source pollution through active policies and measures, which will certainly help to improve the quality of the water environment. Thus, the future water storage and quality in the Han River basin are relatively optimistic.

This study also has several limitations. Specifically, the nitrogen legacy issue in the Hanjiang River basin necessitates a long-term remediation process. Future research could further explore appropriate agricultural management practices to improve soil nitrogen legacy. Additionally, this study did not account for the impact of human activities on water resources in the Hanjiang River basin. Future research should consider the effects of human policies, such as the reduction of fertilizer application, on the control of agricultural non-point source pollution and water environment management.

4. Conclusions

In summary, the SWAT simulation analysis of the Hanjiang River basin indicated an average TN concentration of 2.16 mg/L from 2005 to 2020. Notably, pollution levels were lower during the dry season compared to the wet season, with slightly higher concentrations observed on the north bank than on the south bank. Further source analysis highlighted the significance of the soil nitrogen pool in nitrogen emissions, accounting for 92.78% of the total emissions. Meanwhile, sampling data indicated that during the wet season, the concentration of inorganic nitrogen was relatively lower, likely due to the dilution effect of increased rainfall. Under global climate change, precipitation was forecasted to significantly influence nitrogen emissions in the Hanjiang River basin. Projections suggested that between 2020 and 2050, the annual average runoff in the Hanjiang River would increase by 6.19 m³/s annually, while the average TN concentration was anticipated to decrease by 0.004 mg/L annually. Overall, this trend indicated that the Hanjiang River basin would provide more abundant and cleaner water resources in the future, supporting enhanced sustainability. Implementing precision agriculture practices can help reduce nitrogen and phosphorus emissions into the Hanjiang River associated with precipitation events. By prioritizing these strategies, we may effectively mitigate pollution and promote the long-term health of the Hanjiang River ecosystem, ultimately benefiting both the environment and local society. In summary, ongoing efforts to improve the sustainability of water resources will be vital in ensuring water security for the sustainable economic and social development of the Beijing–Tianjin–Hebei region.