Identifying priority control areas and proposing countermeasures for nitrogen and phosphorus pollution in the Jinjiang River Basin: An integrated remote sensing and hydrological analog approach

1 Introduction

Nitrogen (N) and phosphorus (P) overloading drives widespread aquatic eutrophication worldwide (NRC; National Research Council, 2000; Dimitriou and Moussoulis, 2010; Röman et al., 2018; Chu et al., 2021; Van Heyst et al., 2022; Canning and Death, 2023). River systems integrate land-use nutrient losses and reflect anthropogenic water quality degradation (Finnian et al., 2024; Ganie et al., 2024; Wang et al., 2024; Zhao et al., 2026), with pollutant inputs split into point sources (PS, domestic/industrial effluents) and non-point sources (NPS, agricultural fertilizer runoff, livestock waste, urban stormwater). Quantifying their respective contributions is prerequisite for targeted watershed pollution management (Liu et al., 2017; Li H. L. et al., 2023).

The Jinjiang River Basin (JJRB) is a vital drinking water source in Fujian, supplying nearly 10 million residents in Quanzhou and Kinmen (Zhou et al., 2012; Liu et al., 2022). Long-term monitoring reveals severe N-P degradation: the share of Class I–II water among provincial/national monitoring sections fell from 76.9% (2015) to 46.2% (2020), mainly driven by elevated total phosphorus (TP) (Quanzhou Ecological Environment Bureau (QEEB), 2021). Characterized by mixed agricultural–urban land cover, highly variable seasonal rainfall and extensive ungauged subcatchments, the basin faces major barriers to accurate nutrient load apportionment (Huang et al., 2026; Wang et al., 2025; Chen et al., 2025). Existing literature confirms rainfall-runoff dominates N-P export in mixed land-use watersheds, yet conventional physically based models (SWAT, HSPF) require dense hydro-meteorological and high-resolution land surface datasets; data scarcity in JJRB induces large parameter uncertainty and biased load estimations, further exacerbated by fragmented urban–rural underlying surfaces.

To overcome quantification limitations in data-sparse catchments (Angalika et al., 2022), the hydrological analogy method—using monitored station rainfall and catchment area weighting to extrapolate runoff for ungauged reaches—was adopted here to offset missing discharge data and reduce simulation bias, suitable for medium-small sub-watersheds with stable local meteorology and land cover.

Prior JJRB studies mostly focused on isolated river segments and reservoirs (Dongxi River, Shanmei Reservoir), limited to water quality trend assessment. One basin-wide load evaluation relied on only 15 sampling sections, lacking full subcatchment coverage and spatial source differentiation analysis (Rong et al., 2009; Chen et al., 2012; Ma et al., 2015; Zhang et al., 2018; Li et al., 2021). No comprehensive quantification of point/non-point source contributions and their spatial drivers exist to support spatially precise pollution control, forming a critical research gap.

This study integrates remote sensing land use data, field sampling, laboratory nutrient analysis, and hydrological analogy methods to systematically assess the nitrogen and phosphorus pollution situation in the entire JJRB area, and focuses on three core objectives: (1) Quantifying the pollution levels of each sub-basin; (2) Quantify the contribution of point sources and non-point sources to nutrient loadings in terms of temporal and spatial scales; (3) Identify key control areas and major pollution sources within the watershed.

Results will reveal spatial differentiation mechanisms of N-P pollution, deliver scientific support for refined differentiated watershed governance, and protect regional drinking water security and aquatic ecosystem integrity.

2 Overview of the study area

The JJRB (24°31′-25°32′ N, 117°44′-118°47′ E) lies on Fujian's southeast coast, bounded by the Minjiang Basin to the north and Jiulongjiang Basin to the southwest. Terrain transitions from northwest hills to southeast coastal plains under a subtropical maritime monsoon climate, with annual precipitation of 1,000–1,800 mm strongly partitioned seasonally: 80% falls in the wet season (March–September), and 20% in the dry season (October–February). As Fujian's fourth-longest river, the Jinjiang main channel extends 182 km over a 5,629 km² catchment. Its two headwaters—the Dongxi (Yongchun County) and Xixi (Anxi County)—converge at Shuangxikou (Nan'an) before discharging into Quanzhou Bay. The dendritic fan-shaped river network contains 16 tributaries >100 km², including Lanxi, Taoxi, and Jiushijiuxi (Huang and Cheng, 2004; Chen et al., 2012).

In the JJRB, the dominant soil types include red soil, yellow-red soil, yellow soil, and paddy soil. According to the 2020 land-use classification, forests account for 45%, cultivated land for 29%, grassland for 13%, and built-up (urban) areas for 12% of the total land area (Figure 1). Cultivated land is further categorized into paddy fields, hillside orchards and tree plantations, and coastal vegetable plots.

The long-term monitoring at the Shilong Hydrological Station on the Jinjiang River (from 2016 to 2025; Quanzhou Ecological Environment Bureau (QEEB), 2025) revealed that the risk of eutrophication persisted. Mann-Kendall tests show a mild, non-significant rise in annual mean TP (0.10 mg/L in 2016 to 0.11 mg/L in 2025), while ammonia nitrogen (NH₃-N) decreased significantly (p < 0.01) from 0.43 mg/L to 0.21 mg/L (Figure 2).

Non-point agricultural sources and urban domestic wastewater constitute the primary contributors to nitrogen (N) and phosphorus (P) loading in the river, whereas industrial point-source discharges exert a comparatively minor influence. Nutrient loss intensity varies substantially across regions due to disparities in fertilization practices and irrigation regimes among farmland types. Paddy fields operate under a double-rice cropping system, with annual N fertilizer application rates ranging from 225 to 280 kg/ha and P fertilizer (as P₂O5) application rates of 75–105 kg/ha. Slope-based orchards and tree plantations—predominantly perennial cash crops such as longan, lychee, and tea—are characterized by high nutrient inputs and elevated susceptibility to surface runoff-driven losses. Coastal vegetable production systems employ intensive multi-cropping rotations and exhibit the highest fertilizer application rates among all land uses: N at 450–600 kg/ ha and P₂O5 at 180–240 kg/ha—resulting in pronounced nitrate leaching, particularly in sandy soils (Quanzhou Bureau of Agriculture and Rural Affairs, 2022).

3 Data sources and methods

3.1 Data sources

The dataset employed in this study comprises a 30-meter-resolution Digital Elevation Model (DEM) acquired in 2019 from the Geospatial Data Cloud (http://www.gscloud.cn). The data download date was June 12, 2024. The baseline land use and land cover (LULC) data used in this study is the 2020 layer of the global 10 m annual Sentinel-2 land cover product (Esri & Impact Observatory, 2022). This open dataset is hosted on ArcGIS Living Atlas of the World, derived from Sentinel-2 imagery via deep learning methods with nine land cover classes and available under the CC BY 4.0 license (https://livingatlas.arcgis.com/landcoverexplorer/). For local accuracy optimization, we referenced the China Multi-period Land Use Land Cover Dataset (CNLUCC, Xu et al., 2026) and implemented manual corrections combined with field investigations. The first-level classification includes five types: cropland, urban land, grassland, forest land and water bodies. The data was downloaded on April 3, 2026. Rainfall-driven data originated from the China Meteorological Assimilation Drive Dataset (CMADS, https://www.tpdc.ac.cn/zh-hans/data/faae7605-a0f2-4d18-b28f-5cee413766a2). The complete time series of this dataset covers daily meteorological elements from 1979 to 2020. In this study, the rainfall sequence from 2000 to 2018 was selected for analysis. The data download acquisition date was May 28, 2024.

The flow rates at sampling points were estimated using the hydrological analogy method due to the spatial mismatch between discharge monitoring stations and water quality sampling sites. This method is applicable to the JJRB because of its relatively small catchment areas. Additionally, the meteorological conditions and underlying surface characteristics are similar among adjacent subcatchments. The estimation process included three steps: first, ArcGIS hydrological tools and 12.5 m DEM data were used to delineate and calculate the catchment area of each sampling section and hydrological station (Figure 3); second, the rainfall-area composite parameter for each catchment was computed; finally, flow rates of sampling sites were proportionally estimated based on the parameter ratio relative to reference hydrological stations. Rainfall data from June and October were used to represent wet-season and dry-season hydrological conditions, respectively.

Field sampling and laboratory tests were conducted to obtain total nitrogen (TN) and total phosphorus (TP) concentrations. Following the national surface water monitoring specification (HJ/T 91), 60 sampling sites were set across the Jinjiang River Basin, covering the Xixi River, Dongxi River, Jinjiang main channel, and their multi-grade tributaries. The sampling layout covered upstream mountainous areas, midstream agricultural and town zones, and downstream urban industrial areas, enabling comprehensive analysis of spatial water quality differences and pollution contributions from varied rivers and land use types. Specifically, 20 sites were distributed in the Xixi River basin, 35 in the Dongxi River basin, and 5 along the Jinjiang main channel.

Water samples were collected synchronously in June 2020 (wet season) and October 2020 (dry season) to capture seasonal variations in TN and TP pollution. Separate glass bottles (500 mL for TP, 25 mL for TN) were used for sample collection. All samples were acidified to pH 1–2 with concentrated sulfuric acid and tested within 7 days at ambient temperature to guarantee sample stability.

All samples were analyzed by ALS Chemex (Guangzhou) Co., Ltd. All chemical reagents were of analytical grade and supplied by Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). TN content was determined using the Water quality—Determination of total nitrogen—Alkaline potassium persulfate digestion UV spectrophotometric method (GB/T 11894-1989), with a method detection limit of 0.05 mg/L. TP content was determined using the Water quality—Determination of total phosphorus—Ammonium molybdate spectrophotometric method (GB/T 11893-1989), with a method detection limit of 0.01 mg/L. The detection process strictly followed standard operating procedures. Absorbance was measured using corresponding analytical instruments, and the concentrations of TN and TP in the samples were calculated based on standard curves, ensuring the accuracy of the analytical data. To ensure data quality, blank controls, parallel samples, and standard reference materials were simultaneously set up during the experiment. The relative deviation of parallel samples was controlled within 5%, and the recovery rate of standard reference materials was maintained between 95% and 105%. These measures effectively minimized detection errors and ensured that the data met the precision requirements for nitrogen and phosphorus pollution analysis in the basin and for the identification of priority control areas.

3.2 Methods

3.2.1 Delineation of catchment areas and monitoring design

Subcatchments were delineated through hydrological analysis in ArcGIS using a 12.5 m resolution digital elevation model (DEM). A total of 60 subcatchments were identified, with 20, 35, and 5 located in the Xixi River, Dongxi River, and Jinjiang River mainstem, respectively.At the outlet of each subcatchment, a monitoring section was established (Figure 3), where water samples were analyzed for TN and TP content in both the dry and wet seasons.

3.2.2 TN, TP quality assessment in JJRB

The water quality of the JJRB was evaluated against the Class III criteria (TN: 1.0 mg/L, TP: 0.2 mg/L) stipulated in China's “Environmental Quality Standards for Surface Water” (GB 3838-2002).

This standard was selected because it aligns with Chinese regulatory requirements and the specific objectives of this study. While international eutrophication thresholds for lakes are more stringent (TN: 0.2 mg/L, TP: 0.02 mg/L; UNEP, 2021), the application of the Chinese standard is consistent with the nation's focus on controlling nitrogen and phosphorus pollution. This focus is evident in key policies such as the 14th Five-Year Plan for Ecological Protection, which prioritize total emission control of these nutrients.

3.2.3 Calculation of TN and TP pollution loads

We calculated the TN and TP pollution loads for various river reaches using the hydrological analogy method, and the detailed steps are as follows:

(1) Pollutant flux of a cross-section.

The pollutant flux of a cross-section was calculated using the following Equation 1:

Wi=Qi × Ci(1)

Where:

W_*i*: The pollutant flux of cross-section i (g/s); Q_*i*: The flow of cross-section i (m³/s); C_*i*: The pollutant concentration of cross-section i (mg/L).

(2) Pollution load in the section or catchment area.

Monitoring cross-sections were categorized into two types: (i) A cross-section follows a complete subcatchment area, such as cross-sections 1 and 3 in Figure 4 The pollution load of such a cross-section is the pollutant flux of the subcatchment area; (ii) Two cross-sections are separated by a subcatchment interval. The pollution load between such two cross-sections is the pollutant flux difference between cross-sections downstream and upstream. In the case where tributaries flow into the subcatchment interval, the pollutant fluxes of the tributaries should be subtracted.

Such as, the equations for calculating pollution loads between cross-sections 1 and 2 and between 2 and 4 are W_{1 − 2} = W₂− W₁ and W_{2 − 4} = W₄ − W₂ − W₃, respectively, with the latter accounting for the inflow of tributary 3.

(3) Contribution rate of pollution loads

The contribution rate (%) of a reach or subcatchment area to the pollution load of the cross-section at the outlet of a river basin was determined using the following Equation 2:

Contribution rate= Wi-jWtotal × 100%(2)

Where:

W_{_*i*−j}: The pollution load between cross-sections i and j (g/s). W_total: The total pollution load at the outlet section of the river basin (g/s).

3.2.4 Analysis of pollution contribution rate of land use types

To quantify the contribution rates of different land use types to the TN or TP load in the JJRB, this study employed multiple linear regression. The principle is to use the statistical relationship between the pollutant load measured at the outlet of each sub-basin and the area percentages of various land use types within the watershed to estimate the unit area pollution “output intensity” for each land use type (Tu and Xia, 2008). The specific steps are as follows:

(1) Construction of the Multiple Linear Regression Model. The following model was established (Equation 3):

PL=C%×β1+F%×β2+G%×β3+U%×β4+W%×β5(3)

Where:

PL: Total annual pollutant load at the basin outlet (TN or TP load amount, kg/a). C%, F%, G%, U%, W%: The proportion of cultivated land, forest land, grassland, urban construction land and water bodies within the basin. β₁–β₅: Regression coefficients, denote pollution output intensity of each land use type[kg·(hm²·a)⁻¹]. The magnitude and sign of regression coefficients β₁–β₅ reflect the differential pollution influence range of each land use type. A larger β value indicates longer pollutant transport distance and wider effective pollution impact area within the catchment. By contrast, smaller β values correspond to limited pollution contribution, while negative β values indicate that the corresponding land use type can purify and intercept pollutants.

(2) Calculation of contribution amount and contribution rate.

Multiple linear regression analysis was performed using SPSS software to obtain the output coefficients (β₁ to β₅). Subsequently, the pollution contribution amount and contribution rate for each land use type were calculated according to Equations 4, 5, respectively:

CQi=Li×A×βi(4)

CRi=CQiPL×100%(5)

Where:

CQ_*i*: annual TN/TP pollution contribution load of the i-th land use type (kg/a); A: total area of the subbasin(hm²); . L_*i*: area fraction of the i-th land use type within the subbasin (ranging from 0 to 1); CR_*i*: contribution rate of land type “I” to TN or TP (%).

3.2.5 Risk assessment and Priority control areas identification

Based on the cumulative pollution load curve method (Li W. et al., 2023), we separately assessed TN and TP pollution risks to ensure comparable zoning. Pixel diffuse load intensities for TN and TP were sorted ascendingly to generate independent cumulative curves. Following Wei et al. (2017), unified cumulative load thresholds (20%, 40%, 60%, and 80%) were adopted for both pollutants to define five risk grades: extremely high-risk (80–100%), high-risk (60–80%), moderate-risk (40–60%), low-risk (20–40%), and very low-risk(0–20%). Extremely high-risk and high-risk zones were designated as priority control areas for targeted management. Figure 5 illustrates this zoning framework.

4 Results and analysis

4.1 Concentration of TN and TP

Rainfall dominates the seasonal differentiation of TN pollution in the Jinjiang River Basin (JJRB). Intense precipitation in the wet season generates abundant surface runoff, which washes residual nitrogen fertilizers from farmlands and transports substantial agricultural non-point source nitrogen into river channels, triggering dramatic fluctuations in TN concentrations. The TN concentration ranges from 0.22 to 5.87 mg/L in the wet season, with a far wider variation range than that in the dry season (0.35–4.82 mg/L). Nevertheless, massive river discharge in wet seasons creates strong dilution effects that offset pulsed nitrogen inputs from rainfall and reduce background nitrogen loads in water bodies, resulting in a lower average TN concentration of 2.20 mg/L in wet seasons relative to the dry-season mean of 2.42 mg/L.

Rainfall also regulates seasonal variations in TP pollution, yet phosphorus exhibits distinctly different hydrological responses compared with nitrogen. Urban domestic sewage constitutes the primary source of phosphorus loads across the basin. Although surface runoff in wet seasons carries a small amount of particulate phosphorus eroded from farmlands, strong dilution restricts phosphorus accumulation effectively: wet-season TP concentrations range from 0.02 to 0.58 mg/L with an average value of 0.21 mg/L. In the dry season, scarce precipitation sharply reduces river discharge and nearly eliminates water dilution capacity. Continuous direct discharge of domestic sewage, coupled with internal phosphorus release from river sediments under sluggish flow conditions, jointly exacerbates phosphorus pollution. Dry-season TP concentrations vary between 0.04 and 0.75 mg/L with a mean of 0.30 mg/L and an elevated exceedance rate of 71.7%, indicating markedly severer phosphorus pollution than in wet seasons.

In reference to the Class III water quality thresholds specified in GB3838-2002, nitrogen and phosphorus pollution is prevalent throughout the basin. TN pollution shows extensive over-standard conditions across the whole basin, while TP pollution deteriorates prominently in dry seasons. Spatially, sections E18–E20 along the Gaolong Stream are identified as hotspots of TN contamination. Agricultural discharges from surrounding farmlands are responsible for nitrogen exceedance at the upstream sites, whereas direct domestic sewage discharge accounts for elevated TN levels at the urban section E20.

Based on Figure 6, TN and TP in the JJRB display distinct spatiotemporal patterns. Temporally, TN concentrations in the Xixi River fluctuated markedly between seasons, in contrast to the relatively stable levels in the Dongxi River and the mainstream. Spatially, during the wet season, the maximum, median, and average TN concentrations were highest in the Dongxi River, followed by the mainstream and the Xixi River; however, these spatial differences became less distinct during the dry season.

In contrast, TP exhibited a different distribution. Its concentrations varied seasonally in the Xixi River and the mainstream but remained stable in the Dongxi River. Spatially, the wet season was characterized by the order: Xixi River > mainstream > Dongxi River, whereas the dry season followed the pattern: mainstream > Xixi River > Dongxi River.

4.2 TN/TP ratios

Minimum concentrations of TN and TP during both wet and dry seasons exceeded internationally accepted eutrophication thresholds (TN: 0.2 mg/L, TP: 0.02 mg/L) (Guo et al., 2021). Notably, the maximum recorded concentrations exceeded these thresholds by up to 28 times, indicating that nutrient levels are sufficiently high to support algal growth. The TN/TP ratios ranged from 1.2 to 52.5 (mean: 14.9) in the wet season and from 2.9 to 50.3 (mean: 11.9) in the dry season (Figure 7). Ratios between 10 and 30, observed in 48.3% (wet season) and 33.3% (dry season) of sections, suggest conditions conducive to algal growth and potential cyanobacterial blooms (Qing et al., 2015). In contrast, the higher prevalence of ratios below 10 (41.7% and 58.3% in wet and dry seasons, respectively) indicates widespread nitrogen (N) limitation. This finding aligns with the classic Redfield ratio (optimal N:P mass ratio ≈ 7.2:1), and the frequent occurrence of ratios below 10 strongly suggests that algal growth is N-limited. Consequently, any increase in exogenous nitrogen input could significantly stimulate algal proliferation, thereby elevating the eutrophication risk in these waters. This conclusion is consistent with the findings of Zhong et al. (2024), who similarly identified nutrient concentration as a key driver of phytoplankton dynamics in the JJRB.

4.3 Pollution loads and sources analysis

4.3.1 Spatial distribution of TN pollution loads

In the JJRB, the total TN load was 260.61 g/s during the wet season and 471.87 g/s during the dry season. The Xixi River was the largest contributor during the wet season (119.93 g/s, 46.02%), followed by the Dongxi River (75.39 g/s, 28.93%) and the mainstream (73.29 g/s, 28.12%). Spatially, the J3 subcatchment near the estuary was the most significant source in the wet season, with a load of 56.76 g/s (21.78%), while the W9 subcatchment in the lower Xixi River ranked second (31.33 g/s, 12.02%). In the dry season, the TN load of the Xixi River, Dongxi River and the mainstream was 187.10 g/s, 135.62 g/s and 149.15 g/s respectively, and their contribution rates to the total load were 39.65%, 28.74% and 31.61% respectively. The estuarine subcatchments J3 and J4 became the top contributors, with combined loads accounting for 30.97% of the basin total (Figure 8). Negative TN loads were observed at several cross-sections (e.g., W4, E6, E11, E32, and J1) during both seasons, indicating that the TN output from these areas was lower than the input. This net decrease suggests a reduction in external TN inputs or the removal of nitrogen within the river channel through processes such as sedimentation and denitrification.

4.3.2 Spatial distribution of TP pollution loads

The total TP load in the JJRB was 46.69 g/s during the wet season and 94.84 g/s during the dry season. In the wet season, the Xixi River, Dongxi River, and mainstem contributed 23.06 g/s (49.39%), 8.71 g/s (18.65%), and 14.92 g/s (31.96%), respectively. Spatially, the W9 subcatchment exhibited the highest contribution at 8.36 g/s (17.90%), followed by the J4 subcatchment at 6.23 g/s (13.35%). During the dry season, the Xixi River contributed 39.79 g/s (41.95%), the Dongxi River 27.89 g/s (29.41%), and the mainstem 27.16 g/s (28.64%) (Figure 9). The largest contributing subcatchment shifted to J2, with a load of 25.46 g/s (26.84%), while W9 ranked second at 12.13 g/s (12.79%). Negative TP loads were observed at several cross-sections (e.g., W4, E6, E11, E32, and J1) in both seasons. This phenomenon may be attributed to reduced external TP inputs or in-stream phosphorus removal processes such as sedimentation, biological uptake, and other retention mechanisms.

Throughout the JJRB, subcatchments with high TN and TP loads were predominantly concentrated in the lower reaches of the Xixi River and the mainstream of the Jinjiang River, while the Dongxi River exhibited relatively low TN and TP loads.

4.3.3 Analysis of pollution sources

Results from the multiple linear regression model indicate that cropland and urban land are significant contributors to both TN and TP loads (p < 0.01), with average contribution rates of 38.3% and 33.0%, respectively. Together, these two land use types account for over 70% of the total pollution load in the JJRB, confirming their status as the primary pollution sources. Grassland and water bodies exhibit moderate contributions at 13.5% and 9.5%, respectively, whereas forest land demonstrates the lowest contribution (5.6%), highlighting its function as a pollution sink.

An analysis by pollutant type reveals that TN is primarily driven by cropland, which consistently ranks as the leading contributor during both wet and dry seasons, underscoring the persistent impact of agricultural non-point source pollution (Figures 10a, b). In contrast, TP pollution is more strongly influenced by urban land, particularly during the wet season when its contribution increases significantly, indicating that urban surface runoff plays a critical role in phosphorus export (Figures 10c, d).

Seasonal dynamics further demonstrate that TN loads during the wet season are significantly higher than those observed in the dry season (p < 0.05), emphasizing the influence of rainfall-induced runoff on the mobilization of non-point source pollutants. The contribution of urban land to TP increases substantially during the wet season (p < 0.01), suggesting that stormwater scouring is a key mechanism for phosphorus transport. Conversely, the contribution of water bodies to TP rises significantly during the dry season (p < 0.05), implying that internal phosphorus release from sediments becomes a dominant pathway under low-flow conditions.

The above model-derived conclusions are further verified by field investigation of typical pollution hotspots in the basin. Sections E18, E19, and E20 along the Gaolong Stream (a tributary of the Dongxi River) are the basin-wide hotspots with the highest TN concentrations, where peak TN values were detected in both wet and dry seasons. Field tracing shows that the riparian zones of E18 and E19 are continuously covered by rice, vegetable, longan and banana croplands, and agricultural non-point sources dominate TN exceedance at these two sites. By contrast, Section E20 is located in an urban residential area; direct domestic sewage discharge not only elevates TN levels here, but also delivers continuous phosphorus loads, consistent with the model result that urban land serves as the primary driver of TP pollution.

To further verify the reliability of the source apportionment results, this study conducted cross-validation by incorporating the SWAT-based mechanism simulation results reported by Tian (2022) for the Jiulong River Basin in Fujian Province. Qualitatively, their study confirmed that agricultural fertilization and rural and urban domestic sewage discharge are the dominant sources of nitrate nitrogen and ammonium nitrogen, respectively, which is highly consistent with our findings that cropland and construction land serve as the primary pollution sources, while forestland presents a prominent pollution reduction effect. Quantitatively, the TN contribution rates of cropland and construction land in this study exceed 41% and range from 27% to 31%, which are basically consistent with the major contribution structure of agricultural non-point sources (41.1%) and anthropogenic discharge (23.0%) derived from SWAT simulation. Moreover, the seasonal characteristic that wet-season runoff scouring aggravates pollution output from land use types is in good agreement with the runoff-driven pollutant migration mechanism revealed by the SWAT model. Overall, the two studies achieve highly consistent pollution source structures and spatiotemporal variation patterns. Slight discrepancies in contribution rates can be attributed to differences in statistical calibers, research dimensions, and local background conditions of the study areas, whereas the core pollution characteristics remain unchanged. The above cross-validation has fully demonstrated the scientificity and reliability of the source apportionment results in the southeastern coastal river basin.

The pollution contribution characteristics obtained in this study are applicable to watersheds with similar hydrological and land-use backgrounds to the JJRB. If extrapolated to catchments with vastly different underlying surface and land use patterns, the export intensity and contribution proportion of each land use type will differ significantly.

4.4 TN and TP pollution risk

Figure 11 shows the spatial distribution of TN pollution risk in the JJRB. The risk exhibits significant spatial heterogeneity, being higher in the Xixi River than in the Dongxi River, and higher in the lower reaches than in the upper reaches. Furthermore, the spatial pattern of TN pollution risk exhibits significant differences between the wet and dry seasons.

During the wet season, the high-risk areas for TN pollution are predominantly concentrated in the subcatchments of W6, W9, and J2, while the extremely high-risk area is localized within the J3 subcatchment. The combined TN load contribution from these high-risk and extremely high-risk zones accounts for 45.9% of the total basin load. In the dry season, the spatial distribution of risk shifts, with high-risk areas relocating to the subcatchments of W10, E10, and J3, and the extremely high-risk area emerging in the J4 subcatchment, which alone contributes 40.2% of the total basin load.

Figure 12 shows the spatio-temporal distribution characteristics of TP pollution risk in the JJRB. The spatial pattern of TP pollution risk is highly consistent with that of TN, both showing a gradient distribution of “West Stream > East Stream, Downstream > Upstream,” suggesting that the two may have homologous pollution or similar migration processes. The distribution of areas with high and high-risk TP pollution is relatively stable between the wet season and the dry season. During the wet season, high-risk areas are primarily concentrated in subcatchments J2 and W9, alongside an extremely high-risk area in J4. Together, they are responsible for 43.2% of the basin's total TP load. In the dry season, the spatial distribution of risk shifts, with high-risk areas relocating to the subcatchments of W9 and E13, and the extremely high-risk area emerging in the J2 subcatchment. The combined contribution of these three subcatchments reached 47.7%.

Figures 11, 12 reveal obvious seasonal spatial discrepancies between high-risk zones of TN and TP pollution, which are jointly driven by pollution source composition, nutrient transport characteristics and seasonal hydrological processes, consistent with previous watershed nutrient studies (Zhou et al., 2012; Zhu et al., 2024).

TN pollution originates from a mixture of agricultural non-point sources and variable industrial point sources. In wet seasons, intensive rainfall washes fertilizers from upstream tea plantations and generates abundant agricultural runoff, forming high-TN-risk subcatchments (W6) in the upper basin. In dry seasons, agricultural runoff decreases drastically, while undiluted industrial and domestic sewage accumulates in downstream mainstream reaches (J4), driving the downstream shift of TN high-risk zones. In contrast, TP pollution is dominated by steady urban and industrial point-source discharges with limited agricultural inputs, so its spatial pattern is less sensitive to seasonal changes. Most phosphorus is adsorbed onto sediment particles and rarely transported over long distances in rivers, which leads to stable spatial agglomeration of TP high-risk subcatchments (W9, J2) across both seasons. Only the extremely high-risk hotspots of TP slightly switch between J4 (wet season) and J2 (dry season) due to seasonal sediment resuspension and deposition, without large-scale cross-basin displacement.

The fundamental driver of divergent seasonal distribution patterns lies in the distinct mobility of nitrogen and phosphorus: nitrogen mainly exists as dissolved nitrate with strong transport capacity, while particulate-bound phosphorus tends to deposit near discharge outlets and shows weak long-distance diffusion potential.

4.5 Priority control areas identification and management suggestions

4.5.1 Identification of priority control areas

We extracted TN and TP priority control areas using the risk classification criteria in Section 3.2.5. Figure 13 illustrates their spatial in the JJRB under wet and dry conditions. Even with unified grading thresholds, divergent pollutant transport and source structures create marked spatial differences between TN and TP priority zones.

The priority control areas for TN pollution are primarily located in the Xixi River Basin and the downstream reaches of the mainstream. Compared to the dry season, wet-season TN priority control areas generally exhibit a downstream shift. Subcatchment J3 represents a perennial priority control areas for TN, consistently generating substantial pollution loads in both seasons. In contrast, TP priority control areas demonstrate notable spatial clustering and temporal stability, being highly concentrated in the more urbanized regions of the lower basin, which underscores the dominant influence of anthropogenic activities on the spatial distribution of TP pollution. Subcatchments W9 and J2 are identified as persistent priority control areas for TP pollution, exhibiting significant pollution risks throughout both hydrological seasons. The consistent quantitative grading rule eliminates artificial deviation in TN–TP comparative analysis. The spatial divergence between two pollutants mainly lies in pollution source attributes: TN pollution is jointly driven by upstream agricultural non-point sources, urban runoff and industrial wastewater, showing obvious seasonal downstream migration; TP pollution is dominated by concentrated urban domestic sewage and industrial point-source emissions, presenting stable agglomeration in downstream urbanized reaches throughout the year.

Based on pixel-scale risk grading results (high/extremely high risk grids), overlay analysis of land use, industrial layout and agricultural planting distribution was performed to subdivide all priority control areas. The classification criteria integrate risk levels and characteristics of major pollution sources. For the JJRB, priority control areas of TN and TP pollution were identified and classified into three categories according to dominant pollution sources and spatial characteristics. The first category encompasses downstream urbanized regions (e.g., subcatchments J2, J3, and J4), where pollution originates from a complex combination of domestic sewage, industrial effluents, and urban surface runoff. The second category corresponds to the Nan'an industrial agglomeration zone (e.g., subcatchments W9, W10), which is characterized by significant point-source discharges from manufacturing activities. This region has previously been recognized as the area with the highest industrial environmental risk within the JJRB (Zhou et al., 2012; Zhu et al., 2024), and this study further confirms that such risk is manifested through elevated levels of nitrogen and phosphorus pollution. The third category includes upstream tea plantation zones (e.g., subcatchments W6, E10), where non-point-source pollution-primarily driven by fertilizer runoff-is predominant, particularly during the wet season. The spatial distribution and key attributes of these priority control areas are summarized in Table 1.

Table 1. Summary of priority control areas for TN and TP pollution in the JJRB.

All subcatchments listed in the table are areas graded as high-risk or extremely high-risk via the cumulative pollution load curve classification standard in Section 3.2.5.

Category classification is determined by matching risk grids with regional pollution source types and spatial land-use features.

4.5.2 The management strategy of priority control areas

Based on the above pollution characteristics and the identification results of priority control areas, it is recommended to implement precise governance in accordance with the principle of “zoning, classification and grading.” Specific measures include:

• (1) Downstream urbanized areas: for downstream urbanized areas, efforts should be made to enhance sewage collection and treatment efficiency to reduce pollution at the source. Meanwhile, it is imperative to strictly control the total amount of pollutants entering rivers, ensuring that the effluent quality meets the rigorous standards of TP below 0.3 mg/L and TN below 10 mg/L.
• (2) Nan'an industrial agglomeration area: implement strict industrial emission supervision and industrial structure optimization policies, promote key industries such as electroplating, papermaking, and food processing to carry out clean production renovations, and reduce the generation and discharge of high-concentration organic nitrogen and phosphorus pollutants from the source.
• (3) Upstream tea garden planting area: a comprehensive management approach should be implemented to prevent and control non-point source pollution. On one hand, promote the use of organic fertilizers and soil testing and formulated fertilization techniques to develop eco-agriculture, thereby reducing pollutants at the source. On the other hand, enhance soil and water conservation measures to effectively minimize erosion in sloped tea plantations.
• (4) Implement a pollution risk classification governance mechanism: despite ongoing governance efforts, the findings of this study (based on 2020 data) reveal that the risk of nitrogen and phosphorus pollution in Quanzhou City remains severe. Guided by this risk assessment, a targeted governance strategy is recommended. The primary focus should be on extremely high-risk subcatchments, followed by high-risk subcatchments. Concurrently, it is essential to establish an annual water quality monitoring program, dynamic pollution load modeling, and a risk early-warning mechanism to support the dynamic optimization and precise regulation of control measures.

5 Discussion

5.1 Pollution risk level

Monitoring data from the China National Environmental Monitoring Center (CNEMC) between October 24 and November 4, 2025, recorded average TN and TP concentrations of 2.47 mg/L and 0.08 mg/L, respectively, in the JJRB. Compared with other major basins in Fujian, the TN level in the JJRB was intermediate between the Minjiang (1.21 mg/L) and Jiulong (5.7 mg/L) River Basins, while TP levels were consistently low across all three systems (JJRB: 0.08 mg/L; Minjiang: 0.09 mg/L; Jiulong: 0.07 mg/L).These results suggest a moderate state of eutrophication in the JJRB during the monitoring period.

5.2 Uncertainty analysis

First, the regression model adopted the land-use proportions of the whole basin without distinguishing riparian buffer zones, ignoring spatial distance differences in pollutant transport and interception among different land parcels, which may cause deviations in the quantified contribution intensity of pollution sources. Second, a uniform export coefficient was assigned to all areas of the same land-use type, failing to characterize the internal spatial heterogeneity of land patches, such as variable fertilization levels across farmlands and differentiated pollution generation features of surface runoff in urban built-up areas. Third, seasonal average pollutant loads conceal the fluctuations of source contribution rates under extreme rainstorms or prolonged low-flow conditions. Despite the above uncertainties, the seasonal variation patterns and identification of dominant pollution sources revealed in this study remain reliable for non-point source pollution analysis in the Jinjiang River Basin (JJRB).

5.3 Research limitations and future perspectives

This study presents a comprehensive overview of nutrient pollution characteristics and associated risks in the JJRB across distinct hydrological seasons. Nevertheless, two key limitation must be acknowledged:

First, the analysis relies on sampling data collected over a single year, encompassing one wet and one dry season. Although this design is commonly adopted and effective for identifying initial, seasonally contrasting patterns of pollutant sources and transport dynamics, it inherently constrains the capacity to evaluate long-term pollution trends or inter-annual variability. The dataset does not account for influential factors such as inter-annual climatic variations (e.g., extreme precipitation events or prolonged droughts), shifts in land use practices, or fluctuations in anthropogenic discharge rates—all of which can substantially affect nutrient loading. As a result, the source apportionment outcomes and risk assessments reported in this study should be interpreted as reflective of conditions specific to 2020, rather than as fixed or representative of enduring basin-wide attributes. While the general conclusions—particularly the dominant contribution of agricultural runoff during the wet season and the heightened influence of domestic sewage in the dry season—are likely robust, their quantitative estimates may vary across years. To strengthen the reliability of these findings, future research should integrate long-term, multi-annual monitoring initiatives to validate observed patterns, clarify the drivers of temporal changes, and improve the predictive accuracy of risk assessment models for the JJRB.

Second, this study is the use of hydrological analogy to estimate flow discharge in subcatchments lacking flow monitoring data. This method itself may introduce significant systematic errors, as it is difficult to fully achieve consistency between the reference and target subcatchments in terms of hydrogeological conditions, human water use activities, and other factors. Such errors will directly propagate into pollution load calculations, thereby potentially affecting the risk classification results. Future research should strengthen monitoring network construction and conduct quantitative uncertainty assessments.

6 Conclusions

This study systematically investigates the spatial heterogeneity of TN and TP loads across the JJRB, quantifies the relative contributions of distinct pollution sources, and delineates priority control zones for nutrient pollution mitigation. This analysis combines the water chemical data collected and tested in 2020 with satellite remote sensing images and hydrological analogy methods. Results indicate that the JJRB is predominantly in a moderate eutrophic condition, with TN and TP loads exhibiting pronounced spatial gradients—specifically, concentrations are significantly higher in the West River than in the East River, and markedly elevated in downstream reaches compared to upstream regions. Areas identified as high-risk for nutrient pollution include the downstream urban corridor, the Nan'an industrial zone, and upstream tea plantation districts. Source apportionment analysis demonstrates that anthropogenic activities constitute the dominant drivers of basin-wide TN and TP loading. Agricultural non-point sources contribute the largest share of TN (43.2%), whereas urban point sources account for the greatest proportion of TP (38.4%).

Informed by these findings, we propose a spatially differentiated and source-targeted management framework: (i) in the downstream urban corridor, upgrading wastewater treatment infrastructure—including enhancement of tertiary nitrogen and phosphorus removal efficiency—is recommended to rigorously regulate urban point-source discharges; (ii) in the Nan'an industrial zone, enforcement of stricter industrial effluent standards—particularly for phosphorus-intensive sectors—should be coupled with intensified regulatory oversight and process-level upgrades to on-site wastewater treatment systems; and (iii) in upstream agricultural and tea-growing areas, adoption of sustainable land-use practices—such as establishment of vegetated buffer strips and ecological drainage ditches, alongside precision fertilizer application—should be promoted to mitigate diffuse nitrogen losses.

It is important to acknowledge several methodological limitations: the reliance on single-year (2020) monitoring data constrains temporal representativeness; the hydrological analogy method entails inherent assumptions regarding hydrological similarity across sub-basins. To strengthen robustness and generalizability, future research should incorporate multi-year observational datasets.

Generative AI statement

The author(s) declared that Generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/frwa.2026.1863121/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: María Luz Rodríguez-Blanco, University of Vigo, Spain

Reviewed by: Hualong Hong, Xiamen University, China

Shiyu Xue, National Ecological Research Station of Sanya Bamboo and Rattan Associated Forest Ecosystem, China