Hydrodynamic and water quality simulation of Yangzonghai Lake, Southwest China, using the two-dimensional CE-QUAL-W2 model

Abstract

Lake eutrophication is a major environmental concern in southwest China, where external inputs of nutrients and organic pollutants adversely impact plateau lakes like Yangzonghai (YZH) Lake. This study applied the two-dimensional CE-QUAL-W2 model to simulate the hydrodynamics and water quality of YZH Lake from 2016 to 2018, aiming to (i) assess the model’s performance for water level and surface temperature, (ii) identify spatial and temporal patterns of water quality, and (iii) determine pollutant reduction targets to meet Class II water standards. The model accurately simulated observed water levels (RMSE = 0.078 m, AME = 0.066 m, R = 0.91, p < 0.001) while moderately predicted surface temperatures (RMSE = 1.82 °C, AME = 1.53 °C, R = 0.80, p < 0.001). Total nitrogen (TN), total phosphorus (TP), and chemical oxygen demand (CODMn) exceeded Class III standards during strong thermal stratification periods, especially near major river inflows like the Baiyi and Yangzonghai Rivers. High nutrient and chlorophyll-α levels with the lowest transparency were observed during summer stratification. External inflows contributed over 70% of the total nutrient and organic matter input. Reducing external loads by 43% (TN), 26% (TP), and 10% (CODMn) annually is crucial to meet Class II standards. At the 75th percentile, 50%, 35%, and 40% reductions, respectively, are required. The internal cycling and hydrodynamics result in a non-linear relationship between load reductions and in-lake concentrations. The study demonstrates both the strengths and limitations of CE-QUAL-W2 model, offering a scientific foundation for establishing realistic pollutant reduction targets for sustainable lake management.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-42817-0.

Introduction

Lake eutrophication is a major global environmental concern, especially in regions experiencing rapid urbanization and agricultural intensification. Plateau lakes in southwest China, such as Yangzonghai (YZH) Lake, are increasingly threatened by nutrient enrichment and water quality deterioration due to rising exogenous pollutant loads. The primary drivers of eutrophication include external inputs from domestic sewage, industrial and agricultural wastewater, and diffuse sources such as atmospheric deposition and agricultural runoff^1–3. These pressures have led to elevated concentrations of total nitrogen (TN), total phosphorus (TP), and organic pollutants, posing significant risks to both aquatic ecosystems and public health.

In recent decades, the intensification of agricultural activities has accelerated the export of pollutants into lakes through runoff, carrying nitrogen, phosphorus, heavy metals, and organic contaminants^4,5. For instance, in the Taihu Lake Basin, agricultural runoff accounted for 52% of total nitrogen and 54% of total phosphorus loads⁶. Such elevated pollutant loads have led to widespread water quality degradation and increased eutrophication in lakes across China^7–9.

A variety of pollution mitigation strategies have been implemented to reduce non-point source pollution, including minimizing fertilizer application rates¹⁰, improving field management practices¹¹, adopting innovative planting patterns^12–14, planting vetiver at field boundaries¹⁵, converting steep slopes to terraces, and constructing lakeside wetlands¹⁶. However, the effectiveness of these interventions is highly site-specific and often insufficient to achieve targeted water quality standards, especially in complex plateau lake systems. This challenge underlines the need for robust, site-based scientific assessments to inform sustainable and cost-effective management strategies^17,18.

A critical step in lake management is the accurate quantification of water environmental capacity—the maximum allowable pollutant load that ensures compliance with water quality standards¹⁹. Determining environmental capacity requires identifying key pollution indicators, setting water quality function targets, and applying suitable water quality models to simulate pollutant fate under varying hydrological conditions¹⁸. Forward iterative modeling is often used to calibrate and validate these estimates, providing a scientific foundation for setting reduction targets and management policies.

Hydrodynamic and water quality models have become indispensable tools for understanding pollutant transport, evaluating management scenarios, and supporting evidence-based policy decisions. Models such as ANSWERS, AGPNS, SWAT, PLOAD, and RUSLE have been widely used to predict non-point source pollution in diverse environments^20–22. Among these, the CE-QUAL-W2 model is highly valuable for simulating the hydrodynamics and water quality of long, narrow lakes and reservoirs^23–25. CE-QUAL-W2 enables detailed assessment of water level fluctuations, thermal stratification, and nutrient cycling, which are crucial for evaluating the impacts of exogenous pollutants and for setting realistic load reduction targets. The two-dimensional characteristics of CE-QUAL-W2 support both horizontal and vertical simulation of thermal and hydrodynamic processes, enhancing the understanding of spatial heterogeneity in lake environments^26–28.

Recent applications of CE-QUAL-W2 have demonstrated its utility in diverse contexts, such as assessing the impact of climate change on nutrient dynamics in Lake Diefenbaker²⁹ and simulating temperature and water quality in plateau lakes in Yunnan Province, China^30,31. However, limitations persist, specifically in simulating lake surface temperature. Inaccuracies in boundary conditions, such as inflow/outflow rates and meteorological data, can lead to discrepancies between simulated and observed temperatures. Consequently, a root mean square error (RMSE) of less than 2 °C is considered a moderately acceptable performance for temperature simulations in complex lake systems³².

Despite these advances, most modeling studies in China have focused on lakes in plain regions, such as Chaohu and Taihu in the Yangtze River Basin^13,22,33. Yangzonghai Lake, which is situated at a high altitude on the Yunnan Plateau, differs in hydrodynamic and ecological characteristics from the low-lying lakes that are most commonly researched in China. The high altitude significantly influences atmospheric pressure, dissolved oxygen saturation, and solar radiation regimes, resulting in distinct thermal stratification patterns and oxygen dynamics. In comparison to lowland lakes, the equilibrium dissolved oxygen saturation is lower at high elevations due to lower air pressure, while plateau lakes’ distinctively low water temperatures increase oxygen solubility and delay metabolic oxygen use. Thus, in contrast to lowland systems like the Three Gorges, the combined effects of altitude and temperature form the deep-water redox environment, which in turn modifies sediment nutrient release processes. Additionally, the lake’s complex watershed topography, which is marked by steep slopes and dissected terrain, produces highly variable inflow regimes through rapid surface runoff response to precipitation. As a result, the lake’s water levels and mixing patterns are dynamic and significantly different from those of lakes with gentler catchment gradients, thus necessitating tailored management approaches. These unique attributes justify focused modeling studies in plateau lake systems, and well suited for CE-QUAL-W2 modeling²³ to capture these physiochemical interactions critical to water quality and ecosystem health. Yet, uncertainties due to meteorological input, especially wind data, can still affect model accuracy.

Given the ecological, socio-economic, and strategic importance of plateau lakes, and in line with global sustainability targets such as the United Nations Sustainable Development Goal 6 (Clean Water and Sanitation), there is an urgent need for robust, site-specific scientific assessments to inform effective management and restoration strategies. To address these gaps, this study focuses on Yangzonghai (YZH) Lake, a representative plateau lake in southwest China. The objectives are to: (i) Apply the CE-QUAL-W2 model to simulate hydrodynamics and water quality in YZH Lake; (ii) Evaluate the model’s performance, especially for surface temperature and water level, under real-world data constraints; and, (iii) Quantify pollutant reduction targets necessary to achieve Class II water quality standards following Chinese regulations. By providing a robust scientific basis for lake management, this research supports the development of effective strategies for mitigating eutrophication and safeguarding water resources in plateau lake regions, with implications for sustainable development and regional policy.

Materials and methods

Study site

Yangzonghai (YZH) Lake (102.55°–103.02°N, 24.27°–24.54°E) is a typical plateau lake situated in the low-latitude region of southern China. The lake extends approximately 12.6 km from north to south, with an average width of 2.6 km and a maximum depth of 30 m (mean depth ~ 20 m; Fig. 1). The mean annual temperature of the area is 14.5 °C, while mean annual precipitation is 963.5 mm. YZH Lake shows distinct seasonal stratification, with stable thermal layers in summer and autumn, mixing in winter, and transitional stratification in spring; therefore, it is classified as a warm monomictic lake.

Fig. 1.

Underwater topographic map of Yangzonghai (YZH) Lake.

The YZH lake is a major water source for local communities, supporting fisheries, recreation, and regional biodiversity. Long, narrow morphology and ecological significance of YZH Lake make it an ideal case for advanced hydrodynamic and water quality modeling. The map of Yangzonghai Lake in Fig. 1 was created by one of our co-authors (Haitao Feng) using QGIS, version 3.28, while the maps of China and Yunnan Province were downloaded from the government website of Ministry of Natural Resources of China, which is publicly open website and does not require permission.

Water quality monitoring

The YZH lake was comprehensively monitored for water quality from January 2016 to January 2019 at three representative stations—northern (YZH_North), central (YZH_Center), and southern (YZH_South) regions of the lake—to study spatial variability. Sampling was performed monthly, with three replicates per site, totaling 324 samples. Where possible, samples were collected at the surface, mid-depth, and near-bottom layers to account for vertical gradients. Water transparency (Secchi depth) was measured in situ.

Samples were immediately transported to the laboratory at 4 °C and analyzed following Chinese national standards (Table S1) and the “Water and Wastewater Monitoring and Analysis Methods (Fourth Edition, 2002).” Samples were analyzed for total phosphorus (TP), total nitrogen (TN), potassium permanganate index (CODMn), chlorophyll-a (Chl-α), and water transparency (SD) using standard analytical methods (e.g., spectrophotometry for TP and TN) and QA/QC procedures (use of blanks, replicates, and certified reference materials) to ensure data reliability (Table 1).

Table 1.

Determination methods for water quality index.

Index	Indicator	Test method	Standard coding
CODMn	Potassium permanganate value	Acidic permanganate method	GB 11892 − 1989
TP	Total phosphorus (phosphate)	Ammonium molybdate spectrophotometry	GB 11893 − 1989
TN	Total nitrogen	UV spectrophotometry	GB 11894 − 1989
NH4+-N	Ammonium nitrogen	Nessler’s reagent photometry	GB7479-1987
Chl-α	Chlorophyl	Spectrophotometry	HJ 897–2017
SD	Transparency	Plug disc method	SL87-1994

Model description and rationale

A two-dimensional, CE-QUAL-W2 model (v4.0) was selected for its proven capability to accurately simulate the hydrodynamics and water quality processes of long, narrow, and stratified lakes like Yangzonghai Lake^23,3031. Its two-dimensional longitudinal and vertical framework allows detailed resolution of temperature stratification, flow regimes, and water quality gradients essential to understanding pollutant transport and transformation, essential for evaluating management scenarios in plateau lakes.

The simulation process started with hydrodynamic calibration, comparing simulated water levels and volumes with observed data. The water quality module was then initiated to simulate nutrient-Chl-α relationships, dissolved oxygen dynamics, and thermal stratification. This model adopts a uniform horizontal distribution of velocity, temperature, and water quality components within each cross-sectional segment, and focuses on material changes in both longitudinal and lateral directions, with lateral values representing averages. Governing equations are based on conservation of mass and energy, free surface hydraulics, and water quality kinetics^34,35.

In the water quality module, CE-QUAL-W2 models various key constituents of water quality, including dissolved oxygen (DO), slowly degradable particulate organic carbon (representing CBOD), inorganic nitrogen species (ammonia NH₄⁺-N, nitrate NO₃⁻-N, nitrite NO₂⁻-N), refractory and labile organic nitrogen, inorganic phosphorus (orthophosphate PO₄³⁻-P), refractory and labile organic phosphorus, total nitrogen (TN), total phosphorus (TP), chemical oxygen demand as COD_Mn (permanganate index tracking organic oxidizable matter), and chlorophyll-a (Chl-a) as proxy for phytoplankton biomass. COD_Mn primarily reflects organic matter oxidizable by permanganate and serves as a proxy for biological oxygen demand (BOD), correlating well with organic carbon decay processes simulated in the model²³. Despite the unavailability of observed data on dissolved oxygen (DO) during the monitoring period, the CE-QUAL-W2 used standard methods to simulate DO dynamics. These include reaeration, photosynthesis, respiration, and sediment oxygen demand, which are essential for studying oxygen availability and its impacts on nutrient cycling, especially sediment phosphorus and ammonia release.

Boundary conditions and input data for nutrient species were derived from extensive monitoring of incoming rivers (Yangzonghai, Qixing, and Baiyi Rivers), emphasizing measured concentrations of individual nutrient species rather than bulk TN and TP. As a result, TN and TP within the model are dynamic outputs, calculated as sums of their constituent inorganic and organic forms, reflecting integrated nutrient cycling processes rather than static input parameters.

In accordance with normal CE-QUAL-W2 practice for temperate lakes, the aquatic biological component includes three algal functional categories: diatoms, cyanobacteria, and green algae²³. These groups represent the dominant phytoplankton dynamics found in meso-eutrophic plateau lakes such as Yangzonghai, where transitional states are characterized by green algae, cyanobacteria predominate under nutrient enrichment, and diatoms predominate in cooler, silica-rich conditions^36,37. This allows the model to accurately represent the dynamics of seasonal algal blooms and their effect on water quality.

Although the CE-QUAL-W2 model permits specification of internal nutrient release from sediments, including ammonia and phosphorus fluxes, which are driven by sediment oxygen demand and redox conditions. Such internal loading processes were not explicitly modeled in this study because of the lack of site-specific sediment chemistry, oxygen demand, and redox potential data necessary for parameterization. Given the limitations of the available data, our analysis concentrated mainly on external nutrient load decreases because we recognized that sediment nutrient release can significantly affect in-lake nutrient dynamics and eutrophication processes. In order to fully assess both external and internal nutrient inputs, we admit that internal loading may have a major impact on nutrient cycling and algal development. We suggest focused field measurements and additional modeling that include internal fluxes in future study.

Model setup and discretization

YZH Lake was divided into 76 longitudinal segments (average length: 160 m; range: 100–272 m) using bathymetric data from the 2017 sonar survey and lake boundary vectors. Each longitudinal segment was identified by a water level–area relationship curve. Each segment was separated vertically into 60 layers of 0.5 m thickness, resulting in 4,560 computational grid cells. This resolution balances stratification and flow patterns as well as computational efficiency. During simulation, the thickness of the surface layer was dynamically adjusted to represent real variations in water level.

Hydrological inputs and boundary conditions

YZH Lake receives inflows from three main rivers: Yangzonghai River, Qixing River (south bank), and Baiyi River (north bank). The east, west, and north banks (without significant rivers) were combined as a single inflow source, resulting in six inflow boundaries for the model. The combined catchment area of Yangzonghai, Qixing, and Baiyi Rivers is approximately 200.3 km², accounting for 77% of the YZH basin.

Inflow boundary conditions were determined from observed discharge data where available (e.g., Baiyi River daily discharge). For other catchments, inflows were estimated using the runoff reduction method, distributing monthly land runoff proportionally among inflow areas. Where direct inflow water quality data were unavailable, values were estimated based on land use and regional monitoring reports.

Meteorological data (air temperature, wind speed, humidity, solar radiation, precipitation) were obtained from the Yiliang County Meteorological Station, located 5 km from the lake. Missing data (< 5% of records) were filled using linear interpolation. These data provided surface forcing for the model, which is critical for simulating thermal stratification and surface temperature. The accuracy of meteorological inputs, especially wind data, was recognized as a key factor influencing model performance for temperature simulation.

Water balance calculation

A daily water balance was calculated to ensure mass conservation and accurate hydrodynamic forcing, using the following equation:

where: ΔV= daily change in lake volume (m³/day), W1 = total inflow (m³/day), W2 = precipitation over the lake surface (m³/day), W3 = outflow to Tangchii River (m³/day), W4 = water consumption by industry and agriculture (m³/day), W5 = evaporation loss (m³/day), calculated using the Penman formula.

These computations’ time series cover the whole simulation duration.

Based on the runoff reduction calculations, an annual average runoff from land to the lake in the YZH basin was 52.24 million m³ between 2016 and 2018. Among this, Baiyi River contributed 15.15 million m³, while land runoff into the lake contributed about 37.09 million m³. The estimated annual average land runoff from Yangzonghai River, Qixing River, the north bank, the west bank, and the east bank remained approximately 3.17 million m³, 20.22 million m³, 3.68 million m³, 5.75 million m³, and 4.25 million m³, respectively. During the same period, the average annual evaporation and rainfall were 1167 mm and 850 mm, respectively. Monthly runoff data for the rivers, along with measured daily runoff from Baiyi River, was collected as land inflow boundaries for the model. Daily discharge observations from Tangchi River and daily water intake observations were taken as the outflow boundary. Water quality monitoring data from the Yangzonghai River, Qixing River, and Baiyi River, which included indicators such as water temperature, TN, TP, NH₄⁺-N, and COD, were directly used as concentration boundaries for incoming pollutants. Hourly data on temperature, relative humidity, wind speed, wind direction, cloud cover, and radiation were collected from Kunming meteorological station, which is approximately 27 km away from YZH.

To determine their impact on water level projections, a sensitivity analysis was conducted by adjusting key factors (such as the evaporation coefficient and abstraction rates) within reasonable ranges.

Model calibration and validation

Model calibration was done using data from 2016 to 2017, with 2018 data reserved for validation. Calibration process optimized parameters such as wind sheltering and vertical mixing coefficients to best reproduce observed water levels and surface temperatures. Model performance was assessed by root mean square error (RMSE), mean absolute error (MAE), and correlation coefficient (R). An RMSE < 2 °C for temperature was considered moderately acceptable in line with previous studies³². Calibration and validation results, including time series plots and statistical summaries, are provided in Table S2 and Figure S1 given in the Supplementary Material.

Scenario analysis for pollutant load reduction

To determine the pollution load reductions necessary to meet Class II water quality criteria, a scenario analysis was carried out. Every scenario was simulated over three years, with external TN, TP, and COD loads systematically decreased in 10% increments. Model outputs were compared to regulatory thresholds to identify the minimum reductions necessary for compliance.

Although CE-QUAL-W2 supports simulation of sediment nutrient release processes, sediment nutrient release was not explicitly modeled in this study. This limitation is due to the absence of site-specific sediment chemistry data and redox potential measurements essential for parameterizing sediment-water nutrient fluxes. Consequently, the model focused on external nutrient loadings from tributary inflows as the primary driver of lake water quality dynamics.

Ethical considerations and data availability

All field sampling and monitoring activities complied with local environmental regulations. Data supporting this study are available in the supplementary materials or upon reasonable request from the corresponding author.

Results

Model performance: hydrodynamics and surface temperature

During the study period from 2016 to 2018, the CE-QUAL-W2 model demonstrated strong performance in simulating the hydrodynamics of YZH Lake. With a correlation coefficient (R) of 0.91 (p < 0.001), root mean square error (RMSE) of 0.078 m, and mean absolute error (MAE) of 0.066 m, the simulated water levels are closely related to observed data, indicating strong model fitness for hydrodynamic processes (Fig. 2).

Fig. 2.

The model simulates the daily water level process and observes the daily water level process.

The mean RMSE, MAE, and R values for surface temperature across all monitoring stations were 1.82 °C, 1.53 °C, and 0.80 (p < 0.001), respectively, indicating a moderate fit for the model (Fig. 3). These findings align with earlier studies using CE-QUAL-W2 in plateau lakes, where temperature simulation accuracy is usually constrained by uncertainties in boundary conditions and meteorological inputs^30,32. Although significant variations were observed during periods of rapid change in temperature, mostly due to less precise wind data and inflow temperature estimates, the model successfully reproduced seasonal thermal stratification and vertical mixing events.

Fig. 3.

Model simulation and measurement of surface water temperature.

Model simulations showed accurate seasonal DO patterns closely linked to thermal stratification (similar to reported temperature profiles in Fig. 4), despite the lack of direct DO measurements. Because of air exchange and photosynthesis, surface DO maintained close to saturation (~ 8–10 mg/L) throughout the year, but bottom waters experienced hypoxia (< 2 mg/L) during summer stratification (June–October), especially in deeper parts (> 15 m). In line with plateau lake dynamics, these modeled patterns—which are fueled by stratification-suppressed vertical mixing, algal respiration, and sediment oxygen demand—highlight the dangers of internal nutrient mobilization under low-DO circumstances³⁷.

Fig. 4.

Model simulation of vertical water temperature processes at monitoring stations.

Daily vertical water temperature profiles were recorded at monitoring stations located at the north, middle, and south of YZH Lake (Fig. 4). In January, during the months of winter, the lake showed a uniform water temperature across both horizontal and vertical directions. As temperatures started to rise in February, vertical differentiation in water temperature was observed, with a slight vertical temperature gradient in March. This gradient further increased with increasing temperatures, leading to the formation of temperature stratification by April due to a distinct thermocline between 12 and 20 m depth. Below 20 m, a stable isothermal layer is formed, less influenced by external factors. This vertical temperature stratification persisted until October, which contributed to improving water quality by limiting the vertical transmission of pollutants. Water quality monitoring included systematic sampling at upper, middle, and lower depths of Yangzonghai Lake, providing vertical nutrient concentration profiles that were used to characterize stratification effects. Both measured data and model simulations demonstrate an accumulation of nutrients. During spring and summer, pollutants such as P and N species were observed accumulating in the upper water layers, while the deeper layers remained relatively unaffected. This pattern contrasts with general assumptions of higher nutrient accumulation near the sediments during stratification. However, as stratification began to reduce in November, and by December, the lake water became more uniform again. This process might have facilitated the upward movement of pollutants from deeper layers to the surface, contributing to nutrient supplementation. The relatively well-oxygenated bottom waters and dynamic inflows likely reduce nutrient release from sediments, limiting nutrient accumulation in deeper layers.

While sediment nutrient release under hypoxia is a common phenomenon in many lakes, Yangzonghai Lake’s oxygen regime and inflow dynamics reduce this effect here. Nonetheless, we recognize that more detailed sediment chemistry and vertical biogeochemical sampling would strengthen these findings and recommend such efforts in future studies. The water temperature below 20 m in YZH Lake remained relatively stable throughout the year, typically ranging from 13 to 15 °C.

Variations in water quality by time and space.

Dynamics of organic and nutrient pollution

Throughout the study period, water quality monitoring revealed a consistent increase in chemical oxygen demand (CODMn), total nitrogen (TN), and total phosphorus (TP), all of which exceeded the water quality criteria for Class III. Higher concentrations of TN and TP in the northern and central regions corresponded to the main inflow locations and agricultural activity areas, as indicated by the simulated geographical distributions (Fig. 5). According to temporal analysis, the periods of substantial thermal stratification and decreased vertical mixing corresponded with the summer and early fall peaks of nutrient concentrations.

Fig. 5.

The water quality process was simulated and measured at three monitoring sites.

Chlorophyll-a (Chl-α) concentrations also exhibited significant seasonal variation, with the highest values observed in late summer, reflecting increased phytoplankton growth under high nutrient and temperature conditions. Water transparency (Secchi depth) was the lowest during these periods, indicating increased algal biomass and reduced light penetration.

Contributions from external loads

The water quality of YZH Lake was mostly deteriorated by external loads of TN, TP, and CODMn from riverine inflows, as confirmed by model simulations (Fig. 6). During the period from 2016 to 2018, the average inflow load for TP and TN was found to be 6.72 t and 111.7 t, respectively. More than 70% of all external inputs were exported from the Baiyi and Yangzonghai rivers, which also delivered the highest loads of organic matter and nutrients (Table S3). Although they were not specifically modeled, internal nutrient cycling and sediment discharge might be the secondary reasons for water quality deterioration under current conditions.

Fig. 6.

Statistical results of monthly inflow load into YZH Lake from 2016 to 2018.

The mean values and trends of key water quality indices across the entire lake are shown in Fig. 7. Starting in 2016, TN, TP, and NH₄⁺-N in YZH Lake increased to the highest level in 2017 and gradually declined in subsequent years. On the other hand, COD demonstrated a generally stable and slightly decreasing trend from 2016 to 2020, opposite to the patterns observed for N and P. Chl-α content initially reduced in 2016, increased in 2017, and reached its peak during 2018 and 2019, and then declined in 2020. Due to Chl-α, there was a significant decrease in SD of YZH Lake water in 2017, while maintaining stability from 2018 to 2020.

Fig. 7.

Changes in water quality trends at monitoring stations in YZH Lake.

Model validation focused primarily on water temperature, nutrient species, chemical oxygen demand, and Chl-α due to limited availability of observed dissolved oxygen (DO) data at Yangzonghai Lake during the study period. Despite this limitation, the CE-QUAL-W2 model incorporated comprehensive DO dynamics, simulating processes such as reaeration, photosynthesis, respiration, and sediment oxygen demand, which are necessary for estimating oxygen concentration gradients and their impact on nutrient cycling and aquatic life. Model-predicted DO profiles revealed typical seasonal patterns for high-altitude lakes, with oxygen saturation near the surface during mixing periods and hypoxic conditions developing near the sediments under summer stratification. These simulated DO levels highlight potential zones of stress for benthic organisms and areas where internal nutrient release might be enhanced. The absence of observed DO data is acknowledged as a limitation, and future monitoring campaigns are recommended to strengthen validation and enhance modeling accuracy.

Figure 8 presents the annual statistical outcomes of key water quality indicators in YZH Lake spanning from 2016 to 2020. The average TN value in YZH Lake meets the Class III surface water quality standard but does not qualify as Class II. TP exhibited a significant decrease in 2017, temporarily exceeding Class II standards in 2019 but subsequently meeting the same post-2019. However, it’s crucial to acknowledge that TP monitoring data in YZH retained only one significant digit, introducing a potential for errors in the results. NH⁴⁺-N concentrations in YZH remained low, meeting and complying with Class II quality standards, and in 2018, achieved the standards for Class I. COD qualified for Class III, because it slightly exceeded the requirements for Class II water quality standards.

Fig. 8.

Average water quality and its changing trends in YZH Lake.

Water environmental capacity and reduction targets for pollutants.

The scenario analysis revealed that YZH Lake’s Class II water quality standards can only be met by significant decreases in external pollution loads. To satisfy average Class II standards, annual external loads, particularly, must be reduced by 43% (TN), 26% (TP), and 10% (CODMn) (Fig. 9). The necessary reductions rise to 50% (TN), 35% (TP), and 40% (CODMn) when the 75th percentile criteria are followed.

Fig. 9.

YZH Load reduction and water quality response diagram (the red dots in the figure are mean values, the red horizontal line in the box plot is median values, the upper horizontal line in the box plot is 75% quantile, and the upper horizontal line is 95% quantile).

Particularly for CODMn, the relationship between in-lake concentrations and external load reductions was non-linear, with diminishing effects at greater reduction levels. This pattern demonstrates how the hydrodynamic regime of the lake, legacy loads, and internal nutrient cycling affect water quality.

Model uncertainties and limitations

While the hydrodynamics and water quality dynamics of YZH Lake were better understood by using CE-QUAL-W2, several limitations still need to be considered. The moderate model fitness for surface temperature might be associated with uncertainties in meteorological forcing, particularly wind speed and direction, as well as incomplete inflow temperature data. While CE-QUAL-W2 offers the capacity to simulate internal nutrient release from sediment layers—such as ammonia and phosphorus fluxes driven by redox conditions—these processes were not explicitly modeled in this study. This decision was primarily due to the lack of site-specific sediment chemistry and redox data necessary for accurately parameterizing sediment nutrient fluxes. Previous research³² highlights that internal loading can be significant in lakes with extensive sediments and weak oxygen stratification. However, without detailed sediment chemistry, diagenetic process data, and redox potential measurements, the model’s internal flux estimates would carry high uncertainty and limited reliability. Moreover, the model did not explicitly consider internal nutrient loading from sediments or direct atmospheric deposition, which may contribute to residual nutrient concentrations during low-flow periods.

However, we acknowledge that sediment nutrient release could substantially influence nutrient levels and algal growth, especially under hypoxic or anoxic conditions. Future research should focus on incorporating sediment fluxes into the modeling framework to better quantify both external and internal contributions to water quality deterioration.

Despite these limitations, model validation statistics and scenario outcomes are consistent with earlier research on plateau lakes in southwest China^30,31, which also supports the robustness of the findings for management and policy applications.

Discussion

Interpretation of hydrodynamic and thermal stratification results

The CE-QUAL-W2 model effectively simulated the hydrodynamic behavior and seasonal stratification patterns of Yangzonghai (YZH) Lake, as shown by the close relationship between simulated and observed water levels and temperatures. The model’s ability to reproduce the seasonal cycle—marked by stable stratification in summer and autumn, mixing in winter, and transitional periods in spring—makes YZH a warm monomictic lake. This stratification regime is crucial for understanding nutrient dynamics, as it acts as a physical barrier that limits vertical mixing and the redistribution of nutrients and oxygen between surface and deep layers^36,38.

The thermocline, typically found between 12 and 20 m below the surface, limits the upward flow of nutrients from deeper waters during the stratification phase, making contaminants and phytoplankton accumulate in the epilimnion. Nutrient-rich surface waters encourage phytoplankton development due to increased chlorophyll-α and decreased water transparency during the summer. On the other hand, the interruption of stratification in late autumn and winter promotes vertical mixing, which enables organic matter and nutrients from the hypolimnion to rise to the surface, leading to nutritional supplementation or secondary algal blooms. These results conform with previous research on temperate lakes and plateaus, where mixing cycles and stratification drive seasonal variations in water quality^39–41.

Nutrient mechanisms and organic pollution dynamics

The increasing concentrations of total nitrogen (TN), total phosphorus (TP), and chemical oxygen demand (CODMn), persistently exceeding the Class III standards throughout the study period highlights the dominant role of external loading from riverine inflows, particularly the Baiyi and Yangzonghai Rivers. The spatial distribution of higher nutrient concentrations near inflow points and agricultural catchments underlines the impacts of watershed land use and non-point source pollution^22,42. The temporal variation shows that nutrient levels are higher during the periods of strong stratification, further supporting the mechanism whereby limited vertical mixing during summer allows for the accumulation of nutrients and organic matter near surface waters, increasing phytoplankton proliferation and reducing water transparency.

The observed decreases in TN and TP concentrations during the flood season (May–October) may be associated with dilution effects of increased runoff, while the increase in chlorophyll-α during summer is attributed to high temperatures and nutrient availability, which promotes phytoplankton growth^43–45. The combined effects of hydrological conditions, external loading, and internal cycling processes, such as sediment nutrient release during periods of hypolimnetic anoxia, result in interannual variability in nutrient and chlorophyll-α trends^46,47.

The lack of direct in situ dissolved oxygen data is a limitation constraining quantitative validation of the modeled oxygen dynamics. Nevertheless, incorporation of oxygen flux processes within CE-QUAL-W2 provides mechanistic understanding of oxygen variability and its pivotal role in sediment nutrient release and aquatic ecosystem function. Future research integrating extensive oxygen profiling and redox potential measurements is essential to refine model calibration and enhance predictive confidence for oxygen-related processes in Yangzonghai Lake.

Model results, constraints, and consequences

In complex natural lakes, uncertainties in meteorological forcing (particularly wind speed and direction) and inflow temperature data are frequent challenges that always exist. However, the model’s moderate fit for surface temperature (RMSE ≈ 1.8 °C) for YZH falls within the range reported for similar applications^31,32. Discrepancies that arise from abrupt temperature change necessitate the need for improved boundary condition data and higher-resolution meteorological inputs, even if the model accurately simulates broad temperature patterns and water level dynamics⁴⁸.

Long, narrow lakes like YZH are well suited for the two-dimensional structure of the CE-QUAL-W2 model, which allows for detailed simulation of longitudinal and vertical gradients^23,30. However, the model does not explicitly consider internal nutrient loading from sediments or direct atmospheric deposition, which may contribute to residual nutrient concentrations, particularly during low-flow or stratification periods. Recent advances in CE-QUAL-W2 have improved the simulation of internal nutrient cycling but require high-quality input data and calibration⁴⁹. These limitations should be addressed when interpreting scenario results and establishing management goals.

The exclusion of explicit internal nutrient loading processes in our model, due to insufficient sediment chemistry and redox data, is a limitation that may affect the completeness of nutrient cycling representation. While external load reductions demonstrated clear influence on water quality, sediment nutrient release remains a potential major contributor to nutrient availability and algal blooms, especially in stratified or hypoxic conditions. Incorporating detailed sediment fluxes via CE-QUAL-W2’s internal loading option in future studies, supported by targeted sediment oxygen demand and chemistry measurements, would enhance understanding of the relative importance of internal versus external nutrient sources. This integrated approach is essential for more effective lake management and eutrophication control strategies.

Effectiveness of load reduction scenarios and management implications

Scenario analysis showed that significant reductions in external TN, TP, and CODMn loads are necessary to meet Class II water quality standards. Internal cycling, legacy nutrient pools, and the lake’s hydrodynamic regime influence the non-linear response of in-lake concentrations to load reductions, especially the decreasing benefits at higher reduction levels^50,51. This outcome highlights the need for integrated watershed management strategies focusing on both point and non-point sources, along with in-lake processes.

The results are consistent with findings of studies on other plateau and temperate lakes, where reducing external load reductions are necessary but may not always be impactful to quickly improve water quality due to internal feedbacks and delayed ecosystem responses². Setting specific reduction targets such as 43% for TN, 26% for TP, and 10% for CODMn provides actionable guidance for policymakers and prioritizes interventions in the most impactful sub-catchments.

Mechanistic insights and broader relevance

The study shows that the interaction between hydrodynamics, stratification, and external loading controls the seasonal and spatial patterns of eutrophication in plateau lakes. Depending upon the season, the vertical stratification can either hinder or facilitate nutrient cycling, although external inflows continue to be the major source of nutrient enrichment^27,28. These results highlight the importance of adaptive management strategies that consider in-lake and watershed processes, particularly under changing climate and land use conditions⁵².

Recent research emphasizes that targeting both external and internal nutrient sources are crucial for sustainable lake restoration. This includes using cutting-edge remediation techniques like Phoslock, alum, sediment removal, and vegetative buffers as well as circular management strategies that recover and reuse nutrients⁵³,. Combining remote sensing with high-frequency monitoring further improves the ability to efficiently monitor eutrophication trends and inform management decisions⁵⁴.

By utilizing a robust modeling framework and scenario analysis, this study contributes to the knowledge of eutrophication mechanisms in plateau lakes and proposes a scientific basis for setting practical and efficient water quality targets. Regional and national water management policies can benefit from the method and insights, which are generally relevant to similar lake systems dealing with eutrophication.

Study strengths, limitations, and future directions

One of the main strengths of this study is the integration of high-resolution monitoring data, sophisticated hydrodynamic modeling, and scenario-based management analysis. The use of a multi-year dataset including dry, normal, and wet years improves the robustness and generalizability of the results. However, its limitations include the exclusion of internal nutrient loading processes, uncertainties in meteorological and inflow data, and the moderate fit for surface temperature simulations.

Future research is recommended to focus on integrating sediment nutrient release and atmospheric deposition into the modeling framework³⁷; enhancing the spatial and temporal resolution of meteorological and inflow data; assessing the long-term impacts of climate change and land use dynamics on stratification and nutrient cycling⁵⁵; and evaluating the efficacy of combined watershed and in-lake management interventions, including nutrient recovery and circular economy approaches. By addressing these areas, future research can further refine the prediction capacity of models like CE-QUAL-W2 and boost the effectiveness of lake management technologies under current environmental change.

Conclusions

The two-dimensional CE-QUAL-W2 is a useful model for simulating the hydrodynamics and water quality of long and narrow plateau lakes with complex shapes and stratification patterns. The model produced a good fit for water level changes but a moderate fit for surface temperature, which might be due to complex lake morphology with limited meteorological and inflow data resolution. High levels of TN, TP, and CODMn in Yangzonghai Lake, exceeding water quality standards for Class III, are mainly due to external inputs from river inflows, especially from agricultural catchments.

Significant reductions in external TN, TP, and CODMn loads are crucial to meet Class II water quality standards, with non-linear improvements in in-lake concentrations associated with internal cycling and hydrodynamic processes. The findings of this study underscore the significance of integrated management strategies that address both watershed and in-lake sources, as well as the requirement for comprehensive monitoring, high-quality input data, and adaptable modeling frameworks.

While the CE-QUAL-W2 model provides valuable insights into lake management, its limitations—particularly regarding internal nutrient cycling and meteorological uncertainties—should be acknowledged. To further improve the model’s predictive capacity and management relevance, future work should incorporate sediment nutrient release, atmospheric deposition, and climate change impacts. Overall, this research provides a scientific foundation for setting realistic pollutant reduction targets, informing policy, and supporting the sustainable restoration and management of plateau lakes in China and similar regions globally.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

All authors contributed equally to the manuscript. C. Tang and J. Wang contributed to securing funding, CE-QUAL-W2 model application, data analysis, and manuscript writing. H. Feng created map of Yangzonghai Lake. H. Feng created the map of Yangzonghai Lake. H. feng, C. Yao, and J. Zhao played key roles in developing methodology, administration, and resource provision. L. Zhao and Z. Hussain reviewed and edited the manuscript. All authors have read and agreed to the published version of the manuscript.

Funding

Authors acknowledge the support of funding by Yunnan Provincial Science and Technology Department Provincial and Municipal Integration Project “Yangzonghai Intelligent Supervision and Intelligent Decision-making Platform R & D Application (Grant No. 202202AH210007) and “Study on temporal and spatial variation of COD and its causes in Yangzonghai Lake, Yunnan Province, China”, supported by Yunnan Research Academy of Eco- Environmental Sciences.

Data availability

Data supporting this study are available in the supplementary materials or upon reasonable request from the corresponding author.

Declarations

Competing interests

The authors declare no competing interests.