Next Article in Journal
Hydraulic Performance of a Box Culvert-Type Two-Way Channel Pumping System: An Experimental Study
Previous Article in Journal
Enhancing Planning in the Management of Urban Water Systems to Increase Resilience
Previous Article in Special Issue
Selective Degradation of Organic Pollutants via Peroxymonosulfate-Based Electrochemical Advanced Oxidation Driven by Different Electrodes: Performance and Influencing Factors
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Water
Volume 18
Issue 3
10.3390/w18030389
water-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

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

A Novel Weather Generator and Soil Attribute Database for SWAT to Improve the Simulation Accuracy in the Heilongjiang Region of China

by
Zhihao Zhang
1,2,
Haorui Zhang
1,2,
Xiaoying Yu
3,
Chunyan Yang
4,* and
Tong Zheng
1,2,*
1
School of Environment, Harbin Institute of Technology, Harbin 150090, China
2
Heilongjiang Provincial Key Laboratory of Polar Environment and Ecosystem, Harbin Institute of Technology, Harbin 150090, China
3
Heilongjiang Provincial Research Academy of Environmental Sciences, Harbin 150056, China
4
College of Architecture and Environment, Sichuan University, Chengdu 610207, China
*
Authors to whom correspondence should be addressed.
Water 2026, 18(3), 389; https://doi.org/10.3390/w18030389
Submission received: 27 December 2025 / Revised: 28 January 2026 / Accepted: 29 January 2026 / Published: 3 February 2026
(This article belongs to the Special Issue Advanced Oxidation Technologies for Water and Wastewater Treatment)
Browse Figures
Versions Notes

Abstract

This study addresses the issue of missing basic data and insufficient accuracy in predicting runoff and non-point-source pollution in the Heilongjiang region of China using the Soil and Water Assessment Tool (SWAT) model. Based on the China Ground Climate Data Daily Dataset (V3.0) and SPAW soil characteristic calculation formula, and assisted by the Python V3.0 language for data processing and computation, new high-precision weather generators and soil attribute databases suitable for the Heilongjiang region of China were established. The weather generator is based on daily data and contains detailed meteorological parameters such as temperature, humidity, wind speed, rainfall, etc., used to characterize the periodic changes in meteorological elements. And the differences and fluctuations outside this change curve were also retained in the basic construction of the weather generator. The soil database covers various parameters, such as soil type, texture, structure, nutrient content, organic matter content, etc., enabling the SWAT model to better simulate hydrological and pollutant transport processes in the soil. Additionally, point-source input data, including various industrial and domestic wastewater discharge situations, were collected and organized to improve data quality. Furthermore, a series of agricultural management measures were developed based on the use of fertilizers and pesticides for simulation, providing an important basis for analyzing non-point-source pollution using the SWAT model. By comparing the different results of the simulation using optimized databases, it is shown that the above work improved the simulation accuracy of the SWAT model in predicting runoff and pollution load in Heilongjiang, China. The NSE of runoff simulation increased from 0.923 to 0.988, and the NSE of ammonia nitrogen and CBOD simulation increased from 0.852 and 0.758 to 0.930 and 0.902, respectively. It is expected that these efforts will provide strong data support for subsequent research and provide a theoretical basis for government decision-makers to build scientifically rigorous and effective pollution control strategies.

1. Introduction

Non-point-source pollution, as a significant source of water pollution, has serious impacts on ecosystems and human health [1]. It is characterized by a wide distribution, complex causes, and numerous influencing factors, making its prediction and control extremely challenging [2,3]. Since the 1990s, the mainstream approach for predicting non-point-source pollution has been using mathematical models, remote sensing technology, and geographic information systems [4]. Among these models, the Soil and Water Assessment Tool (SWAT) stands out as a prominent representative. It is a widely distributed hydrological model [5,6] with powerful capabilities for simulating non-point-source pollution, capable of simulating hydrological and pollutant transport processes at the watershed scale [7]. The SWAT model has been widely recognized and adopted in the international water resource and environmental fields. The significance of the SWAT model in predicting non-point-source pollution manifests in several main aspects: first, simulating the relationship between soil and water quality [8]; second, accurately predicting non-point-source pollution [6]; and third, predicting the impact of management measures [9,10]. Additionally, the SWAT model is particularly useful as a decision-making tool [11], as it can assess the potential impact of land use, land management, and climate change on river ecosystems through scenario simulation [12,13], and can be used to identify the best management practices for watersheds under different hydrological, climatic, and other parameter conditions [14].
The SWAT model is an important tool for estimating the release intensity of riverside pollutants and non-point-source pollutant loads, and evaluating the utilization of watershed water resources [15]. It can be applied across different spatial and temporal scales, with excellent simulation accuracy [16,17]. However, the application of the SWAT model is often unsatisfactory in most regions (especially remote areas), mainly due to the prediction accuracy of non-point-source pollution, which depends on the completeness of input model data and parameters and the degree to which the model reproduces watershed characteristics [18]. Yet, the limited availability and difficulty in obtaining basic data (including flow data, soil data, weather data, etc.) in remote areas are major obstacles to model calibration and validation [19], making it difficult for the available data to meet the requirements for constructing the SWAT model. Moreover, the precipitation data of CFSR_World used by many scholars is calculated by numerical forecasting models based on the basic real circulation field, which has lower accuracy compared to the measured precipitation at ground rainfall stations [20]. On a monthly scale, CFSR precipitation data overestimates weak precipitation and underestimates heavy precipitation; it is also impossible to accurately capture short-term climate variability. This has led to limitations in the application of SWAT in some regions.
In order to accurately predict non-point-source pollution in the Heilongjiang region of China, formulate effective pollution prevention and control strategies, and ensure ecological environment safety and human health [21], this study focuses on the Heilongjiang region. It aims to systematically optimize the reliability and optimization of the SWAT model in predicting runoff and dynamically simulating non-point-source pollution, with the purpose of assessing the impact of different soil attribute databases and weather generators on river flow and pollutant concentration when used as SWAT inputs. The study compares simulations using the Food and Agriculture Organization of the United Nations soil database (HWSD_Soil) and a self-built soil attribute database for Heilongjiang soil data (SOIL_HLJ), as well as the simulation accuracy of the CFSR_World weather generator model and a weather generator developed for Heilongjiang meteorological data (WGEN_HLJ). We also conduct a comprehensive evaluation of the simulation accuracy of the model in terms of runoff, ammonia nitrogen, and CBOD under various scenarios. Additionally, point-source input data, including various industrial and domestic wastewater discharge situations, are collected and organized to improve data quality. Furthermore, a series of agricultural management measures are developed based on the use of fertilizers and pesticides for simulation, providing an important basis for analyzing non-point-source pollution by the SWAT model. We expect that through this exploration, the analytical ability of the local SWAT model for hydrological cycles in cold regions of China and the accuracy of river water quality reproduction will be significantly enhanced, thereby providing strong data support and a theoretical basis for environmental protection departments to construct and implement scientifically rigorous and effective pollution prevention and control strategies.

2. Materials and Methods

2.1. Study Area

The study area (45.52–46.62° N and 126.47–128.86° E) is located in the northeastern region of China, specifically in the southern part of Heilongjiang Province (Figure 1c), with a total area of 14,151 square kilometers. The study area traverses five counties, namely Hulan District (western), Bayan County (northwestern), Bin County (southern), Mulan County (central), and Tonghe County (eastern). Situated in the middle and upper reaches of the Songhua River, which has a total length of 1900 km and two tributaries, the Nen River originates from the Greater Khingan Mountains, and the second tributary, the southern source of the Songhua River, originates from Tianchi of Changbai Mountain. They converge in Fuyu to form the main stream of the Songhua River. There are many hydrological stations along the Songhua River in the study area, including Dadingshan Station, Baiduzhen Station, and Tonghe Station (Figure 1d), providing sufficient hydrological and water quality information for the study. The study area has a temperate humid and semi-humid continental monsoon climate, with four distinct seasons. The average monthly maximum temperature is about 28 °C, the average monthly minimum temperature is about −23 °C, and the average annual precipitation is about 500 mm, with most rainfall occurring in July and August [22]. The river water is frozen in winter. Approximately 51.3% of the land in the study area is used for cultivated land, 31.9% is forest land, 8.8% is grassland, 4.3% is wetland, 2.2% is residential land, and the remaining 1.5% is water bodies (Figure 2b).

2.2. Data Sources

The scope of this study is extensive, covering a large area, thus requiring the collection and utilization of a large amount of data to meet research needs. We have conducted comprehensive data collection work and ensured the reliability of the data sources. Detailed information on data usage and sources is presented in Table 1. Data used for estimating planting areas and fertilizer application measures mainly come from provincial and municipal yearbooks; data on human and livestock numbers and industrial production are mainly referenced from the results of the Second National Pollution Source Census of China; climate data, including rainfall, temperature, wind direction, wind speed, and sunshine-induced evaporation, can be obtained from the official website of the China Meteorological Administration (http://data.cma.cn (accessed on 12 December 2018)); digital elevation maps and land use data can be found on the China National Earth System Science Data Center website (http://www.geodata.cn/ (accessed on 26 December 2025)); hydrological and water quality data come from local environmental science institutes and water conservancy departments, respectively; and soil attribute data can be queried through the official website of the Food and Agriculture Organization (FAO) of the United Nations (https://www.fao.org/ (accessed on 12 December 2018)). As for other related data, such as atmospheric nitrogen deposition, they can be obtained from the published literature.

2.3. SWAT Model Software

The SWAT model, developed by the United States Department of Agriculture (USDA), is a model for calculating non-point-source pollution loads applicable to large watershed scales [23,24]. Compared to traditional monitoring methods, it has higher efficiency and accuracy [25]. It is a physically distributed model capable of simulating non-point-source pollution over continuous time periods. The model uses a large amount of basic data as input, including meteorological, soil, topographical, vegetation, and land management information, to simulate processes such as water movement, sediment transport, plant growth, and nutrient cycling [26,27]. When integrated with GIS, SWAT can better handle spatial variability [28], dividing the watershed into sub-basins for analysis, and ultimately calculating the water and sediment yield and nutrient content at the watershed outlet.
SWAT software (SWAT2012) requires a series of basic data inputs, including topographic data, meteorological data, soil attribute libraries, land use data, point-source data, station observation data, and management measure settings [29]. Among them, meteorological data such as rainfall, average temperature, and solar radiation have important impacts on hydrological processes, crop growth, and nutrient degradation and transformation. Soil properties determine the movement of water and substances in the soil, and both play important roles in the water and substance cycling in hydrological response units (HRUs). Hence, in this study, we focus on the environmental conditions of Heilongjiang Province, using weather generator databases and soil attribute databases reconstructed for local characteristics. Additionally, point-source emissions and specific agricultural management measures will be incorporated into the simulation to analyze the improvement in SWAT model simulation accuracy.

2.4. Uncertainty Analyses

In this study, the model’s simulation accuracy is primarily assessed using the Nash-Sutcliffe efficiency (NSE) coefficient to enhance the reliability of the validation results [30]. Below are the formulas used to calculate these statistical indicators:
N S E = 1 − ∑ i = 1 n ( X o b s , i − X s i m , i ) 2 ∑ i = 1 n ( X o b s , i − X ¯ o b s ) 2
where X o b s , i represents the ith observed value, X s i m , i represents the ith simulated value, and X ¯ o b s represents the mean value of the observed values.
We separately validate the simulated values of runoff and water quality, adjusting the relevant parameters continuously using the SUFI-2 algorithm in the SWAT-CUP (SWAT-CUP2012) software [31] for calibration and completing uncertainty analysis. Each calibration uses 2000 cycles to improve the statistical significance of the simulation results and prevent any results from occurring accidentally or randomly fluctuating. Generally, for monthly scale runoff and water quality simulation, if the following condition is met, we consider the simulation results to meet the accuracy requirements and reflect the real hydrological environment: NSE > 0.5 [32].

2.5. Pollution Source Data Integration

The acquisition of pollution source data typically relies on national or regional pollution source survey projects [33,34], such as the Second National Pollution Source Census of China (referred to as “Second Pollution Census” in this study) [35]. These data cover detailed information on various types of pollution sources, including industrial, agricultural, and domestic sources [36]. When processing these data, researchers need to classify and filter the data according to specific research objectives and model application requirements. Data points located outside the study area or those insignificantly affecting the current research objectives can be reasonably excluded. Moreover, attention should be paid to data quality issues, and appropriate methods should be adopted for data cleaning and preprocessing, addressing problems such as missing values, outliers, and coordinate loss.
In the application of the SWAT model, pollution sources with large emissions, clear geographical locations, and significant impacts on water environmental quality are designated as point sources, such as direct discharge outlets of industrial wastewater and effluent outlets of urban sewage treatment plants. Additionally, considering that accurate geographic coordinates of fecal sewage from livestock and poultry farming are available, they are also included in the category of point sources for consideration. The SWAT point-source database typically includes information such as types of pollutants, emission amounts, and the time and geographical locations of emissions. For diffuse pollution sources with dispersed emissions and difficult-to-locate specific emission locations, such as surface runoff and residual fertilizers and pesticides in farmland, the SWAT model cannot directly simulate by adding point sources, and other simulation methods and technical means are required. It is important to note that when using point-source data for simulation, factors such as the temporal variability and spatial distribution characteristics of point-source emissions and their interactions with other pollution sources should be fully considered to improve the model’s predictive accuracy.

2.6. Database Reconstruction

In this study, we produced targeted reconstructions of the basic input data for SWAT in the Heilongjiang region of China, mainly covering the following core aspects:
(1)
Weather generator database: SWAT has an embedded WXGEN weather generator [37] that can generate meteorological data or supplement missing data, which can meet the simulation requirements for most areas in the United States. Currently, most scholars use the CFSR_World weather generator to create simulated meteorological data. It is based on the Climate Forecast System Reanalysis (CFSR) global meteorological dataset, which is a high-resolution meteorological reanalysis dataset provided by the National Oceanic and Atmospheric Administration (NOAA) of the United States. If constructing other weather generators, various calculation formulas must be used based on the observation data of meteorological stations to generate multi-year monthly average characteristics of various meteorological parameters. In this study, the China Surface Climate Daily Dataset, Version 3.0 (1951–2020), was used to construct a weather generator (WGEN_HLJ) suitable for the Heilongjiang region of China. The specific technical roadmap is shown in Figure 3.
(2)
Soil database: In calculating parameters such as soil available water capacity and saturated hydraulic conductivity, we utilized the SPAW soil characteristic calculation formula [38]. This formula, based on extensive soil science theory and abundant experimental data, can accurately calculate various soil properties and substance contents based on different soil types and environmental conditions. In this study, we used this formula to construct the soil attribute database SOIL_HLJ suitable for the Heilongjiang region of China. The specific technical roadmap for constructing the database is shown in Figure 4.
(3)
Agricultural management measures: We reviewed historical data from the region over the past few years and actively sought opinions from relevant experts in the field and local farmers to extract the main types of crops and corresponding fertilization measures in the Heilongjiang region. Relevant collected data were incorporated into the model to calibrate it, improving its ability to simulate real-world scenarios (Table 2).

3. Results and Discussion

3.1. Hydrological Regime and Its Characteristics

After an in-depth study of the trends in runoff and water quality at the Tonghe station in the study area from 2014 to 2017, we obtained the trajectory of its changes, as shown in Figure 5b. The location of Tonghe Station within the study area is shown in Figure 5a.
Based on the runoff records at the Tonghe hydrological station from 2014 to 2017, the middle reaches of the Heilongjiang River exhibit a typical cold-temperate bimodal hydrological regime (Figure 5b). In spring (March–May), runoff gradually increases with rising air temperature and snowmelt, forming a distinct spring freshet around May, with peak discharges generally of approximately 1500 m3 s−1. This meltwater-dominated process corresponds to the so-called “peach blossom flood.” Notably, the spring flood in 2014 was exceptionally pronounced, with peak discharge exceeding 3000 m3 s−1, indicating a large snowpack and/or accelerated snowmelt rates that triggered a strong runoff response. During summer (June–August), runoff rises rapidly to annual maxima under the influence of concentrated heavy rainfall, with peak discharges during the main flood season typically ranging from 2200 to 2500 m3 s−1; in 2014, the peak further increased to 3220 m3 s−1, reflecting the superimposed effects of extreme precipitation events. This pattern is closely associated with the climatic characteristic of the Heilongjiang Basin, where approximately 50% of the annual precipitation occurs during July–August. In autumn (September–October), decreasing precipitation leads to recession of runoff and a transition toward low-flow conditions. In winter (November to the following March), river ice cover develops, surface runoff is largely suppressed, and discharge is mainly sustained by upstream baseflow, remaining relatively stable at low levels of about 200–300 m3 s−1. Overall, despite interannual variability—particularly the anomalously wet conditions in 2014 associated with climatic extremes—the runoff processes over the four years constantly follow a characteristic pattern of “spring freshet–summer flood–winter low flow.” This reflects the continental hydrological features of the Heilongjiang River Basin, jointly regulated by air temperature, precipitation, and ice conditions. Such hydrological seasonality highlights the high sensitivity of cold-region rivers to climate variability and provides a critical scientific basis for regional water resource management and flood control operations.

3.2. Water Quality Concentration and Runoff Dynamics

The above section described the spatiotemporal dynamics of watershed water volume, and the following will further discuss the water quality status within the watershed. The study will combine observations of water quality data to explore the coupling relationship between concentration changes and hydrodynamics.
Figure 6 illustrates the observed variations in concentrations and pollutant fluxes of ammonia nitrogen (NH3-N) and carbonaceous biochemical oxygen demand (CBOD) at the Tonghe hydrological station from 2014 to 2017. In contrast to the typical bimodal hydrological regime of runoff, the temporal dynamics of water quality parameters exhibit pronounced non-periodicity and greater complexity, reflecting the coupled influences of pollutant source inputs, hydrological conditions, and in-stream self-purification capacity.
From a concentration perspective, NH3-N remained persistently elevated during January–March each year, reaching a peak of 2.3 mg L−1 in January 2015, whereas summer concentrations were relatively stable at 0.5–1.0 mg L−1. This pattern indicates that under low-temperature and low-flow conditions, reduced dilution and biodegradation capacity promotes the accumulation of ammonia nitrogen during winter. In comparison, CBOD concentrations showed much larger fluctuations (1.0–3.0 mg L−1), with multiple episodes exceeding 3.0 mg L−1 during January–April and October of 2015–2016, suggesting a high sensitivity to short-term anthropogenic disturbances, such as domestic wastewater overflows or agricultural non-point-source runoff. Notably, although September–October corresponds to the peak period of autumn fertilization in the Heilongjiang region, synchronous increases in NH3-N and CBOD concentrations were not observed. This implies a pronounced lag effect in the transport of fertilizer-derived nitrogen and organic matter into surface waters, likely mediated by soil leaching and groundwater pathways before becoming detectable in the river system.
In contrast, pollutant fluxes (concentration × discharge) were predominantly controlled by hydrological forcing. During the spring freshet in May 2014, despite a moderate NH3-N concentration of approximately 1.2 mg L−1, the sharp increase in discharge to over 3000 m3 s−1 resulted in a flux of up to 8 × 106 g d−1, representing the annual maximum. Similarly, CBOD fluxes peaked at 1.4 × 107 g d−1 in May 2014 and 1.2 × 107 g d−1 in July 2015, both coinciding with snowmelt- or rainfall-driven high-flow events. These results indicate that although high-flow conditions may dilute pollutant concentrations, they can substantially amplify actual mass exports, giving rise to a “high-flux–low-concentration” phenomenon. Conversely, during low-flow and ice-covered periods (November to the following March), pollutant fluxes remained constantly below 2 × 106 g d−1, despite occasionally elevated concentrations, due to discharges of generally lower than 300 m3 s−1. This highlights the dominant role of hydrological conditions in regulating pollutant transport. Overall, NH3-N dynamics are characterized mainly by seasonal accumulation, whereas CBOD is more susceptible to episodic pollution inputs. The flux maxima of both pollutants are concentrated during spring and summer high-flow periods, underscoring the need for cold-region river management strategies to move beyond static concentration-based assessments and to place greater emphasis on pollutant export risks under extreme hydrological events, while explicitly accounting for the lagged response of agricultural non-point-source pollution.

3.3. Sensitivity of Runoff and Water Quality to Meteorological and Point Sources

To quantify the impact of different inputs on runoff and water quality simulation results, this study will evaluate the influence of weather generator database on runoff processes; assess the impact of soil attribute databases on hydrological and water quality simulations; organize and refine the point-source emission inventory and test the sensitivity of human activities to water quality prediction; and explore the impact of non-point sources such as agricultural management measures on water quality.

3.3.1. Influence of Weather Generator Database on Runoff Simulation Accuracy

To evaluate the impact of different weather generators on runoff simulation accuracy, this study conducted two comparative experiments: one driven by the self-developed weather generator WGEN_HLJ, and the other by CMADS (China Meteorological Assimilation Driving Dataset). Except for the meteorological input data, all other model structures, parameters, and boundary conditions were kept constant to ensure comparability of the results.
As shown in Figure 7, runoff simulated using WGEN_HLJ closely matches the observed values, outperforming the CMADS-driven simulations. The NSE increased from 0.923 under the CMADS scenario to 0.985 with WGEN_HLJ, indicating a significant enhancement in model performance. The improvement in the NSE is mainly due to the higher accuracy of the model in simulating key hydrological events such as flood peaks and dry seasons, especially in capturing high-flow peaks. During high-flow events such as heavy rainfall in May 2014 and May 2017, WGEN-HLJ is able to more accurately capture the height and timing of flood peaks, while simulations driven by CMADS have a certain degree of underestimation. This difference reflects the limitations of global data in characterizing extreme precipitation and its hydrological response in some regions.
However, the simulation results of WGEN-HLJ do not perfectly match the observed values throughout the entire time period. For example, there is still a gap between peak and observed values, such as the slight overestimation of runoff in July 2014 and July 2016, which may be related to the high sensitivity of the rainfall-runoff response. However, the simulated values in September 2015 and September 2017 were both lower; there is a slight tendency for underestimation of flow during the spring snowmelt period (April), which may be due to uncertainties in other parameters such as snow melting rate. Although these local biases did not significantly affect the overall statistical indicators, they suggest that in the future, the performance of the model under specific conditions can be further improved by optimizing modules or introducing other conditions.
In summary, WGEN_HLJ demonstrates higher accuracy and reliability in runoff simulation, particularly under extreme high-flow conditions. Its refined representation of regional climatic characteristics effectively improves hydrological model performance under complex meteorological scenarios, providing a more robust meteorological forcing basis for applications such as watershed flood forecasting and water resource management.

3.3.2. Influence of Soil Attribute Database on Runoff and Water Quality Simulation Accuracy

To assess the influence of different soil property databases on the accuracy of runoff and water quality simulations, a comparative experiment was conducted in this study. One scenario employed a self-developed soil database (SOIL_HLJ), while the other used the Harmonized World Soil Database (HWSD). Except for the soil input data, all model structures, parameters, and boundary conditions were kept identical to ensure the comparability of the results.
For runoff simulation (Figure 8), the use of SOIL_HLJ increased the NSE from 0.923 under the HWSD-based scenario to 0.945, indicating higher accuracy in simulating runoff generation and routing processes. Time-series comparisons show that the improvement in the NSE coefficient is mainly due to the ability to more accurately reproduce the magnitude and occurrence time of flood peaks in multiple high-flow events. In contrast, the simulation results of the HWSD driver are generally lower at the peak. In contrast, HWSD-driven simulations generally underestimate peak discharges or exhibit phase lag, reflecting limitations in representing regional soil water retention and hydraulic conductivity characteristics. These results demonstrate that, owing to its refined characterization of local soil physical properties, SOIL_HLJ exhibits clear advantages under high-runoff conditions.
In contrast, improvements in water quality simulation are relatively limited (Figure 9). Although the NSE for simulated CBOD concentrations increased from 0.758 with HWSD to 0.812 with SOIL_HLJ, the improvement (0.054) is modest. Both scenarios broadly capture the seasonal variation in CBOD; however, notable discrepancies remain in simulating peak concentrations, decay rates, and low-concentration periods. While SOIL_HLJ performs slightly better than HWSD, the overall enhancement is not substantial. This can be attributed to several factors: (i) water quality processes are jointly controlled by pollutant source loading, in-stream hydrodynamics, and biogeochemical reactions, with soil properties acting only as indirect drivers; (ii) soil parameters in SOIL_HLJ that are relevant to pollutant transport (e.g., organic matter content, cation exchange capacity, and adsorption–desorption characteristics) have not yet been specifically optimized for the water quality modules; and (iii) simplified representations of non-point-source pollutant transport and transformation in the model may constrain the potential benefits derived from improved soil data.
Overall, the SOIL_HLJ database demonstrates pronounced superiority in runoff simulation, particularly under extreme high-flow events, whereas its contribution to water quality simulation, although positive, remains limited. These findings indicate that soil data exert a much stronger direct effect on hydrological processes than on the indirect regulation of water quality dynamics. Future work should focus on establishing explicit linkages between soil properties and pollutant-related parameters for water quality modeling, integrating field monitoring data to calibrate key attributes, and developing more refined non-point-source pollution modules, thereby fully exploiting the potential of high-resolution regional soil databases for integrated watershed management.

3.3.3. The Impact of Point-Source Completeness on the Accuracy of Water Quality Simulation

Figure 10 and Figure 11 illustrate the impact of the completeness of point-source input on the accuracy of water quality simulation. When studying the water quality of large-flow water bodies like the main stream of the Songhua River, the influence of point-source water on the overall flow is negligible. Therefore, we focus on its potential impact on water quality. During the SWAT simulation process, we used refined point-source input data based on the second national pollution source inventory while keeping other data and parameters unchanged for comparative analysis with the observed values of ammonia nitrogen and CBOD.
Compared to the basic simulation, the systematic incorporation of point-source pollution loads—including direct industrial wastewater discharges, municipal sewage outfalls, and livestock farming effluents—significantly improved model performance. Specifically, the NSE coefficient for ammonia nitrogen increased from 0.852 to 0.930, underscoring the decisive role of point sources in nitrogen dynamics in the river system. The time series of the refined simulation exhibits markedly better agreement with observations during both high-flow periods (e.g., July 2015 and July 2016) and low-flow conditions, with simulated peaks closely matching measured values in both magnitude and timing.
Similarly, for CBOD, the NSE improved from 0.758 to 0.885 following the inclusion of point-source inputs, indicating a substantial enhancement in simulation fidelity. The overall temporal trend aligns more closely with field observations, particularly during high-concentration events in May 2015 and May 2016, where simulated CBOD concentrations accurately capture the observed peaks—highlighting the critical contribution of point sources to CBOD variability. Nevertheless, notable discrepancies persist in certain periods; most prominently, in January 2015, the simulated CBOD concentration is significantly higher than the observed value and remains deviated over an extended duration. This mismatch is likely attributable to an insufficient model spin-up period, which fails to establish representative initial water quality conditions reflective of the actual environmental background. Future work should consider extending the spin-up duration to mitigate such initial-condition-related biases and further enhance early-phase simulation accuracy. Collectively, these results demonstrate that comprehensive representation of point-source inputs substantially strengthens the model’s capacity to reproduce pollutant dynamics, thereby offering a more reliable tool for watershed-scale water quality management and decision-making.

3.3.4. The Impact of Agricultural Management Measures on Water Quality

When exploring the impact of agricultural management measures on water quality, we particularly focused on the effect of increasing inputs of pollutants such as fertilizers on ammonia nitrogen simulation. In the process of simulation using the SWAT model, local agricultural management practices were refined in the model through the Input menu, while all other datasets and parameter settings were kept unchanged, to enable a direct comparison between simulated and observed ammonia nitrogen concentrations. As shown in Figure 12, with the introduction of agricultural management measures based on field research, a substantial improvement in NH3-N simulation accuracy was observed, with the NSE increasing from the initial value of 0.852 to 0.930. Compared with the scenario without considering detailed agricultural management, the introduction of localized agricultural management parameters significantly enhances the model’s ability to simulate pollutant dynamics. This result confirms the important role of refined agricultural management databases in improving the physical authenticity and predictive reliability of SWAT models in local watersheds. These findings also emphasize the importance of agricultural management measures in water quality protection, especially in controlling and reducing the input of pollutants such as fertilizers. Further optimization of agricultural management measures and simulation parameter settings is expected to continuously improve simulation accuracy to better reflect reality, contributing to the protection of water resources and the improvement of water quality.

3.4. Implications for Prediction and Management

In this study, the SWAT model was used to simulate hydrological processes and pollution loads in the Heilongjiang region. The model input database was localized, parameter calibration and validation were completed, and hydrological and water quality characteristics were preliminarily evaluated under different databases. Based on the above results, it is possible to further explore the dominant controlling factors that affect the accuracy of SWAT runoff and water quality simulation in the study area, in order to reveal the contributions of natural and human factors to runoff and pollutants, and provide a scientific basis for regional water environment management.
As shown in Table 3, there is a significant correlation between runoff data and the weather generator database, and the association with the soil characteristic database is particularly close. However, the influence on runoff calculation results is not significant after adding point-source pollution inputs and agricultural management measures. In the simulation of runoff in large rivers such as the Songhua River, natural factors, especially rainfall and soil conditions, play a dominant role [39]. Based on this conclusion, we can reasonably infer that when predicting the future trends of runoff in major rivers, relying on future meteorological data such as rainfall (multi-scenario multi-model monthly precipitation forecast data from 2021 to 2100), we can accurately deduce their long-term dynamic changes [40].
With the self-built databases and parameter inputs mentioned above, we predicted the evolution of runoff volume for the years 2018 and 2019, as shown in Figure 13. From the figure, it can be observed that the trajectory of runoff volume changes shows some similarities each year: during the peach blossom flood period, the runoff volume slightly increases, and during the rainy season, the runoff volume significantly rises; however, after October and during the freezing period, the water volume notably decreases to its lowest level. Although we currently lack actual runoff observation data for 2018 and 2019 to compare and accurately assess the deviation of simulations using the aforementioned data and parameters, we have gained insight into some real situations through the relevant literature. For example, in July and August 2018, floods occurred in Heilongjiang Province, while the rainfall in 2019 was relatively low. Our simulation results also show that the peak runoff volume during the rainy season in 2018 was significantly higher than that in other years, while the peak in 2019 was relatively lower, which is consistent with the actual situation. In the future, if we can obtain specific runoff volume data, we will be able to more accurately assess the accuracy of simulations using our self-built databases and set parameters.
In terms of water quality, CBOD concentration is most significantly correlated with the increase in point-source inputs, followed by soil attribute databases, and exhibits a weak correlation with the weather generator formulation. Ammonia nitrogen concentration is most significantly correlated with the increase in point-source inputs, closely related to the formulation of agricultural management measures, followed by soil attribute databases, and weakly correlated with weather generator databases. This means that by scientifically and reasonably con trolling the input of anthropogenic point-source pollution, the overall flux level of pollutants in rivers can be greatly reduced. However, there is a relationship between pollutant concentration, flux, and runoff volume, so it is not wise to solely emphasize strict control of point-source inputs. Effective management of point-source pollution should be strengthened during periods of relatively low runoff volume, while during periods of high runoff volume, the river’s own environmental carrying capacity should be fully utilized for dynamic regulation and management to achieve more comprehensive and precise protection of river ecosystems.
Additionally, the connection between CBOD concentration and soil characteristics is more direct and significant, whereas the relationship between ammonia nitrogen concentration and soil characteristics appears relatively indirect. This can be explained by the fact that CBOD can migrate from soil to water bodies relatively quickly through impermeable layers or surface runoff, leading to water quality degradation in rivers. On the other hand, the concentration of ammonia nitrogen in soil largely stems from agricultural activities, especially fertilizer application. The process of ammonia nitrogen migration from soil to water bodies is usually slower and involves higher loss rates compared to CBOD, exhibiting a significant lag effect. However, human activities play a crucial role in this migration process, especially when nitrogen fertilizers are excessively applied in farmlands. This not only accelerates the rapid accumulation of nitrogen elements in the soil, but also increases the risk of surface-attached nitrogen loss with runoff. Consequently, excessive nitrogen may rapidly migrate from soil and surface runoff to water bodies, severely affecting water quality and threatening the health of aquatic ecosystems. Due to the significant impact of influent water quality on water quality simulation, we are temporarily unable to carry out water quality simulation for subsequent years (after 2017). Once we obtain specific water quality observation data, we will be able to more accurately assess the accuracy of simulations using the SOIL_HLJ database and set parameters.

4. Conclusions

This study has significantly enhanced the accuracy of the SWAT model in predicting runoff and pollutant dynamics in the cold region of Northeast China by utilizing the independently reconstructed weather generator WGEN_HLJ and soil attribute database SOIL_HLJ, optimizing the completeness of point-source data, and incorporating agricultural management measures. Specifically, the accuracy of runoff simulation has been significantly improved from 0.923 to 0.988, with the improvement in the weather generator playing a decisive role, followed closely by the influence of the soil attribute database. The accuracy of ammonia nitrogen simulation has also increased from 0.852 to 0.930, mainly attributed to the introduction of agricultural management measures. Furthermore, there has been a significant improvement in the prediction accuracy of CBOD (Chemical Oxygen Demand), rising from 0.758 to 0.902, with the completion of point-source data contributing most prominently, and the optimization of the soil attribute database also playing an important role.
The above achievements demonstrate that the comprehensive measures adopted in this study have effectively improved the simulation accuracy of the SWAT model in the cold environment of Northeast China. It is hoped that this will provide strong data support for subsequent researchers and also lay a solid theoretical foundation for policymakers to formulate scientific, rigorous, and effective environmental protection policies. A more significant benefit is that it can provide some inspiration for researchers with specific research needs or who encounter challenges in simulating accuracy when using global databases for SWAT models, and can encourage them to independently rebuild databases tailored to local conditions and establish region-specific management practices to meet the requirements of SWAT model construction and application. But when applying this method, it is necessary to clarify the following: (1) The improvement of the database depends on a relatively complete observation dataset. In areas with data scarcity, the construction of weather generators and soil databases may involve challenges. (2) The setting of agricultural management measures in this study is only applicable to typical cultivated land in the Heilongjiang region, and cannot be applied in other regions. Researchers should seek the opinions of local experts and farmers to establish local crops and management measures.
Looking to the future, the following improvement measures can be taken: (1) Studies should consider high-intensity human activity scenarios such as large-scale reservoir scheduling and groundwater extraction. (2) At the data level, it is possible to consider combining remote sensing data to achieve annual dynamic updates to land and crop types. (3) Regarding model level, the performance of improved SWAT models or other distributed models (such as MIKE, WASP, etc.) could be tested in the local area.
This study not only provides a reliable dataset for water environment management in the Heilongjiang region, but also provides a localized modeling paradigm that can be used as a reference for other remote watersheds with moderate data conditions but certain observational foundations. We encourage future researchers to independently build adapted input databases based on regional characteristics, rather than blindly relying on global products—this may be an effective path to enhance the application of SWAT in “non-standard” environments.

Author Contributions

Software, Z.Z.; Formal analysis, Z.Z.; Investigation, H.Z. and C.Y.; Data curation, Z.Z. and X.Y.; Writing—original draft, Z.Z.; Writing—review & editing, C.Y. and T.Z.; Supervision, C.Y. and T.Z.; Project administration, T.Z.; Funding acquisition, T.Z. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the National Natural Science Foundation of China [Grant No. 52500009] and the Heilongjiang Provincial Key Laboratory of Polar Environment and Ecosystem (HPKL-PEE) [Grant No. 2021010].

Data Availability Statement

The original contributions presented in this study are included in the article. Further inquiries can be directed to the corresponding authors.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Geissen, V.; Mol, H.; Klumpp, E.; Umlauf, G.; Nadal, M.; van der Ploeg, M.; van de Zee, S.E.A.T.; Ritsema, C.J. Emerging Pollutants in the Environment: A Challenge for Water Resource Management. Int. Soil Water Conserv. Res. 2015, 3, 57–65. [Google Scholar] [CrossRef]
  2. Jiang, S.; Qiu, S.; Zhou, J. Re-Examination of the Relationship between Agricultural Economic Growth and Non-Point Source Pollution in China: Evidence from the Threshold Model of Financial Development. Water 2020, 12, 2609. [Google Scholar] [CrossRef]
  3. Wang, S.; Rao, P.; Yang, D.; Tang, L. A Combination Model for Quantifying Non-Point Source Pollution Based on Land Use Type in a Typical Urbanized Area. Water 2020, 12, 729. [Google Scholar] [CrossRef]
  4. Chen, L.; Qi, X.; Zhang, X.; Li, Q.; Zhang, Y. Effect of Agricultural Land Use Changes on Soil Nutrient Use Efficiency in an Agricultural Area, Beijing, China. Chin. Geogr. Sci. 2011, 21, 392–402. [Google Scholar] [CrossRef]
  5. Ahmadi, M.; Ascough, J.C.; DeJonge, K.C.; Arabi, M. Multisite-Multivariable Sensitivity Analysis of Distributed Watershed Models: Enhancing the Perceptions from Computationally Frugal Methods. Ecol. Modell. 2014, 279, 54–67. [Google Scholar] [CrossRef]
  6. Arnold, J.G.; Srinivasan, R.; Muttiah, R.S.; Williams, J.R. Large Area Hydrologic Modeling and Assessment Part I: Model Development1. JAWRA J. Am. Water Resour. Assoc. 1998, 34, 73–89. [Google Scholar] [CrossRef]
  7. Arabi, M.; Frankenberger, J.R.; Engel, B.A.; Arnold, J.G. Representation of Agricultural Conservation Practices with SWAT. Hydrol. Process. 2008, 22, 3042–3055. [Google Scholar] [CrossRef]
  8. Neitsch, S.L.; Arnold, J.G.; Kiniry, J.R.; Williams, J.R. Soil and Water Assessment Tool: Theoretical Documentation, Version 2009; Texas Water Resources Institute Technical Report No. 406; Texas Water Resources Institute: College Station, TX, USA, 2011; pp. 1–647.
  9. Yalew, S.; van Griensven, A.; Ray, N.; Kokoszkiewicz, L.; Betrie, G.D. Distributed Computation of Large Scale SWAT Models on the Grid. Environ. Model. Softw. 2013, 41, 223–230. [Google Scholar] [CrossRef]
  10. Zhang, X.; Zhang, M. Modeling Effectiveness of Agricultural BMPs to Reduce Sediment Load and Organophosphate Pesticides in Surface Runoff. Sci. Total Environ. 2011, 409, 1949–1958. [Google Scholar] [CrossRef]
  11. Femeena, P.V.; Chaubey, I.; Aubeneau, A.; McMillan, S.K.; Wagner, P.D.; Fohrer, N. An Improved Process-Based Representation of Stream Solute Transport in the Soil and Water Assessment Tools. Hydrol. Process. 2020, 34, 2599–2611. [Google Scholar] [CrossRef]
  12. Haas, M.B.; Guse, B.; Fohrer, N. Assessing the Impacts of Best Management Practices on Nitrate Pollution in an Agricultural Dominated Lowland Catchment Considering Environmental Protection versus Economic Development. J. Environ. Manag. 2017, 196, 347–364. [Google Scholar] [CrossRef]
  13. Hoque, Y.M.; Raj, C.; Hantush, M.M.; Chaubey, I.; Govindaraju, R.S. How Do Land-Use and Climate Change Affect Watershed Health? A Scenario-Based Analysis. Water Qual. Expo. Health 2014, 6, 19–33. [Google Scholar] [CrossRef]
  14. Daniel, E.B.; Camp, J.V.; LeBoeuf, E.J.; Penrod, J.R.; Dobbins, J.P.; Abkowitz, M.D. Watershed Modeling and Its Applications; a State-of-the-Art Review. Open Hydrol. J. 2011, 5, 26–50. [Google Scholar] [CrossRef]
  15. Hoang, L.; Schneiderman, E.M.; Moore, K.E.B.; Mukundan, R.; Owens, E.M.; Steenhuis, T.S. Predicting Saturation-excess Runoff Distribution with a Lumped Hillslope Model: SWAT-HS. Hydrol. Process. 2017, 31, 2226–2243. [Google Scholar] [CrossRef]
  16. Behera, S.; Panda, R.K. Evaluation of Management Alternatives for an Agricultural Watershed in a Sub-Humid Subtropical Region Using a Physical Process Based Model. Agric. Ecosyst. Environ. 2006, 113, 62–72. [Google Scholar] [CrossRef]
  17. Lenhart, T.; Fohrer, N.; Frede, H.G. Effects of Land Use Changes on the Nutrient Balance in Mesoscale Catchments. Phys. Chem. Earth Parts A/B/C 2003, 28, 1301–1309. [Google Scholar] [CrossRef]
  18. Geza, M.; McCray, J.E. Effects of Soil Data Resolution on SWAT Model Stream Flow and Water Quality Predictions. J. Environ. Manag. 2008, 88, 393–406. [Google Scholar] [CrossRef]
  19. Ayana, E.K.; Dile, Y.T.; Narasimhan, B.; Srinivasan, R. Dividends in Flow Prediction Improvement Using High-Resolution Soil Database. J. Hydrol. Reg. Stud. 2019, 21, 159–175. [Google Scholar] [CrossRef]
  20. Kalnay, E.; Kanamitsu, M.; Kistler, R.; Collins, W.; Deaven, D.; Gandin, L.; Iredell, M.; Saha, S.; White, G.; Woollen, J.; et al. The NCEP/NCAR 40-Year Reanalysis Project. Bull. Am. Meteorol. Soc. 1996, 77, 437–471. [Google Scholar] [CrossRef]
  21. Yu, C.; Huang, X.; Chen, H.; Godfray, H.C.J.; Wright, J.S.; Hall, J.W.; Gong, P.; Ni, S.; Qiao, S.; Huang, G.; et al. Managing Nitrogen to Restore Water Quality in China. Nature 2019, 567, 516–520. [Google Scholar] [CrossRef]
  22. Gao, Y.; Tian, Y.; Zhan, W.; Li, L.; Sun, H.; Zhao, T.; Zhang, H.; Meng, Y.; Li, Y.; Liu, T.; et al. Characterizing Legacy Nitrogen-Induced Time Lags in Riverine Nitrogen Reduction for the Songhuajiang River Basin: Source Analysis, Spatio-Seasonal Patterns, and Impacts on Future Water Quality Improvement. Water Res. 2023, 242, 120292. [Google Scholar] [CrossRef]
  23. Guo, H.; Hu, Q.; Jiang, T. Annual and Seasonal Streamflow Responses to Climate and Land-Cover Changes in the Poyang Lake Basin, China. J. Hydrol. 2008, 355, 106–122. [Google Scholar] [CrossRef]
  24. Singh, A.P.; Dhadse, K.; Ahalawat, J. Managing Water Quality of a River Using an Integrated Geographically Weighted Regression Technique with Fuzzy Decision-Making Model. Environ. Monit. Assess. 2019, 191, 378. [Google Scholar] [CrossRef]
  25. Borah, D.; Bera, M. Watershed-Scale Hydrologic and Nonpoint-Source Pollution Models: Review of Mathematical Bases. Trans. ASAE 2003, 46, 1553–1566. [Google Scholar] [CrossRef]
  26. Jeyrani, F.; Morid, S.; Srinivasan, R. Assessing Basin Blue–Green Available Water Components under Different Management and Climate Scenarios Using SWAT. Agric. Water Manag. 2021, 256, 107074. [Google Scholar] [CrossRef]
  27. Ouyang, W.; Song, K.; Wang, X.; Hao, F. Non-Point Source Pollution Dynamics under Long-Term Agricultural Development and Relationship with Landscape Dynamics. Ecol. Indic. 2014, 45, 579–589. [Google Scholar] [CrossRef]
  28. Arnold, J.G.; Fohrer, N. SWAT2000: Current Capabilities and Research Opportunities in Applied Watershed Modelling. Hydrol. Process. 2005, 19, 563–572. [Google Scholar] [CrossRef]
  29. Afonso De Oliveira Serrão, E.; Silva, M.T.; Ferreira, T.R.; Paiva De Ataide, L.C.; Assis Dos Santos, C.; Meiguins De Lima, A.M.; de Paulo Rodrigues Da Silva, V.; de Assis Salviano De Sousa, F.; Cardoso Gomes, D.J. Impacts of Land Use and Land Cover Changes on Hydrological Processes and Sediment Yield Determined Using the SWAT Model. Int. J. Sediment Res. 2022, 37, 54–69. [Google Scholar] [CrossRef]
  30. Shokrana, M.S.B.; Ghane, E.; Qi, Z. Calibration and Validation of RZWQM2-P Model to Simulate Phosphorus Loss in a Clay Loam Soil in Michigan. J. ASABE 2023, 66, 1–12. [Google Scholar] [CrossRef]
  31. Yang, J.; Reichert, P.; Abbaspour, K.C.; Xia, J.; Yang, H. Comparing Uncertainty Analysis Techniques for a SWAT Application to the Chaohe Basin in China. J. Hydrol. 2008, 358, 1–23. [Google Scholar] [CrossRef]
  32. Moriasi, D.N. Model Evaluation Guidelines for Systematic Quantification of Accuracy in Watershed Simulations. Trans. ASABE 2007, 50, 885–900. [Google Scholar] [CrossRef]
  33. MEE; NBS; MOA Bulletin on the First National General Survey of Pollution Sources (Fiscal Year:2007) 2010, 2021. Available online: https://www.mee.gov.cn/gkml/hbb/bgg/201002/t20100210_185698.htm (accessed on 12 December 2018).
  34. NBS National Statistical Yearbook (2000–2019) 2020, 2021. Available online: https://www.stats.gov.cn/sj/ndsj/ (accessed on 12 December 2018).
  35. MEE; NBS; MOA Bulletin on the Second National General Survey of Pollution Sources (Fiscal Year: 2017) 2020, 2021. Available online: https://www.mee.gov.cn/xxgk2018/xxgk/xxgk01/202006/t20200610_783547.html (accessed on 12 December 2018).
  36. Xie, M.; Duan, H.; Kang, P.; Qiao, Q.; Bai, L. Toward an Ecological Civilization: China’s Progress as Documented by the Second National General Survey of Pollution Sources. Engineering 2021, 7, 1336–1341. [Google Scholar] [CrossRef]
  37. Williams, J.R.; Dyke, P.T.; Jones, C.A. Epic—A Model for Assessing the Effects of Erosion on Soil Productivity1 1Paper Presented at the Third International Conference on State-of-the-Art in Ecological Modelling, Colorado State University, May 24–28, 1982. In Analysis of Ecological Systems; Lauenroth, W.K., Skogerboe, G.V., Flug, M., Eds.; Developments in Environmental Modelling; Elsevier: Amsterdam, The Netherlands, 1983; Volume 5, pp. 553–572. [Google Scholar]
  38. Saxton, K.E.; Rawls, W.J.; Romberger, J.S.; Papendick, R.I. Estimating Generalized Soil–Water Characteristics from Texture. Soil Sci. Soc. Am. J. 1986, 50, 1031–1036. [Google Scholar] [CrossRef]
  39. Xu, Z.X.; Zhao, F.F.; Li, J.Y. Response of Streamflow to Climate Change in the Headwater Catchment of the Yellow River Basin. Quat. Int. 2009, 208, 62–75. [Google Scholar] [CrossRef]
  40. Hu, N.; Shu, K.; Zhang, Y.; Alfonso, L.; Li, C.; Zheng, T. Prediction of Natural Runoff in China Based on Multi-Scenario Climate Models with Self-Attention Neural Networks. Water Res. 2025, 282, 123768. [Google Scholar] [CrossRef]
Water 18 00389 g001aWater 18 00389 g001b
Figure 1. Research area. (a) World map, with China highlighted; (b) map of china, with Heilongjiang province highlighted; (c) research area with meteorological stations, rivers, and watersheds indicated; (d) research area with rivers, divisions, and meteorological stations indicated.
Figure 1. Research area. (a) World map, with China highlighted; (b) map of china, with Heilongjiang province highlighted; (c) research area with meteorological stations, rivers, and watersheds indicated; (d) research area with rivers, divisions, and meteorological stations indicated.
Water 18 00389 g001aWater 18 00389 g001b
Water 18 00389 g002
Figure 2. (a) DEM. (b) Land use (AGRL: agricultural; FRST: forest—mixed; RNGE: range—grasses; WETL: wetlands—mixed; WATR: water; URBN: urban). (c) Soil (self-built databases: SOIL_HLJ). (d) Slope ranges.
Figure 2. (a) DEM. (b) Land use (AGRL: agricultural; FRST: forest—mixed; RNGE: range—grasses; WETL: wetlands—mixed; WATR: water; URBN: urban). (c) Soil (self-built databases: SOIL_HLJ). (d) Slope ranges.
Water 18 00389 g002
Water 18 00389 g003
Figure 3. Technical roadmap for weather generator construction.
Figure 3. Technical roadmap for weather generator construction.
Water 18 00389 g003
Water 18 00389 g004
Figure 4. Technical roadmap for soil attribute database construction.
Figure 4. Technical roadmap for soil attribute database construction.
Water 18 00389 g004
Water 18 00389 g005
Figure 5. (a) The location of Tonghe station; (b) runoff observations at Tonghe Station.
Figure 5. (a) The location of Tonghe station; (b) runoff observations at Tonghe Station.
Water 18 00389 g005
Water 18 00389 g006
Figure 6. Observation values of water quality concentration and pollutant flux at Tonghe Station. (a) ammonia nitrogen concentration observations, (b) CBOD concentration observations, (c) ammonia nitrogen pollutant flux observations and (d) CBOD pollutant flux observations.
Figure 6. Observation values of water quality concentration and pollutant flux at Tonghe Station. (a) ammonia nitrogen concentration observations, (b) CBOD concentration observations, (c) ammonia nitrogen pollutant flux observations and (d) CBOD pollutant flux observations.
Water 18 00389 g006
Water 18 00389 g007
Figure 7. Comparison between observed and simulated runoff values at Tonghe Station (using WGEN_HLJ).
Figure 7. Comparison between observed and simulated runoff values at Tonghe Station (using WGEN_HLJ).
Water 18 00389 g007
Water 18 00389 g008
Figure 8. Comparison between observed and simulated runoff values at Tonghe Station (using soil database).
Figure 8. Comparison between observed and simulated runoff values at Tonghe Station (using soil database).
Water 18 00389 g008
Water 18 00389 g009
Figure 9. Comparison between observed and simulated CBOD values at Tonghe Station (using soil database).
Figure 9. Comparison between observed and simulated CBOD values at Tonghe Station (using soil database).
Water 18 00389 g009
Water 18 00389 g010
Figure 10. Comparison between observed and simulated ammonia nitrogen values at Tonghe Station (with point sources added).
Figure 10. Comparison between observed and simulated ammonia nitrogen values at Tonghe Station (with point sources added).
Water 18 00389 g010
Water 18 00389 g011
Figure 11. Comparison between observed and simulated CBOD values at Tonghe Station (with point sources added).
Figure 11. Comparison between observed and simulated CBOD values at Tonghe Station (with point sources added).
Water 18 00389 g011
Water 18 00389 g012
Figure 12. Comparison between observed and simulated ammonia nitrogen flux values at Tonghe Station (with agricultural facilities added).
Figure 12. Comparison between observed and simulated ammonia nitrogen flux values at Tonghe Station (with agricultural facilities added).
Water 18 00389 g012
Water 18 00389 g013
Figure 13. Comparison between observed and simulated runoff at Tonghe Station (all measures).
Figure 13. Comparison between observed and simulated runoff at Tonghe Station (all measures).
Water 18 00389 g013
Table 1. List of basic data.
Table 1. List of basic data.
Dataset NameData TypeData SourceData Collection DateSpatial Resolution
ASTER GDEM V2Digital Elevation ModelGeographic Spatial Data Cloud201530 m
Global Land30Land Use DataNational Geographic Information Resource Catalog Service System201030 m
HWSD_SoilSoil DataFood and Agriculture Organization (FAO)20131 km
Hydrological Data of Heilongjiang BasinHydrological DataChina Hydrological Yearbook2008–2016
River Network Data of Heilongjiang ProvinceRiver Network DataHeilongjiang Province Water Resources and Hydropower Design Institute2017
Water Quality Monitoring Data of Heilongjiang ProvinceMonitoring DataHeilongjiang Province Water Resources and Hydropower Design Institute2014–2018
Monitoring Sections of the 14th Five-Year Plan RiversMonitoring SectionsHeilongjiang Provincial Monitoring Center Station2019
Statistical Yearbook of Heilongjiang ProvinceStatistical DataHeilongjiang Provincial Bureau of Statistics2018
Water Function Classification of Heilongjiang ProvinceVector DataHeilongjiang Province Water Resources and Hydropower Design Institute2017
Second Pollution Census DataCensus DataHeilongjiang Provincial Department of Ecology and Environment2017
China Surface Meteorological Data Daily Value Dataset (V3.0)Meteorological DataChina Meteorological Administration website2020
Table 2. Main crop types and corresponding fertilization measures.
Table 2. Main crop types and corresponding fertilization measures.
Crop TypesFertilizerFertilization TimeFertilizer DepthFertilizer Quality of N (kg/ha)Fertilizer Quality of P (kg/ha)Fertilizer Quality of K (kg/ha)
SoybeanBase30 April5 cm343
Non-Additional\\\\
RiceBase30 April10 cm554
Additional30 May300
CornBase30 April5 cm452
Additional10 May502
Table 3. Comparison table of simulation accuracy.
Table 3. Comparison table of simulation accuracy.
NSEInitial ConditionsSelf-Built Weather Generator DatabaseSoil Physicochemical Property DatabasePoint SourceAgricultural Management MeasuresUsing All Measures
Runoff0.9230.9850.9450.9230.9230.988
NH3-N0.852 0.8980.9300.9300.930
CBOD0.758 0.8120.8850.7580.902
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Zhang, Z.; Zhang, H.; Yu, X.; Yang, C.; Zheng, T. A Novel Weather Generator and Soil Attribute Database for SWAT to Improve the Simulation Accuracy in the Heilongjiang Region of China. Water 2026, 18, 389. https://doi.org/10.3390/w18030389

AMA Style

Zhang Z, Zhang H, Yu X, Yang C, Zheng T. A Novel Weather Generator and Soil Attribute Database for SWAT to Improve the Simulation Accuracy in the Heilongjiang Region of China. Water. 2026; 18(3):389. https://doi.org/10.3390/w18030389

Chicago/Turabian Style

Zhang, Zhihao, Haorui Zhang, Xiaoying Yu, Chunyan Yang, and Tong Zheng. 2026. "A Novel Weather Generator and Soil Attribute Database for SWAT to Improve the Simulation Accuracy in the Heilongjiang Region of China" Water 18, no. 3: 389. https://doi.org/10.3390/w18030389

APA Style

Zhang, Z., Zhang, H., Yu, X., Yang, C., & Zheng, T. (2026). A Novel Weather Generator and Soil Attribute Database for SWAT to Improve the Simulation Accuracy in the Heilongjiang Region of China. Water, 18(3), 389. https://doi.org/10.3390/w18030389

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

Article Metrics

Back to TopTop