Multi-scale coupled drainage basin pollution step-by-step traceability method, device, equipment and medium

Abstract

The invention relates to the technical field of pollution traceability, and discloses a multi-scale coupled drainage basin pollution step-by-step traceability method, device and equipment and a medium, and the method comprises the steps: obtaining an initial simulation result of each sub-drainage basin outputted by a target SWAT model; determining a key sub-basin based on the initial simulation result; calling an APEX model to simulate by taking the key simulation result corresponding to the key sub-basin as a boundary condition to obtain an intermediate simulation result of the key sub-basin; performing grid unit division on the sub-region of the intermediate simulation result; determining a grid environment index of each grid unit; taking the initial simulation result and the intermediate simulation result as dependent variables, taking the grid environment index as an independent variable, and constructing a quantitative relationship between the grid environment index and the pollution load capacity; constructing a spatial distribution map based on the quantitative relationship; and tracing the target pollution load capacity of the target grid unit based on the spatial distribution diagram. According to the scheme, step-by-step traceability from the drainage basin to the sub-drainage basin and then to the grid can be realized, the problem of traceability intermediate fault in the prior art is solved, and the traceability accuracy is improved.

Description

Technical Field

[0001] This invention belongs to the field of pollution source tracing technology, specifically involving a multi-scale coupled method, device, equipment, and medium for tracing watershed pollution sources step by step. Background Technology

[0002] Non-point source pollution in watersheds is characterized by its complex causes, strong spatiotemporal variability, and difficulty in precise location, making it a key focus and challenge in current water environment governance. The SWAT (Soil and Water Assessment Tool) model, a widely used watershed hydrological and water quality simulation tool, can effectively simulate the generation and migration of pollutants (such as total nitrogen (TN) and total phosphorus (TP)) under different land uses. However, its output results are based on sub-watersheds, resulting in low spatial resolution (typically tens to hundreds of square kilometers), making it impossible to accurately locate pollution sources and hindering precise governance.

[0003] Although parameter sensitivity analysis, calibration, and verification can be performed using tools such as SWAT-CUP (e.g., using the coefficient of determination R), 2 While using the Nash efficiency coefficient (NSE) as an evaluation index to improve the model's simulation accuracy at the watershed outlet section, the problem of insufficient spatial resolution within the model persists. Existing methods to improve resolution, such as increasing the number of sub-watersheds, significantly increase the computational burden and uncertainty; or simply overlaying model results with remote sensing images, lack mechanistic and quantitative support. The SWAT model performs well in macro-scale simulations, but its accuracy is low in micro-scale simulations, especially in sub-watersheds with intense human activity and concentrated pollution, where uniform parameters fail to reflect the complex internal mechanisms.

[0004] While more refined field / sub-watershed scale models like APEX (Agricultural Policy / Environmental eXtender Model) exist, the increasing demand for precise source tracing and refined management under the current policies of integrated watershed management and green agricultural development presents challenges to existing methods in terms of spatial resolution and application accuracy. Therefore, there is an urgent need to develop a method that, while ensuring the accuracy of watershed simulation, downscales pollution loads from sub-watersheds to a finer spatial grid (e.g., 1km × 1km), enabling high-resolution identification and source tracing of non-point source pollution hotspots. This would provide scientific support for watershed-specific management and precise control.

[0005] Existing watershed non-point source pollution tracing technologies based on models such as SWAT mainly suffer from the following problems: (1) Low spatial resolution and insufficient source tracing accuracy: The output of the SWAT model is based on the average value of the "sub-basin" and cannot distinguish the differences in pollution contribution of different plots within a sub-basin. As a result, the treatment measures can only be implemented for the entire sub-basin and cannot accurately locate the most polluted "hot spots", which leads to high treatment costs and low treatment efficiency.

[0006] (2) The existence of "scale discontinuity" and lack of mesoscale verification: Existing technologies often jump directly from macroscopic SWAT sub-basins (tens of square kilometers) to microscopic site analysis, lacking the support of "mesoscale" models that bridge the gap. At the same time, although site analysis has local accuracy, it is difficult to achieve continuous spatial coverage. As a result, the final downscaling analysis, especially in key areas with complex pollution mechanisms, has a weak physical basis and the reliability of the results is limited.

[0007] (3) Insufficient combination of mechanism and experience: Traditional hydrological and water quality models focus on mechanism processes, but when refined to a small scale, data acquisition is difficult; while spatial interpolation methods based solely on topography or land use can provide high-resolution results, but they lack physical descriptions of hydrological and water quality processes, which may lead to insufficient physical rationality of the results.

[0008] (4) Weak ability to trace sources: Due to the limitation of spatial resolution, existing technologies are unable to accurately identify specific pollution sources and contributing areas within the watershed, resulting in a lack of targeted and inefficient formulation of pollution control measures. Summary of the Invention

[0009] To address the shortcomings of existing technologies, the present invention aims to provide a multi-scale coupled method, apparatus, equipment, and medium for tracing watershed pollution sources step by step, in order to solve the problems of intermediate source tracing gaps and low source tracing accuracy in existing technologies.

[0010] According to one aspect of this application, a multi-scale coupled method for tracing watershed pollution sources at each level is disclosed, the method comprising: Obtain the target simulation results for each sub-basin output by the target SWAT model. The sub-basin is obtained by dividing the target basin based on the elevation information data of the target basin by the original SWAT model. The target simulation result is the average pollution load per sub-basin. The average pollution load of the sub-basin is determined based on the hydrological target simulation data and water quality target simulation data of the sub-basin. Based on the target simulation results, key sub-basins are identified from multiple sub-basins of the target watershed; After receiving the key simulation results corresponding to the key sub-basin, the APEX model is invoked with the key simulation results as the boundary conditions of the APEX model. The key sub-basin is then divided into sub-regions and simulated to obtain the optimized simulation results of the key sub-basin. The optimized simulation results include the average pollution load per sub-region, which is obtained by dividing the key sub-basin based on the APEX model. Divide the sub-region into grid cells of the target rectangular size; Extract each environmental factor from the grid, perform correlation analysis between each environmental factor and the correlation between each environmental factor and the total amount of each target pollutant molecule in the target watershed; Based on the correlation analysis results, the grid environment indicators for each grid cell are determined; Using the target simulation results and the optimized simulation results as dependent variables, and the grid environmental index as independent variables, a quantitative relationship between the grid environmental index and the pollution load is constructed. Based on the quantitative relationship, a spatial distribution map of the pollution load of the target watershed is constructed. The spatial distribution map is used to characterize the spatial set of pollution loads of each grid unit in the target watershed. The target pollution load of the target grid unit is traced based on the spatial distribution map.

[0011] In some embodiments, before obtaining the target simulation results for each sub-basin output by the target SWAT model, the method further includes: Obtain watershed attribute data for the target watershed during the target time period. The watershed attribute data includes elevation information data, land use map, soil type map, and meteorological data. Based on the elevation information data, the land use map, and the soil type map, the original SWAT model is initialized and constructed to obtain the initial SWAT model of the target watershed. During the initialization process, the original SWAT model divides the target watershed into multiple sub-watersheds based on the elevation information data and then defines the hydrological response simulation unit of the target SWAT model in combination with the land use map and the soil type map. After obtaining the meteorological data, the initial SWAT model is called to simulate each sub-basin, and the initial hydrological simulation data and initial water quality simulation data of each sub-basin at the target location are obtained. Acquire actual monitoring data for each sub-basin at the target location, including hydrological monitoring data and water quality monitoring data; Based on the hydrological monitoring data, the water quality monitoring data, the initial hydrological simulation data, and the initial water quality simulation data, the initial SWAT model is optimized to obtain the target SWAT model for the target watershed.

[0012] In some embodiments, obtaining the target simulation results for each sub-basin output by the target SWAT model includes: After acquiring elevation information data, land use map, soil type map and meteorological data of the target watershed, the target SWAT model is called to simulate and output the target simulation results for each sub-watershed of the target watershed.

[0013] In some embodiments, optimizing the initial SWAT model based on the hydrological monitoring data, the water quality monitoring data, the initial hydrological simulation data, and the initial water quality simulation data to obtain the target SWAT model for the target watershed includes: The initial hydrological simulation data is compared with the hydrological monitoring data to obtain a first comparison result, and the initial water quality simulation data is compared with the water quality monitoring data to obtain a second comparison result. Based on the first comparison result and the second comparison result, the sensitive parameters in the initial SWAT model are adjusted using parameter optimization tools, and the model that minimizes the error between the initial simulation results and the actual monitoring data is used as the calibrated SWAT model. Obtain the verification meteorological data corresponding to the verification time period and the verification monitoring data for the verification time period; The verification meteorological data is input into the calibration SWAT model to obtain the verification simulation results; A third comparison result is obtained by comparing the verification simulation results with the verification monitoring data. Obtain evaluation metrics; When the third comparison result meets the preset threshold and the evaluation index, the calibrated SWAT model is determined to be the target SWAT model.

[0014] In some embodiments, determining the key sub-basins from multiple sub-basins of the target watershed based on the target simulation results includes: Determine the pollution load contribution rate of each sub-basin to the pollution compliance of the target basin and the pollution load intensity per unit area of ​​each sub-basin; Based on the pollution load contribution rate and the load intensity, key sub-basins are identified from multiple sub-basins of the target watershed.

[0015] In some embodiments, constructing a quantitative relationship between the grid environmental index and the pollution load, using the target simulation result and the optimized simulation result as dependent variables and the grid environmental index as independent variables, includes: Using the target simulation results and the optimized simulation results as dependent variables, and the grid environmental index as independent variables, a downscaling equation is established by combining a regression model or machine learning method. This downscaling equation characterizes the quantitative relationship between environmental indicators and pollution load. The expression of the downscaling equation is as follows: ; ; In the formula: L N , L P These represent the unit loads of nitrogen and phosphorus at the grid scale, respectively. α 0 , β 0 For constant terms; X i , Y j Environmental indicators at the grid scale; α i , β j These are the regression coefficients for each environmental indicator.

[0016] In some embodiments, the method further includes: Obtain the key simulation results output by the APEX model, including key hydrological simulation results and key water quality simulation results; The key hydrological simulation results are compared with the hydrological monitoring data to obtain a fourth comparison result, and the key water quality simulation results are compared with the water quality monitoring data to obtain a fifth comparison result. Based on the fourth and fifth comparison results, the sensitive parameters in the initial SWAT model are adjusted using parameter optimization tools, and the model that minimizes the error between the key simulation results and the measured values ​​is used as the calibration SWAT model.

[0017] According to another aspect of this application, a multi-scale coupled watershed pollution source tracing device is also disclosed, the device comprising: The target simulation result acquisition module is used to acquire the target simulation results of each sub-basin output by the target SWAT model. The sub-basin is obtained by dividing the target basin based on the elevation information data of the target basin by the original SWAT model. The target simulation result is the average pollution load per sub-basin. The average pollution load of the sub-basin is determined based on the hydrological target simulation data and water quality target simulation data of the sub-basin. The key sub-basin determination module is used to determine the key sub-basins from multiple sub-basins of the target basin based on the target simulation results; The optimization simulation result acquisition module is used to acquire the key simulation results corresponding to the key sub-basin, call the APEX model with the key simulation results as the boundary conditions of the APEX model, divide the key sub-basin into sub-regions and simulate to obtain the optimization simulation results of the key sub-basin. The optimization simulation results include the average pollution load per sub-region, which is obtained by dividing the key sub-basin based on the APEX model. The mesh cell generation module is used to divide the sub-region into mesh cells with a target rectangular size; The correlation analysis module is used to extract environmental factors from the grid, perform correlation analysis between environmental factors, and analyze the correlation between each environmental factor and the total amount of each target pollutant molecule in the target watershed. The grid environment index determination module is used to determine the grid environment index of each grid cell based on the results of correlation analysis. The quantitative relationship construction module is used to construct a quantitative relationship between the grid environmental indicators and the pollution load, with the target simulation results and the optimized simulation results as dependent variables and the grid environmental indicators as independent variables. A spatial distribution map construction module is used to construct a spatial distribution map of the pollution load of the target watershed based on the quantitative relationship. The spatial distribution map is used to characterize the spatial set of pollution loads of each grid unit in the target watershed. The source tracing module is used to trace the target pollution load of the target grid unit based on the spatial distribution map.

[0018] According to another aspect of this application, an electronic device is also disclosed, the electronic device including a memory and at least one processor, the memory storing instructions; the at least one processor invokes the instructions in the memory to cause the electronic device to perform the steps of the multi-scale coupled watershed pollution tracing method as described in any of the preceding claims.

[0019] According to another aspect of this application, a computer-readable storage medium is also disclosed, wherein instructions are stored on the computer-readable storage medium, characterized in that, when executed by a processor, the instructions implement the various steps of the multi-scale coupled watershed pollution source tracing method as described in any of the preceding claims.

[0020] The present invention includes, but is not limited to, the following beneficial effects: (1) The present invention constructs a multi-level analysis chain of watershed → sub-watershed → sub-region → grid unit, which effectively solves the problem of information loss or insufficient accuracy when the traditional method jumps scale, realizes the step-by-step fine-grained source tracing from the whole watershed to the specific grid unit, solves the problem of intermediate gap in source tracing in the prior art, and improves the accuracy of source tracing; (2) The pollution load spatial distribution map generated by the present invention can be directly used to identify key pollution grids, help environmental managers to accurately implement governance measures to specific locations, improve governance efficiency and resource utilization, and the method finally outputs an intuitive spatial distribution map, which transforms complex pollution data into visual information, which helps decision-makers to quickly understand the spatial pattern of pollution and formulate targeted watershed management strategies; (3) The initial simulation results of the model are optimized by actual monitoring data, which improves the simulation accuracy of the final target SWAT model, and the simulation output of the physical mechanism model (SWAT) is dynamically compared and the parameters are optimized by actual monitoring data to form A simulation-monitoring-correction feedback loop was established, which not only improved the local adaptability of the model, but also enabled the model to continuously approximate the hydrological and water quality processes in the real world, enhancing the model's predictive ability and applicability in specific watersheds; (4) This scheme transforms the complex problem of spatial identification of watershed pollution into a quantifiable, operable, and goal-oriented scientific decision-making process. Specifically, through the dual screening mechanism of contribution rate and load intensity, it can more comprehensively identify different types of key areas, realize the linkage between macro-simulation and micro-governance actions, and enhance the practical value and engineering guidance significance of the entire source tracing method; (5) This scheme standardizes the downscaling process into a clear equation construction step and is compatible with regression models or machine learning methods, reflecting the flexibility and scalability of the method. This allows the step to be used as a standardized technical module and applied to the source tracing work of other watersheds or other pollutants. As long as the input data and environmental indicators are replaced, a new downscaling relationship can be quickly established, which has good universality and promotion value. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0022] Figure 1 This is a flowchart of a multi-scale coupled watershed pollution source tracing method according to an embodiment of this application; Figure 2 This is a flowchart illustrating the construction of a target SWAT model according to an embodiment of this application; Figure 3 This is a flowchart illustrating a specific method for constructing a target SWAT model according to an embodiment of this application; Figure 4This is another flowchart of the multi-scale coupled watershed pollution source tracing method according to the embodiments of this application; Figure 5 This is a structural block diagram of a multi-scale coupled watershed pollution source tracing device according to an embodiment of this application; Figure 6 This is a schematic diagram of the structure of the electronic device provided in an embodiment of the present invention; Figure 7 This is a distribution diagram of the average pollution load according to an embodiment of this application; Figure 8 This is a schematic diagram of sub-basin division according to an embodiment of this application; Figure 9 This is a schematic diagram of the quantitative relationship in an embodiment of this application; Figure 10 This is a spatial distribution diagram of an embodiment of this application. Detailed Implementation

[0023] This invention provides a multi-scale coupled method for tracing watershed pollution sources at each level. The method includes: acquiring the target simulation results for each sub-watershed output by a target SWAT model, wherein the sub-watershed is obtained by dividing the target watershed based on elevation information data of the target watershed using the original SWAT model; the target simulation result is the average pollution load per sub-watershed, determined based on hydrological and water quality target simulation data for that sub-watershed; identifying key sub-watersheds from multiple sub-watersheds based on the target simulation results; receiving the key simulation results corresponding to the key sub-watersheds; and calling an APEX model with the key simulation results as boundary conditions for the APEX model to process the key sub-watersheds. After subdividing the watershed into sub-regions, simulations are performed to obtain optimized simulation results for the key sub-watersheds. These optimized simulation results include the average pollution load per sub-region, which is obtained by dividing the key sub-watersheds based on the APEX model. The sub-regions are then divided into grid cells with target rectangular dimensions. Environmental indicators for each grid cell are determined. A quantitative relationship between the environmental indicators and pollution load is constructed, using the target simulation results and the optimized simulation results as dependent variables and the environmental indicators as independent variables. Based on this quantitative relationship, a spatial distribution map of the pollution load in the target watershed is constructed. This spatial distribution map characterizes the spatial set of pollution loads in each grid cell of the target watershed. The target pollution load of the target grid cell is then traced based on the spatial distribution map. This method enables step-by-step source tracing from watershed to sub-watershed to grid, solving the problem of intermediate gaps in source tracing in existing technologies and improving source tracing accuracy.

[0024] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0025] For ease of understanding, the specific process of the embodiments of the present invention will be described below. Figure 1 For the multi-scale coupled watershed pollution source tracing method of this application embodiment, see [link to relevant documentation]. Figure 1 It includes the following steps: S100: Obtain the target simulation results for each sub-basin from the target SWAT model output.

[0026] The sub-basins are derived from the original SWAT model by dividing the target watershed based on elevation information data. The target simulation result is the average pollution load per sub-basin, determined based on the hydrological and water quality target simulation data for that sub-basin. Specifically, the hydrological target simulation data refers to the runoff (e.g., cubic meters per second) simulated by the model, while the water quality target simulation data refers to the concentration of pollutants in the water, such as total nitrogen (TN) concentration in milligrams per liter. Pollution load = runoff × average pollutant concentration. For example, if the annual average flow at the outlet of a sub-basin is 10 million cubic meters and the average TN concentration in the water is 2 milligrams per liter, then the annual TN load of that sub-basin is: (10 million m³) × (2 mg / L) = 20 tons / year. The original SWAT model (Soil and Water Assessment Tool) is a distributed watershed hydrological model that simulates the migration processes of water, sediment, and chemicals in complex watersheds based on a GIS platform, suitable for long-term land use management impact prediction.

[0027] Specifically, obtaining the target simulation results for each sub-basin from the target SWAT model can be achieved by receiving elevation data, land use maps, soil type maps, and meteorological data for the target time period of the target watershed, and then calling the target SWAT model to simulate and output the results for each sub-basin, thus obtaining the target simulation results for each sub-basin. The elevation data, land use maps, soil type maps, and meteorological data can be obtained from environmental monitoring departments or from geographic information data centers. After obtaining the elevation data, coordinate unification, depression filling, and slope calculation can be performed to extract key hydrological and geomorphological indicators such as watershed boundaries, river network paths, and topographic humidity index. Simultaneously, the land use, soil type, and meteorological data are formatted and missing values ​​are handled to provide high-quality basic data for model input. In this example, the target SWAT model first simulates hydrological processes to obtain runoff, then simulates water quality processes to obtain pollutant concentrations, and finally multiplies the two to obtain the final output for each sub-basin—the average pollution load.

[0028] S102. Based on the target simulation results, key sub-basins are identified from multiple sub-basins of the target watershed.

[0029] For example, the pollution load contribution rate of each sub-basin in the pollution load of the target basin and the pollution load intensity per unit area of ​​each sub-basin can be determined first. Then, based on the pollution load contribution rate and load intensity, key sub-basins can be identified from multiple sub-basins of the target basin.

[0030] In this example, when the target time period is in years, the pollution load contribution rate = annual pollution load of the sub-basin / total annual pollution load of the entire basin, and the load intensity = annual pollution load of the sub-basin / area of ​​the sub-basin. It is understandable that sub-basins with high pollution load contribution rates are the main sources of downstream water pollution and require priority control. Load intensity is used to offset the impact of area size, reflecting the pollution risk or management intensity of the land itself. A small sub-basin with extremely high load intensity (such as an intensive agricultural area or industrial area) may contain strong pollution sources and migration pathways, making it a potential "pollution hotspot" requiring in-depth investigation. Therefore, when either a high pollution load contribution rate or a high pollution load intensity is met, the corresponding sub-basin is designated as a critical sub-basin.

[0031] S104. After obtaining the key simulation results corresponding to the key sub-basins, call the APEX model and use the key simulation results as the boundary conditions of the APEX model to divide the key sub-basins into sub-regions and then simulate them to obtain the optimized simulation results of the key sub-basins.

[0032] Specifically, the Agricultural Policy / Environmental eXtender (APEX) model is an agricultural model published by Texas A&M University, applicable to the assessment of agricultural non-point source pollution and the effectiveness of agricultural management measures at the field and small-to-medium watershed scales. This example uses key hydrological and water quality simulation data from key sub-watersheds as boundary conditions for the APEX model, ensuring that the starting point of the APEX model simulation remains consistent with the key simulation results of the corresponding key sub-watersheds, thus avoiding discrepancies between scales.

[0033] The optimized simulation results include average pollution loads at the sub-region level, which are derived from the APEX model by dividing key sub-watersheds. Specifically, the APEX model can further subdivide each sub-watershed into smaller sub-regions. This refinement allows APEX to more realistically simulate the internal dynamics of sub-watersheds, better reflecting the varying impacts of soil, slope, and management practices on pollutants in different sub-regions. It also provides a more precise description of the complete migration path of pollutants from one field to another, and then to ditches and rivers.

[0034] S106. Divide the sub-region into grid cells of the target rectangular size.

[0035] For example, the grid cell size can be 1km×1km, 0.8km×0.8km, or 2km×2km. The specific size can be set based on actual needs. In this example, the grid cell size is 1km×1km.

[0036] S108. Extract each environmental factor from the grid, perform correlation analysis between each environmental factor and the correlation between each environmental factor and the total amount of each target pollutant molecule in the target watershed.

[0037] Specifically, in this example, the target pollutants are total nitrogen and total phosphorus.

[0038] S110. Based on the correlation analysis results, determine the grid environment indicators for each grid cell.

[0039] For example, environmental factors include, but are not limited to: (1) Land use and cover index: Agriculture (Agr), Forest (For), Grassland (Grass), Open water (Ope), Aquaculture (Aqu), Urban (Urb), Suburban (Sub) and Industrial (Ind) systems; This index is used to characterize the potential pollution output capacity of grid cells, for example, the nitrogen and phosphorus loads of farmland grids are much higher than those of forest grids.

[0040] (2) Topographic and geomorphological indicators: Aspect, Slope, Elevation; these indicators determine the ease of pollutant generation and the speed of migration. For example, in places with steep slopes and high TWI values, the runoff generation capacity is strong, and pollutants are more easily washed into the river.

[0041] (3) Vegetation and ecological indicators: Planar curvature (Pl_Cur), effective flow length (EFL), and flow rate (FA); these indicators are used to characterize the water flow convergence capacity of grid cells.

[0042] (4) Hydrological and topographic humidity indicators: Topographic humidity index (TWI), topographic location index (TPI), and modified catchment area (MCA); these indicators reflect the location of the grid unit in the watershed (such as whether it is located on a ridge or in a valley), and affect the convergence path of water flow and pollutants.

[0043] (5) Comprehensive indicators and others: convergence index (CI), contour curvature (Pr_Cur).

[0044] Specifically, environmental factors for each grid cell can be extracted based on the SAGA-GIS geographic information system.

[0045] After identifying the environmental factors for each grid cell, the correlation between the environmental factors and the correlation between each environmental factor and the total amount of the target pollutant molecules are analyzed to obtain the correlation analysis results.

[0046] Specifically, redundant and highly collinear environmental factors can be identified and eliminated based on the Pearson correlation coefficient matrix. For example, if two environmental factors (such as the topographic moisture index TWI and catchment area) are highly correlated, it means that they convey similar environmental information. Retaining both may lead to model overfitting and interpretation difficulties. In this case, one of them can be eliminated. Furthermore, the Pearson correlation coefficient between each environmental factor and total nitrogen (TN) load and total phosphorus (TP) load is calculated, and its significance (p-value) is tested. Based on the analysis results of the above two levels, a comprehensive screening is performed. For example, the environmental factor must have a high correlation coefficient (large absolute value) with TN or TP load and a significant p-value (usually p < 0.05 or 0.01). In the group of highly correlated environmental factors, only the one with the strongest correlation with pollution load or the most explicit physical meaning is retained. Indicators that have been widely proven in environmental science and hydrology to drive pollutant generation and migration processes (such as slope, topographic moisture index TWI, and proportion of agricultural land) are preferred. Through the above screening, a set of environmental indicators with a moderate number, good independence, and effective explanation of the spatial variability of nitrogen and phosphorus loads was finally obtained, which is the grid environmental indicator for each grid cell.

[0047] S112. Using the target simulation results and the optimized simulation results as dependent variables and the grid environmental indicators as independent variables, construct a quantitative relationship between the grid environmental indicators and the pollution load.

[0048] Specifically, step S112 can involve using the target simulation results and optimized simulation results as dependent variables, and the grid environmental indicators as independent variables, to establish a downscaling equation using a regression model or machine learning method. The downscaling equation is used to characterize the quantitative relationship between environmental indicators and pollution load. The expression for the downscaling equation is: ; ; In the formula: L N , L P These represent the unit loads of nitrogen and phosphorus at the grid scale, respectively. α 0 , β 0 For constant terms; X i , Y j Environmental indicators at the grid scale; α i , β j These are the regression coefficients for each environmental indicator.

[0049] S114. Based on quantitative relationships, construct a spatial distribution map of the pollution load in the target watershed. The spatial distribution map is used to characterize the spatial set of pollution loads in each grid unit of the target watershed.

[0050] Specifically, the target simulation results and optimized simulation results are downscaled and distributed to a 1km×1km grid based on their quantitative relationship with environmental factors, ultimately forming a spatial distribution map. This enables a progressively refined source tracing from the watershed to the grid, visually displaying pollution hotspots.

[0051] S116. Target pollution load of target grid unit based on spatial distribution map.

[0052] Specifically, after determining the spatial distribution map, the target grid unit that needs to be traced can be directly traced based on the spatial distribution map to obtain the target pollution load.

[0053] Furthermore, Figure 2 This is a flowchart of constructing a target SWAT model according to an embodiment of this application. This construction process is applied before step S100. See also... Figure 2It includes the following steps: S200. Obtain watershed attribute data for the target watershed during the target time period. The watershed attribute data includes elevation information data, land use map, soil type map, and meteorological data.

[0054] Specifically, a land use map is a thematic map showing land cover types, such as farmland, forest, grassland, urban areas, and water bodies. Different land use types have distinctly different hydrological and pollution characteristics. For example, farmland is a major source of nitrogen and phosphorus pollution, while forests have excellent water conservation and purification capabilities. A soil type map is a thematic map showing the spatial distribution of different soil types and their physicochemical properties (such as texture, organic matter content, and permeability).

[0055] S202. Based on elevation information data, land use map, and soil type map, the original SWAT model is initialized and constructed to obtain the initial SWAT model of the target watershed.

[0056] Understandably, elevation data is used to delineate watershed boundaries, river networks, and sub-watersheds; land use maps are used to determine key parameters for processes such as surface evapotranspiration, runoff generation, and pollutant production; and soil type maps provide information on soil hydrological grouping, permeability, and water-holding capacity, directly impacting the hydrological cycle. Meteorological data, including daily precipitation, temperature, wind speed, solar radiation, and relative humidity, are used to drive the initialization of the SWAT model.

[0057] During initialization, upon receiving elevation data, the original SWAT model automatically divides the target watershed into multiple sub-watersheds. Within each sub-watershed, land use and soil type maps are overlaid to further define hydrological response simulation units, thus defining the simulation units of the target SWAT model. It can be understood that a hydrological response simulation unit is the smallest spatial unit with the same combination of land use and soil types, and most of the model's calculations (such as runoff generation, evaporation, and pollutant formation) are performed on this hydrological response simulation unit.

[0058] S204. After acquiring meteorological data, call the initial SWAT model to simulate each sub-basin, and obtain the initial hydrological simulation data and initial water quality simulation data of each sub-basin at the target location.

[0059] Specifically, initial hydrological simulation data includes topography, meteorology, and runoff, primarily used to simulate the movement of water bodies—that is, how water flows and converges in spaces of different scales, serving as a carrier for pollutant migration. Initial water quality simulation data includes nitrogen or phosphorus concentrations and related environmental data, such as land use and soil, used to simulate the sources, transformations, and loads of pollutants.

[0060] S206. Obtain the actual monitoring data of each sub-basin at the target location. The actual monitoring data includes hydrological monitoring data and water quality monitoring data.

[0061] For example, the target location could refer to the outlet location of each sub-basin or the location of a key control section. A key control section can refer to a river section or monitoring point within the basin (not just the final outlet) that has significant hydrological and water quality representativeness. Distributed downstream of key tributaries or important areas within the basin, these sections are used to "control" or "represent" the water quantity and quality of the entire upstream sub-basin, such as locating pollution sources. By comparing monitoring data from different key sections, it can be determined which sub-basin is the main pollution contributor. For example, if the nitrogen concentration at section A is much higher than at section B, it indicates that the sub-basin upstream of section A is a key pollution area.

[0062] S208. Based on hydrological monitoring data, water quality monitoring data, initial hydrological simulation data, and initial water quality simulation data, optimize and initialize the SWAT model to obtain the target SWAT model for the target watershed.

[0063] Furthermore, Figure 3 This is a flowchart illustrating a specific method for constructing a target SWAT model according to an embodiment of this application. The flowchart is an exemplary description of step S208. (See attached document.) Figure 3 It includes the following steps: S300. Compare the initial hydrological simulation data with the hydrological monitoring data to obtain the first comparison result, and compare the initial water quality simulation data with the water quality monitoring data to obtain the second comparison result.

[0064] S302. Based on the first and second comparison results, the sensitive parameters in the initial SWAT model are adjusted using parameter optimization tools, and the model that minimizes the error between the initial simulation results and the actual monitoring data is used as the calibrated SWAT model.

[0065] For example, in this example, based on the first and second comparison results, combined with the SUFI-2 algorithm module built into this SWAT-CUP tool, parameter calibration and uncertainty analysis can be performed on the model. First, analyze the sensitive parameters in the model as shown in Table 1 to identify which parameters are most sensitive to and have the greatest impact on the simulation results (such as flow rate and total nitrogen concentration). This allows for focusing on adjusting key parameters to improve efficiency. Furthermore, within the parameter value range given in Table 1, the SUFI-2 algorithm automatically generates multiple different parameter combinations, then automatically runs the model, comparing the simulation results of each set with the measured data. Based on the comparison results, the algorithm continuously narrows down the parameter range and generates new, better parameter combinations, repeating this process hundreds or even thousands of times (i.e., iterations) until it finds the optimal parameter combination that best matches the simulation results with the measured data, i.e., minimizes the error.

[0066] Table 1 Common Sensitive Parameters and Their Value Ranges

[0067] S304. Obtain the verification meteorological data corresponding to the verification time period and the verification monitoring data for the verification time period.

[0068] Specifically, the validation period refers to a period of time that is different from the target period, during which meteorological data was not input into the original SWAT model.

[0069] S306. Input the verification meteorological data into the calibration SWAT model to obtain the verification simulation results.

[0070] S308. Based on the comparison between the verification simulation results and the verification monitoring data, the third comparison result is obtained.

[0071] S310. Obtain evaluation indicators.

[0072] S312. When the third comparison result meets the preset threshold and the evaluation index, the SWAT model is determined as the target SWAT model.

[0073] For example, evaluation metrics may include the coefficient of determination (R²). 2 Statistical indicators include the Nash efficiency coefficient (NSE) and percentage of deviation (P-bias). Among these, the coefficient of determination (R²) is the most important. 2 The closer the NSE is to 1, the higher the overall reliability of the simulation results. Generally, NSE > 0.5 is considered acceptable. The closer the percentage of bias (P-bias) is to 0, the smaller the systematic bias of the simulation values.

[0074] The evaluation results for this example are shown in Table 2: Table 2 Evaluation of Simulation Results of the Original SWAT Model

[0075] Based on the above evaluation results, the calibrated SWAT model in this example meets the requirements and can be used as the target SWAT model.

[0076] Furthermore, Figure 4 This is another flowchart of the multi-scale coupled watershed pollution source tracing method according to embodiments of this application. This flowchart is used to optimize the target SWAT model based on key simulation results output by the APEX model. For details, please refer to [link to relevant documentation]. Figure 4 It includes the following steps: S400: Obtain the key simulation results output by the APEX model. The key simulation results include key hydrological simulation results and key water quality simulation results.

[0077] S402. Compare the key hydrological simulation results with the hydrological monitoring data to obtain the fourth comparison result, and compare the key water quality simulation results with the water quality monitoring data to obtain the fifth comparison result.

[0078] S404. Based on the fourth and fifth comparison results, the sensitive parameters in the initial SWAT model are adjusted using parameter optimization tools, and the model that minimizes the error between the key simulation results and the measured values ​​is used as the calibrated SWAT model.

[0079] Understandably, steps S304-S312 can be executed directly after step S404 to finally determine the target SWAT model. In this example, the evaluation results of the SWAT model combined with the APEX model are shown in Table 3: Table 3 SWAT-APEX Model Simulation Evaluation Table

[0080] Based on the above evaluation results, the calibrated SWAT model in this example meets the requirements and can be used as the target SWAT model.

[0081] Furthermore, for ease of understanding, this application provides the following example: Taking the Qiantang River Basin as an example, this technology is applied to conduct refined source tracing of non-point source pollution: Elevation data, land use maps, soil type maps, and meteorological observation data were obtained from national environmental monitoring departments and geographic information data centers. Coordinate system optimization, depression filling, and slope calculations were performed on the elevation data to extract key hydrological and geomorphological indicators such as watershed boundaries, river network paths, and topographic humidity index. Simultaneously, the land use, soil type, and meteorological data underwent format standardization and missing value processing to provide high-quality basic data for model input.

[0082] Furthermore, the original SWAT model was used to simulate the overall hydrological processes and nitrogen and phosphorus loads of the target watershed. Specifically, elevation information, land use maps, soil type maps, and meteorological data shown in the attribute data distribution map of the target watershed were input into the original SWAT model for simulation, and the initial simulation data of each sub-watershed were output. Subsequently, sensitive parameters were identified and SUFI-2 calibrated using the SWAT-CUP tool. Key parameters such as soil hydraulic parameters, runoff generation coefficient, and nitrogen and phosphorus cycle rates were optimized, and the performance of the model was evaluated using R2, NSE, and P-bias indices (Table 2). After calibration and parameter tuning, the outputs showed a good fit, such as... Figure 7 The monthly average load distribution maps of nitrogen (TN) and phosphorus (TP) shown provide a reliable hydrological and water quality benchmark at the macro scale, offering a scientific basis for refined mesoscale simulation and downscaling extrapolation.

[0083] Furthermore, for key sub-basins where water quality monitoring stations are located, the APEX model is introduced for unit-level process simulation, using SWAT output as boundary conditions to simulate the sub-basins. Figure 8 After the division shown, a refined estimate of nitrogen and phosphorus loads within the sub-basins is achieved by combining land use, soil type, and meteorological data; secondary corrections (Ri) are then performed on these key sub-basins by comparing them with measured water quality data from various stations. 2 The NSE and P-bias indices (Table 3) are used to ensure that the mesoscale simulation conforms to the physical mechanism and has reliable data support.

[0084] Furthermore, the study area was divided into 1km×1km grid units, and environmental indicators such as land use ratio, slope, aspect, vegetation cover index, topographic humidity index, effective flow length, and flow convergence index were extracted. Through correlation analysis of these parameters and their correlation with logTN and logTP (if the TN and TP load values ​​are large, logarithmic transformation can be performed), indicators that can express the TN and TP load of each grid were selected, as in this case. Figure 9 The indicators Agr, For, Ope, FA, and TPI are shown. First, a refined simulation is performed using APEX within the key sub-basins where the monitoring stations are located. The results are then used to replace the simulated values ​​of the corresponding sub-basins in the SWAT model, thus obtaining a more accurate and precise benchmark for nitrogen and phosphorus loads in the macroscopic model. Subsequently, using the corrected SWAT output as the training data source, and with 1km×1km grid environmental indicators as independent variables, a quantitative relationship model for nitrogen and phosphorus loads is established using regression analysis or machine learning methods. This achieves downscaling from sub-basin to grid scale extrapolation, utilizing relative mean residuals (RSE) and coefficients of determination (R²). 2 The downscaling simulation results were evaluated using metrics such as the Nash efficiency coefficient (NSE) (Table 4). The final generated... Figure 10 The spatial distribution map of nitrogen and phosphorus loads shown is combined with land use management information to identify pollution hotspots and major sources. Through multi-scale coupling and progressive refinement methods, the model can accurately reflect the distribution characteristics of TN and TP under different topographical and land use types, achieving precise identification of pollution hotspot areas and providing a reliable scientific basis for watershed management and policy formulation.

[0085] Table 4 Evaluation Table of Downscaling Simulation Results

[0086]

[0087] Furthermore, Figure 5 The structural block diagram of the multi-scale coupled watershed pollution source tracing device according to the application embodiment is as follows: Figure 5 As shown, the device includes: The target simulation result acquisition module is used to acquire the target simulation results of each sub-basin output by the target SWAT model. The sub-basin is obtained by dividing the target basin based on the elevation information data of the target basin in the original SWAT model. The target simulation result is the average pollution load per sub-basin. The average pollution load of the sub-basin is determined based on the hydrological target simulation data and water quality target simulation data of that sub-basin. The key sub-watershed determination module is used to determine the key sub-watershed from multiple sub-watersheds of the target watershed based on the target simulation results; The optimization simulation result acquisition module is used to acquire the key simulation results corresponding to the key sub-basins, call the APEX model with the key simulation results as the boundary conditions of the APEX model, divide the key sub-basins into sub-regions and simulate them to obtain the optimized simulation results of the key sub-basins. The optimized simulation results include the average pollution load per sub-region, which is obtained by dividing the key sub-basins based on the APEX model. The mesh cell generation module is used to divide the sub-region into mesh cells with a target rectangular size; The grid environment index determination module is used to determine the grid environment index for each grid cell. The quantitative relationship construction module is used to construct a quantitative relationship between grid environmental indicators and pollution load, with target simulation results and optimized simulation results as dependent variables and grid environmental indicators as independent variables. The spatial distribution map construction module is used to construct a spatial distribution map of the pollution load of the target watershed based on quantitative relationships. The spatial distribution map is used to characterize the spatial set of pollution loads of each grid unit in the target watershed. The source tracing module is used to trace the target pollution load of the target grid unit based on the spatial distribution map.

[0088] The application of the relevant modules of the device in this example can be referred to the relevant introduction of the method principle above, and will not be repeated here.

[0089] This approach constructs a multi-level analysis chain from watershed to sub-watershed to sub-region to grid unit, effectively addressing the issues of information loss or insufficient accuracy in traditional methods when scales jump. It achieves progressively refined source tracing from the overall watershed to specific grid units, resolving the intermediate gaps in source tracing in existing technologies and improving accuracy. Furthermore, the pollution load spatial distribution map generated by this approach can be directly used to identify key pollution grids, helping environmental managers to precisely implement remediation measures at specific locations, improving remediation efficiency and resource utilization. The method ultimately outputs an intuitive spatial distribution map, transforming complex pollution data into visualized information, which helps decision-makers quickly understand the spatial pattern of pollution and formulate targeted watershed management strategies. Moreover, optimizing the initial simulation results of the model using actual monitoring data significantly improves the simulation accuracy of the final target SWAT model. By dynamically comparing and optimizing the simulation output of the physical mechanism model (SWAT) with actual monitoring data, a simulation-monitoring system is formed. - The feedback loop of the correction not only improves the model's local adaptability but also enables it to continuously approximate real-world hydrological and water quality processes, enhancing its predictive ability and applicability in specific watersheds. Furthermore, this scheme transforms the complex problem of spatial identification of watershed pollution into a quantifiable, operable, and goal-oriented scientific decision-making process. Specifically, through a dual screening mechanism of contribution rate and load intensity, it can more comprehensively identify different types of key areas, achieving a linkage between macro-simulation and micro-level governance actions, thus enhancing the practical value and engineering guidance significance of the entire source tracing method. Furthermore, this scheme standardizes the downscaling process into a clear equation construction step, compatible with regression models or machine learning methods, demonstrating the method's flexibility and scalability. This allows this step to be applied as a standardized technical module to source tracing work in other watersheds or for other pollutants. By simply replacing the input data and environmental indicators, new downscaling relationships can be quickly established, demonstrating excellent universality and promotional value.

[0090] above Figure 5 The companion robot interaction device in the embodiments of the present invention will be described in detail from the perspective of modular functional entities. The electronic device in the embodiments of the present invention will be described in detail from the perspective of hardware processing.

[0091] Figure 6This is a schematic diagram of the structure of an electronic device 600 provided in an embodiment of the present invention. The electronic device 600 can vary significantly due to different configurations or performance characteristics. It may include one or more central processing units (CPUs) 610 (e.g., one or more processors) and a memory 620, and one or more storage media 630 (e.g., one or more mass storage devices) for storing application programs 633 or data 632. The memory 620 and storage media 630 can be temporary or persistent storage. The program stored in the storage media 630 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the electronic device 600. Furthermore, the processor 610 may be configured to communicate with the storage media 630 and execute the series of instruction operations in the storage media 630 on the electronic device 600.

[0092] Electronic device 600 may also include one or more power supplies 640, one or more wired or wireless network interfaces 650, one or more input / output interfaces 660, and / or one or more operating systems 631, such as Windows Server, MacOSX, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 6 The illustrated electronic device structure does not constitute a limitation on electronic devices and may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0093] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of any of the above-described multi-scale coupled watershed pollution source tracing methods.

[0094] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system, device, or unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0095] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0096] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-scale coupled method for tracing watershed pollution sources step by step, characterized in that, The method includes: Obtain the target simulation results for each sub-basin output by the target SWAT model. The sub-basin is obtained by dividing the target basin based on the elevation information data of the target basin by the original SWAT model. The target simulation result is the average pollution load per sub-basin. The average pollution load of the sub-basin is determined based on the hydrological target simulation data and water quality target simulation data of the sub-basin. Based on the target simulation results, key sub-basins are identified from multiple sub-basins of the target watershed; After obtaining the key simulation results corresponding to the key sub-basin, the APEX model is called with the key simulation results as the boundary conditions of the APEX model to divide the key sub-basin into sub-regions and then simulate to obtain the optimized simulation results of the key sub-basin. The optimized simulation results include the average pollution load per sub-region, which is obtained by dividing the key sub-basin based on the APEX model. Divide the sub-region into grid cells of the target rectangular size; Extract each environmental factor from the grid, perform correlation analysis between each environmental factor and the correlation between each environmental factor and the total amount of each target pollutant molecule in the target watershed; Based on the correlation analysis results, the grid environment indicators for each grid cell are determined; Using the target simulation results and the optimized simulation results as dependent variables, and the grid environmental index as independent variables, a quantitative relationship between the grid environmental index and the pollution load is constructed. Based on the quantitative relationship, a spatial distribution map of the pollution load of the target watershed is constructed. The spatial distribution map is used to characterize the spatial set of pollution loads of each grid unit in the target watershed. The target pollution load of the target grid unit is traced based on the spatial distribution map.

2. The multi-scale coupled watershed pollution source tracing method according to claim 1, characterized in that, Before obtaining the target simulation results for each sub-basin from the target SWAT model output, the method further includes: Obtain watershed attribute data for the target watershed during the target time period. The watershed attribute data includes elevation information data, land use map, soil type map, and meteorological data. Based on the elevation information data, the land use map, and the soil type map, the original SWAT model is initialized and constructed to obtain the initial SWAT model of the target watershed. During the initialization process, the original SWAT model divides the target watershed into multiple sub-watersheds based on the elevation information data and then defines the hydrological response simulation unit of the target SWAT model in combination with the land use map and the soil type map. After obtaining the meteorological data, the initial SWAT model is called to simulate each sub-basin, and the initial hydrological simulation data and initial water quality simulation data of each sub-basin at the target location are obtained. Acquire actual monitoring data for each sub-basin at the target location, including hydrological monitoring data and water quality monitoring data; Based on the hydrological monitoring data, the water quality monitoring data, the initial hydrological simulation data, and the initial water quality simulation data, the initial SWAT model is optimized to obtain the target SWAT model for the target watershed.

3. The multi-scale coupled watershed pollution source tracing method according to claim 2, characterized in that, The target simulation results for each sub-basin output by the target SWAT model include: After acquiring elevation information data, land use map, soil type map and meteorological data of the target watershed, the target SWAT model is called to simulate and output the target simulation results for each sub-watershed of the target watershed.

4. The multi-scale coupled watershed pollution source tracing method according to claim 2, characterized in that, The step of optimizing the initial SWAT model based on the hydrological monitoring data, the water quality monitoring data, the initial hydrological simulation data, and the initial water quality simulation data to obtain the target SWAT model for the target watershed includes: The initial hydrological simulation data is compared with the hydrological monitoring data to obtain a first comparison result, and the initial water quality simulation data is compared with the water quality monitoring data to obtain a second comparison result. Based on the first comparison result and the second comparison result, the sensitive parameters in the initial SWAT model are adjusted using parameter optimization tools, and the model that minimizes the error between the initial simulation results and the actual monitoring data is used as the calibrated SWAT model. Obtain the verification meteorological data corresponding to the verification time period and the verification monitoring data for the verification time period; The verification meteorological data is input into the calibration SWAT model to obtain the verification simulation results; A third comparison result is obtained by comparing the verification simulation results with the verification monitoring data. Obtain evaluation metrics; When the third comparison result meets the preset threshold and the evaluation index, the calibrated SWAT model is determined to be the target SWAT model.

5. The multi-scale coupled watershed pollution source tracing method according to claim 1, characterized in that, The process of identifying key sub-basins from multiple sub-basins of the target watershed based on the target simulation results includes: Determine the pollution load contribution rate of each sub-basin to the pollution compliance of the target basin and the pollution load intensity per unit area of ​​each sub-basin; Based on the pollution load contribution rate and the load intensity, key sub-basins are identified from multiple sub-basins of the target watershed.

6. The multi-scale coupled watershed pollution source tracing method according to claim 1, characterized in that, The construction of a quantitative relationship between the grid environmental index and the pollution load, using the target simulation results and the optimized simulation results as dependent variables and the grid environmental index as independent variables, includes: Using the target simulation results and the optimized simulation results as dependent variables, and the grid environmental index as independent variables, a downscaling equation is established by combining a regression model or machine learning method. This downscaling equation characterizes the quantitative relationship between environmental indicators and pollution load. The expression of the downscaling equation is as follows: ； ； In the formula: L N , L P These represent the unit loads of nitrogen and phosphorus at the grid scale, respectively. α 0 , β 0 For constant terms; X i , Y j Environmental indicators at the grid scale; α i , β j These are the regression coefficients for each environmental indicator.

7. The multi-scale coupled watershed pollution source tracing method according to claim 4, characterized in that, The method further includes: Obtain the key simulation results output by the APEX model, including key hydrological simulation results and key water quality simulation results; The key hydrological simulation results are compared with the hydrological monitoring data to obtain a fourth comparison result, and the key water quality simulation results are compared with the water quality monitoring data to obtain a fifth comparison result. Based on the fourth and fifth comparison results, the sensitive parameters in the initial SWAT model are adjusted using parameter optimization tools, and the model that minimizes the error between the key simulation results and the actual monitoring data is used as the calibrated SWAT model.

8. A multi-scale coupled watershed pollution source tracing device, characterized in that, The device includes: The target simulation result acquisition module is used to acquire the target simulation results of each sub-basin output by the target SWAT model. The sub-basin is obtained by dividing the target basin based on the elevation information data of the target basin by the original SWAT model. The target simulation result is the average pollution load per sub-basin. The average pollution load of the sub-basin is determined based on the hydrological target simulation data and water quality target simulation data of the sub-basin. The key sub-basin determination module is used to determine the key sub-basins from multiple sub-basins of the target basin based on the target simulation results; The optimization simulation result acquisition module is used to acquire the key simulation results corresponding to the key sub-basin, call the APEX model with the key simulation results as the boundary conditions of the APEX model, divide the key sub-basin into sub-regions and simulate to obtain the optimization simulation results of the key sub-basin. The optimization simulation results include the average pollution load per sub-region, which is obtained by dividing the key sub-basin based on the APEX model. The grid cell division module is used to divide each sub-region into grid cells with a target rectangular size; The correlation analysis module is used to extract the environmental factors of the grid, perform correlation analysis between the environmental factors, and analyze the correlation between each environmental factor and the total amount of each target pollutant molecule in the target watershed. The grid environment index determination module is used to determine the grid environment index of each grid cell based on the results of correlation analysis. The quantitative relationship construction module is used to construct a quantitative relationship between the grid environmental indicators and the pollution load, with the target simulation results and the optimized simulation results as dependent variables and the grid environmental indicators as independent variables. A spatial distribution map construction module is used to construct a spatial distribution map of the pollution load of the target watershed based on the quantitative relationship. The spatial distribution map is used to characterize the spatial set of pollution loads of each grid unit in the target watershed. The source tracing module is used to trace the target pollution load of the target grid unit based on the spatial distribution map.

9. An electronic device, characterized in that, The electronic device includes a memory and at least one processor, the memory storing instructions; the at least one processor invokes the instructions in the memory to cause the electronic device to perform the steps of the multi-scale coupled watershed pollution source tracing method as described in any one of claims 1-7.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the various steps of the multi-scale coupled watershed pollution source tracing method as described in any one of claims 1-7.

Images