A method for quantitatively analyzing driving mechanism of water, sediment, nitrogen and phosphorus in a river basin

Abstract

The application discloses a watershed water and sediment nitrogen and phosphorus driving mechanism quantitative analysis method, relates to the field of quantitative analysis, acquires target watershed basic data and constructs a SWAT model, calibrates and verifies the model; climate change and agricultural management scenarios are set, the temporal and spatial loss processes of the water and sediment nitrogen and phosphorus of the watershed under different scenarios are simulated through the SWAT model; a prediction model is constructed based on the simulation results of the SWAT, and the influence degree of each driving factor on the loss of the water and sediment nitrogen and phosphorus is quantified in combination with the SHAP method; based on the spatial distribution characteristics and the SHAP quantification results, high loss areas and key driving factors are identified, and a partitioned management strategy is proposed. Through the comprehensive physical process simulation capability of the SWAT model, the nonlinear relationship capturing advantage of the XGBoost model and the driving mechanism analysis function of the SHAP method, systematic quantitative analysis of the driving mechanism of the loss of the water and sediment nitrogen and phosphorus of the watershed is realized, and scientific basis is provided for precise control and adaptive strategy making of the watershed.

Description

Technical Field

[0001] This application relates to the field of quantitative analysis technology, and in particular to a quantitative analysis method for the driving mechanism of water, sediment, nitrogen and phosphorus in watersheds. Background Technology

[0002] The intensification of global climate change and the increase in human activity have led to growing uncertainty in the loss of watershed runoff, sediment, total nitrogen, and total phosphorus, becoming a significant issue constraining sustainable development in watersheds. Climate change significantly enhances the volatility of water, sediment, nitrogen, and phosphorus losses by altering precipitation patterns and temperature patterns; while human activities, particularly adjustments to crop planting structures and excessive fertilizer application, exacerbate this trend. Changes in crop types affect land cover, thereby regulating runoff and sediment production; simultaneously, changes in crop nutrient requirements directly impact soil nitrogen and phosphorus uptake efficiency. Excessive fertilization leads to soil nitrogen and phosphorus accumulation, significantly increasing the risk of loss under rainfall-runoff erosion. Therefore, a thorough analysis of the impacts of climate change and human activities on watershed water, sediment, nitrogen, and phosphorus losses is crucial for developing scientific integrated watershed management and adaptation strategies.

[0003] Currently, traditional research often uses SWAT models to assess watershed water, sediment, nitrogen, and phosphorus loss, focusing on process simulation driven by physical mechanisms. However, watershed systems are complex systems coupled with multiple factors such as climate change and human activities, making them difficult to fully analyze through a single physical process. While relying solely on SWAT model simulations can reveal loss patterns at the macroscopic level, it neglects the nonlinear correlations, multidimensional couplings, and microscopic mechanisms among driving factors, thus limiting the targetedness and effectiveness of watershed management measures. Summary of the Invention

[0004] To address the aforementioned challenges, this application provides a quantitative analysis method for the driving mechanisms of water, sediment, nitrogen, and phosphorus loss in watersheds. By integrating the physical process simulation capabilities of the SWAT model, the nonlinear relationship capture advantages of the XGBoost model, and the driving mechanism analysis function of the SHAP method, this method systematically covers the core aspects of physical processes, driving factor identification, and quantification of water, sediment, nitrogen, and phosphorus loss. It compensates for the shortcomings of single models in the analysis of complex watershed systems, providing a quantitative basis with both physical rationality and interpretability for precise control and adaptive strategy formulation of watershed water, sediment, nitrogen, and phosphorus loss, thereby improving the scientificity and effectiveness of watershed ecological environment management.

[0005] To achieve the above objectives, this application provides a quantitative analysis method for the driving mechanisms of nitrogen and phosphorus in watershed water and sediment, comprising the following steps: S1: Obtain basic data for the target watershed, including digital elevation model data, land use data, soil data, meteorological data, water, sediment, nitrogen and phosphorus data, climate change data, and agricultural management data; S2: Construct a SWAT model of the target watershed based on the basic data of the target watershed, perform parameter sensitivity analysis on the model to identify key sensitive parameters, use the SUFI2 algorithm to complete the model parameter calibration and verification based on the key sensitive parameters, and use the coefficient of determination, Nash-Sercliff efficiency and percentage of bias to comprehensively evaluate the model simulation results. S3: Based on climate change data and agricultural management data, set up scenarios of climate change, planting structure adjustment and fertilizer reduction to obtain scenario data; S4: Input the scenario data into the validated target watershed SWAT model, and by modifying the crop growth parameters, nitrogen and phosphorus fertilizer application rate and application time in the target watershed SWAT model, simulate the watershed water, sediment, nitrogen and phosphorus loss process under different scenarios with historical periods as the baseline scenario, obtain the spatiotemporal change results under each scenario and draw a spatial distribution map. S5: Based on the simulation results of each scenario, the scenario number, sub-basin information, and baseline scenario water, sediment, nitrogen, and phosphorus data are selected as input features, and the rate of change of each variable relative to the baseline period is used as the output variable to construct a prediction model. The prediction model is then validated using the coefficient of determination and root mean square error. The SHAP analysis method is used to calculate the marginal contribution of each input feature in different combinations. The contribution value of each feature to the model prediction results is obtained by weighted averaging and the importance is ranked to quantify the degree of influence of each driving factor on the loss of water, sediment, nitrogen, and phosphorus in the basin. S6: Based on the spatial distribution characteristics simulated by the SWAT model, identify the watershed water, sediment, nitrogen, and phosphorus loss patterns, combine the SHAP quantitative analysis results to screen the scenarios of loss changes, analyze the impact mechanism of planting structure and fertilization pattern on loss, and propose a zoned management strategy for watershed water, sediment, nitrogen, and phosphorus in high loss areas to achieve refined management of the watershed water ecological environment.

[0006] Preferably, the meteorological data in S1 includes precipitation, maximum temperature, minimum temperature, relative humidity, and wind speed; the water, sediment, nitrogen, and phosphorus data include monthly runoff, monthly sediment, monthly total nitrogen, and monthly total phosphorus monitoring data; the climate change data includes precipitation and temperature data from the CMIP6 integrated data; and the agricultural management data includes crop planting area, planting structure, nitrogen fertilizer application rate, phosphorus fertilizer application rate, and fertilization time.

[0007] Preferably, the parameter sensitivity analysis and model parameter calibration verification in S2 are performed using SWAT-CUP software.

[0008] Preferably, the coefficient of determination is expressed as: ; In the formula, Let i be the observation value at time i. Let be the predicted value at time i. This is the average value of the observations; Nash Sergey's efficiency is expressed as: ; In the formula, This is the average value of the observations; The percentage deviation is expressed as: .

[0009] Preferably, the climate change scenarios in S3 specifically include: based on the integrated data of five global climate models in CMIP6, the SSP2-4.5 medium emission scenario and the SSP5-8.5 high emission scenario are selected, and three future periods are set for each emission pathway, forming a total of six climate change scenarios; Preferred scenarios include planting rice and peanuts on all cultivated land in the entire watershed; and scenarios for reducing fertilizer application include reducing fertilizer application by 10%, 30%, and 50%.

[0010] Preferably, in S4, the simulation of watershed water, sediment, nitrogen, and phosphorus loss processes under different scenarios is based on historical periods. By inputting precipitation and temperature data from different climate models, the growth parameters of rice and peanuts in the SWAT model are modified, and the application rates and timing of nitrogen and phosphorus fertilizers are adjusted to obtain the spatiotemporal variations of runoff, sediment, total nitrogen, and total phosphorus at the sub-basin scale under each scenario.

[0011] Preferably, the prediction model in S5 is constructed by using the XGBoost algorithm to iteratively generate multiple decision trees, which characterizes the nonlinear relationship of watershed hydrology and nutrient transport processes.

[0012] Preferably, the SHAP analysis method in S5 is based on the Shapley value principle of game theory.

[0013] Therefore, this application adopts the above-mentioned quantitative analysis method of watershed water, sediment, nitrogen, and phosphorus driving mechanisms. By identifying the driving factors affecting the loss of watershed water, sediment, nitrogen, and phosphorus, it can make up for the shortcomings of single models in the analysis of complex watershed systems. It provides more reliable scientific support for the precise control and adaptive strategy formulation of watershed water, sediment, nitrogen, and phosphorus loss, and also provides a reference paradigm for the quantitative research of similar complex watershed ecological and environmental problems. It has profound significance for optimizing watershed ecological and environmental management and ensuring the sustainable development of watersheds. Attached Figure Description

[0014] Figure 1 This is a flowchart illustrating the overall process of a quantitative analysis method for watershed water, sediment, nitrogen, and phosphorus driving mechanisms as described in this application. Figure 2The figures shown are the interpretability analysis results of the prediction model in the embodiments of this application, wherein (a) is the interpretability analysis result of the changes in runoff, (b) is the interpretability analysis result of the changes in sediment, (c) is the interpretability analysis result of the changes in total nitrogen, and (d) is the interpretability analysis result of the changes in total phosphorus. Detailed Implementation

[0015] The following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely illustrates selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0016] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning as understood by a person of ordinary skill in the art to which this application pertains.

[0017] The terms "comprising" or "including," as used in this application, mean that the element preceding the term encompasses the element listed after the term, and do not exclude the possibility of encompassing other elements as well. The terms "inner," "outer," "upper," and "lower," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application. When the absolute position of the described object changes, the relative positional relationship may also change accordingly. In this application, unless otherwise expressly specified and limited, the term "attached," etc., should be interpreted broadly. For example, it can refer to a fixed connection, a detachable connection, or an integral part; it can refer to a direct connection or an indirect connection through an intermediate medium; it can refer to the internal communication of two elements or the interaction relationship between two elements. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0018] Example 1: A quantitative analysis method for watershed water, sediment, nitrogen, and phosphorus driving mechanisms, such as Figure 1 As shown, it includes the following steps: S1: Obtain basic data for the target watershed, including digital elevation model data, land use data, soil data, meteorological data, water, sediment, nitrogen and phosphorus data, climate change data, and agricultural management data; Specifically, the data collected in this application are detailed in Table 1. 30m DEM data were obtained from the geospatial data cloud; land use data for the target watershed in 2005 were obtained from the land use remote sensing monitoring dataset; daily meteorological data included precipitation, maximum and minimum temperatures, relative humidity, and wind speed; and water, sediment, nitrogen, and phosphorus data included monthly runoff from 1997 to 2009, monthly sediment from 2017 to 2022, monthly total nitrogen and total phosphorus from 2018 to 2023, and daily outflow data from the reservoir. To assess the impacts of climate change and agricultural management practices on water, sediment, nitrogen, and phosphorus, CMIP6 ensemble data from five models, as well as data on planting structure and fertilization, were obtained.

[0019] Table 1

[0020] S2: Construct a SWAT model for the target watershed based on basic data, perform parameter sensitivity analysis on the model to identify key sensitive parameters, use the SUFI2 algorithm to complete the model parameter calibration and verification based on the key sensitive parameters, and use the coefficient of determination, Nash-Sercliff efficiency and percentage of bias to comprehensively evaluate the model simulation results. Specifically, this application constructs a SWAT model for the target watershed based on DEM, land use, soil, and meteorological and hydrological data. Monthly data on water, sediment, nitrogen, and phosphorus were collected and used for model calibration and validation. To accurately select parameters that significantly affect the simulation results of water, sediment, nitrogen, and phosphorus in the target watershed, parameter sensitivity analysis was first conducted using SWAT-Cup software to identify key sensitive parameters, thereby optimizing the calibration process and improving model efficiency. Subsequently, based on the selected sensitive parameters, the SUIF2 algorithm in the software was used to complete the model parameter calibration and validation. Simultaneously, the coefficient of determination (R²) was selected. 2 The model uses three evaluation metrics—Nash-Sercliffe efficiency (NS) and percentage bias (PBIAS)—to comprehensively assess the consistency, accuracy, and degree of bias in the simulation results, ensuring that the established SWAT model for the target watershed has reliable simulation capabilities. 2 The closer NS is to 1, the stronger the linear correlation between simulated and observed values. The closer NS is to 1, the higher the accuracy of the model simulation. The theoretical optimal value of PBIAS is 0. The smaller the absolute value, the lower the overall bias of the simulation.

[0021] The coefficient of determination is expressed as: ; In the formula, Let i be the observation value at time i. Let be the predicted value at time i. This is the average value of the observations; Nash Sergey's efficiency is expressed as: ; In the formula, This is the average value of the observations; The percentage deviation is expressed as: .

[0022] S3: Based on collected climate change data and agricultural management data, set up scenarios for climate change, planting structure adjustment, and fertilizer reduction; Specifically, based on the collected data on climate change and agricultural management practices, three scenarios were set up: climate change, planting structure adjustment, and fertilizer reduction. Each scenario has a corresponding number, for a total of 83 numbers.

[0023] As shown in Table 2, SSP2-4.5 and SSP5-8.5, representing moderate and high emission scenarios, were selected. Based on these two emission pathways, integrated precipitation and temperature data were obtained from five global climate models of CMIP6. Three future periods (2025-2034, 2035-2044, and 2045-2054) were selected to analyze the impact of climate change on water, sediment, nitrogen, and phosphorus losses in the target watershed. To assess the impact of agricultural management practices, data on the planting area of ​​rice and peanuts and the total application of nitrogen and phosphorus fertilizers from 1995 to 2022 were systematically collected. Climate change and agricultural management data were input into the established SWAT model. Growth parameters and fertilization management schemes for rice and peanuts were set separately, including the application rates and timing of nitrogen and phosphorus fertilizers, to simulate watershed water, sediment, nitrogen, and phosphorus losses under different climate scenarios and agricultural practices.

[0024] Table 2

[0025] S4: Input the scenario data into the validated SWAT model, and by modifying the crop growth parameters, nitrogen and phosphorus fertilizer application rate and application time in the SWAT model, simulate the loss process of water, sediment, nitrogen and phosphorus in the watershed under different scenarios with historical periods as the baseline scenario, obtain the spatiotemporal change results under each scenario and draw a spatial distribution map. Specifically, after completing model construction, calibration, validation, and scenario setting, the validated SWAT model was used to simulate water, sediment, nitrogen, and phosphorus transport in the target watershed under changing environments. The model used historical periods as baseline scenarios and, by inputting different climate changes, modified crop growth parameters and nitrogen and phosphorus fertilizer application rates to obtain the spatiotemporal variations of water, sediment, nitrogen, and phosphorus transport in the watershed under each scenario, and plotted spatial maps of water, sediment, nitrogen, and phosphorus loss in the target watershed.

[0026] S5: Based on the simulation results of each scenario, the scenario number, sub-basin information, and baseline scenario water, sediment, nitrogen, and phosphorus data are selected as input features, and the rate of change of each variable relative to the baseline period is used as the output variable to construct a prediction model. The prediction model is then validated using the coefficient of determination and root mean square error. The SHAP analysis method is used to calculate the marginal contribution of each input feature in different combinations. The contribution value of each feature to the model prediction results is obtained by weighted averaging and the importance is ranked to quantify the degree of influence of each driving factor on the loss of water, sediment, nitrogen, and phosphorus in the basin. Specifically, based on the results of the SWAT model scenario simulation, scenario number, information on 47 sub-basins, and baseline scenario water, sediment, nitrogen, and phosphorus data were selected as input features. The rate of change of each variable relative to the baseline period was used as the output variable. The XGBoost algorithm was employed to iteratively generate multiple decision trees to construct a model, characterizing the complex nonlinear relationships of watershed hydrology and nutrient transport processes. R² and RMSE were used to validate the model. Subsequently, the SHAP analysis method based on the Shapley value principle of game theory was used to calculate the marginal contribution of each feature in different combinations and perform a weighted average, quantifying the contribution and ranking of each feature to the model's prediction results. Figure 2 As shown, A, B, and C represent scenarios of reduced climate, planting structure, and fertilizer application, respectively. A1-A6 correspond to different climate paths and times, including SSP2-4.5 (A1, A2, A3) and SSP5-8.5 (A4, A5, A6), corresponding to the three periods of 2025-2034, 2035-2044, and 2045-2054, respectively. B1 and B2 represent two planting structures of rice and peanuts in the entire watershed. C1, C2, and C3 represent three fertilization levels of 10%, 30%, and 50% reduction in conventional fertilization, respectively.

[0027] S6: Based on the spatial distribution characteristics simulated by the SWAT model, identify the watershed water, sediment, nitrogen, and phosphorus loss patterns, combine the SHAP quantitative analysis results to screen the scenarios of loss changes, analyze the impact mechanism of planting structure and fertilization pattern on loss, and propose a zoned management strategy for watershed water, sediment, nitrogen, and phosphorus in high loss areas to achieve refined management of the watershed water ecological environment.

[0028] Specifically, based on spatial distribution characteristics, the patterns of water, sediment, nitrogen, and phosphorus loss in the target watershed are identified. Scenario changes in loss are screened using SHAP quantification results, and planting structure and fertilization patterns are analyzed based on existing watershed management data. Based on the above analysis, targeted zoned management strategies are proposed: under historical baseline and future scenarios, precise management is implemented in areas with high nitrogen and phosphorus loss loads, achieving refined management of the watershed's aquatic ecological environment through optimizing planting structure and regulating fertilization amounts.

[0029] Therefore, this application adopts the above-mentioned quantitative analysis method for the watershed water, sediment, nitrogen and phosphorus driving mechanism. By combining the SWAT model with machine learning methods for simulation and quantification, it can not only accurately depict the spatial pattern of water, sediment, nitrogen and phosphorus loss, but also quantify the contribution of different scenarios to the loss intensity, thereby accurately locating high-risk areas and key driving factors, avoiding the subjectivity and one-sidedness of traditional methods.

[0030] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application and not to limit them. Although this application has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of this application, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of this application.

Claims

1. A quantitative analysis method for the driving mechanisms of water, sediment, nitrogen, and phosphorus in watersheds, characterized in that, Includes the following steps: S1: Obtain basic data for the target watershed, including digital elevation model data, land use data, soil data, meteorological data, water, sediment, nitrogen and phosphorus data, climate change data, and agricultural management data; S2: Construct a SWAT model of the target watershed based on the basic data of the target watershed, perform parameter sensitivity analysis on the model to identify key sensitive parameters, use the SUFI2 algorithm to complete the model parameter calibration and verification based on the key sensitive parameters, and use the coefficient of determination, Nash-Sercliff efficiency and percentage of bias to comprehensively evaluate the model simulation results. S3: Based on climate change data and agricultural management data, set up scenarios of climate change, planting structure adjustment and fertilizer reduction to obtain scenario data; S4: Input the scenario data into the validated target watershed SWAT model, and by modifying the crop growth parameters, nitrogen and phosphorus fertilizer application rate and application time in the target watershed SWAT model, simulate the watershed water, sediment, nitrogen and phosphorus loss process under different scenarios with historical periods as the baseline scenario, obtain the spatiotemporal change results under each scenario and draw a spatial distribution map. S5: Based on the simulation results of each scenario, the scenario number, sub-basin information, and baseline scenario water, sediment, nitrogen, and phosphorus data are selected as input features, and the rate of change relative to the baseline period is used as the output variable to construct a prediction model. The prediction model is then validated using the coefficient of determination and root mean square error. The SHAP analysis method is used to calculate the marginal contribution of each input feature in different combinations. The contribution value of each feature to the model prediction results is obtained by weighted averaging and ranked to quantify the influence of each driving factor on the loss of water, sediment, nitrogen, and phosphorus in the basin. S6: Based on the spatial distribution characteristics simulated by the SWAT model, identify the watershed water, sediment, nitrogen, and phosphorus loss patterns, combine the SHAP quantitative analysis results to screen the scenarios of loss changes, analyze the impact mechanism of planting structure and fertilization pattern on loss, and propose a zoned management strategy for watershed water, sediment, nitrogen, and phosphorus in high loss areas to achieve refined management of the watershed water ecological environment.

2. The method for quantitative analysis of watershed water, sediment, nitrogen, and phosphorus driving mechanisms according to claim 1, characterized in that, Meteorological data in S1 includes precipitation, maximum temperature, minimum temperature, relative humidity, and wind speed; water, sediment, nitrogen, and phosphorus data include monthly runoff, monthly sediment, monthly total nitrogen, and monthly total phosphorus monitoring data; climate change data includes precipitation and temperature data from the CMIP6 integrated data; and agricultural management data includes crop planting area, planting structure, nitrogen fertilizer application rate, phosphorus fertilizer application rate, and fertilization time.

3. The method for quantitative analysis of watershed water, sediment, nitrogen, and phosphorus driving mechanisms according to claim 2, characterized in that, In S2, parameter sensitivity analysis and model parameter calibration verification were performed using SWAT-CUP software.

4. The method for quantitative analysis of watershed water, sediment, nitrogen, and phosphorus driving mechanisms according to claim 3, characterized in that, The coefficient of determination is expressed as: ； In the formula, Let i be the observation value at time i. Let be the predicted value at time i. This is the average value of the observations; Nash Sergey's efficiency is expressed as: ； In the formula, This is the average value of the observations; The percentage deviation is expressed as: 。 5. The method for quantitative analysis of watershed water, sediment, nitrogen, and phosphorus driving mechanisms according to claim 4, characterized in that, The S3 climate change scenarios specifically include: based on the integrated data of five global climate models in CMIP6, the SSP2-4.5 intermediate emission scenario and the SSP5-8.5 high emission scenario are selected, with three future periods set for each emission pathway, forming a total of six climate change scenarios.

6. The method for quantitative analysis of watershed water, sediment, nitrogen, and phosphorus driving mechanisms according to claim 5, characterized in that, The planting structure adjustment scenario includes planting rice and peanuts on all cultivated land in the entire watershed; the fertilizer reduction scenario includes reducing fertilizer application by 10%, 30%, and 50%.

7. The method for quantitative analysis of watershed water, sediment, nitrogen, and phosphorus driving mechanisms according to claim 6, characterized in that, S4 simulates the loss of water, sediment, nitrogen, and phosphorus in watersheds under different scenarios. It uses historical periods as the baseline scenario and inputs precipitation and temperature data from different climate models. It modifies the growth parameters of rice and peanuts in the SWAT model and adjusts the application amount and time of nitrogen and phosphorus fertilizers to obtain the spatiotemporal variation results of runoff, sediment, total nitrogen, and total phosphorus at the sub-watershed scale under each scenario.

8. The method for quantitative analysis of watershed water, sediment, nitrogen, and phosphorus driving mechanisms according to claim 7, characterized in that, The prediction model in S5 is constructed by iteratively generating multiple decision trees using the XGBoost algorithm to characterize the nonlinear relationships of watershed hydrology and nutrient transport processes.

9. A method for quantitative analysis of watershed water, sediment, nitrogen, and phosphorus driving mechanisms according to claim 8, characterized in that, The SHAP analysis method in S5 is based on the Shapley value principle of game theory.

Images