Method and device for evaluating influence of human activities on drainage basin total phosphorus pollution

Abstract

The invention relates to the technical field of material testing and analysis, in particular to a method and device for evaluating the influence of human activities on drainage basin total phosphorus pollution, and the method comprises the steps: obtaining influence factors related to the human activities in a drainage basin; carrying out collinearity screening on the influence factors to obtain candidate factors; based on a reference total phosphorus concentration and the candidate factor, determining a grey correlation degree between the candidate factor and the reference total phosphorus concentration; screening the candidate factors based on the grey relational degree to obtain a target factor; and predicting the total phosphorus pollution influence of the target factor on the drainage basin by adopting a geographically weighted random forest model to obtain a prediction result. According to the method, the prediction precision of the geographically weighted random forest model on the local region is improved, so that the influence of the target factors related to human activities on the total phosphorus pollution of the drainage basin can be evaluated more accurately.

Description

Technical Field

[0001] The invention relates to the technical field of material testing and analysis, and in particular to a method and a device for evaluating the impact of human activities on total phosphorus pollution in a watershed in the technical field of material testing and analysis. Background Art

[0002] In related technologies, especially in applications based on statistical models or machine learning models, there are strong collinearity parameters between multiple human activity factors that affect total phosphorus pollution. This will lead to unstable prediction results of the model and it is difficult to effectively judge the true contribution of each factor. Traditional factor screening methods have failed to fully deal with this problem, thus affecting the accuracy and reliability of the model. Most current studies use empirical or simple correlation analysis to screen human activity factors that may affect total phosphorus pollution. This method fails to systematically and scientifically extract the factors most relevant to total phosphorus pollution from a large number of candidate factors, which may cause some important factors to be ignored or irrelevant factors to be incorrectly selected, further reducing the prediction performance of the model. Traditional statistical methods have limited ability to handle complex and nonlinear spatial watershed pollution, and although geographically weighted regression models and some machine learning models combined with spatial analysis take into account regional differences, their limitations are that these methods often fail to fully consider the mutual influence between monitoring points in different control units in the watershed when constructing local models. This may lead to inaccurate prediction results in local areas, especially in watersheds with complex total phosphorus pollution and large spatial heterogeneity, the prediction performance of the model still needs to be improved. When conducting local model training, the relevant technology did not take differentiated treatment of the monitoring points within the same control unit, and failed to fully consider the interaction between points within the control unit, making it impossible to reflect the local pollution situation in a refined manner. Summary of the invention

[0003] The purpose of the present invention is to provide a method and device for evaluating the impact of human activities on total phosphorus pollution in a watershed. The technical scheme adopted is as follows: In a first aspect, an embodiment of the present invention provides a method for evaluating the impact of human activities on total phosphorus pollution in a watershed, the method comprising: Obtain influencing factors related to human activities in the watershed; Performing collinearity screening on the influencing factors to obtain candidate factors; Based on the reference total phosphorus concentration and the candidate factor, determining the grey correlation degree between the candidate factor and the reference total phosphorus concentration; Screening the candidate factors based on the grey relational degree to obtain the target factor; A geographically weighted random forest model is used to predict the impact of the target factor on the total phosphorus pollution in the watershed to obtain a prediction result; wherein, in the geographically weighted random forest model, the first weight of the monitoring points within the same control unit is greater than the second weight of the monitoring points outside the control unit.

[0004] In a second aspect, an embodiment of the present invention provides a device for evaluating the impact of human activities on total phosphorus pollution in a watershed, the device comprising: The acquisition module is used to obtain the influencing factors related to human activities in the watershed; A first screening module is used to perform collinearity screening on the influencing factors to obtain candidate factors; A determination module, used for determining the grey correlation degree between the candidate factor and the reference total phosphorus concentration based on the reference total phosphorus concentration and the candidate factor; A second screening module is used to screen the candidate factors based on the grey relational degree to obtain the target factor; A prediction module is used to use a geographically weighted random forest model to predict the impact of the target factor on the total phosphorus pollution in the watershed and obtain a prediction result; wherein, in the geographically weighted random forest model, the first weight of the monitoring points within the same control unit is greater than the second weight of the monitoring points outside the control unit.

[0005] In a third aspect, a computer program product is provided, comprising: a computer program code, which, when executed on a computer, enables the computer to execute the method in the first aspect or any possible implementation manner of the first aspect.

[0006] In a fourth aspect, a computer-readable storage medium is provided, which stores a computer program code. When the computer program code runs on a computer, the computer executes the method in the first aspect or any possible implementation manner described in the first aspect.

[0007] The present invention has the following beneficial effects: after obtaining the influencing factors related to human activities in the watershed, the influencing factors are subjected to collinearity screening to obtain candidate factors, thereby effectively reducing the collinearity problem in the influencing factors related to human activities, so that the collinearity between the obtained candidate factors is relatively weak. Afterwards, based on the reference total phosphorus concentration and the candidate factors, the grey correlation between the candidate factors and the reference total phosphorus concentration is determined, and the candidate factors are screened based on the grey correlation to obtain the target factors; in this way, by calculating the grey correlation between each candidate factor and the reference total phosphorus concentration, the target factors that have an important impact on the reference total phosphorus concentration can be screened out from the candidate factors. Finally, the geographically weighted random forest model is used to predict the impact of the target factor on the total phosphorus pollution in the watershed to obtain a prediction result; wherein, in the geographically weighted random forest model, the first weight of the monitoring points within the same control unit is greater than the second weight of the monitoring points outside the control unit. In this way, since the geographically weighted random forest model gives greater weight to the monitoring points within the same control unit, it fully considers the interaction between the monitoring points, especially the impact between the monitoring points within the unit, thereby improving the prediction accuracy of the geographically weighted random forest model for local areas, and thus being able to more accurately evaluate the impact of target factors related to human activities on the total phosphorus pollution in the basin. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0009] Figure 1 It is a schematic diagram of the implementation process of a method for evaluating the impact of human activities on total phosphorus pollution in a watershed provided by an embodiment of the present invention; Figure 2 It is a schematic diagram of the result of preprocessing the watershed data of Province A provided by an embodiment of the present invention; Figure 3 It is another implementation flow diagram of a method for evaluating the impact of human activities on total phosphorus pollution in a watershed provided by an embodiment of the present invention; Figure 4 It is a result schematic diagram of the variance inflation factor of each influencing factor provided by an embodiment of the present invention; Figure 5 This is another implementation flow diagram of a method for evaluating the impact of human activities on total phosphorus pollution in a watershed provided by an embodiment of the present invention; Figure 6 is a schematic diagram of the result of the grey relational degree of the candidate factors provided by an embodiment of the present invention; Figure 7 is an importance distribution diagram of each control unit within the province A provided by an embodiment of the present invention; Figure 8 is a bar chart of the importance ranking of each target factor provided by an embodiment of the present invention; Fig. 9 It is a schematic diagram of the composition structure of a device for evaluating the impact of human activities on total phosphorus pollution in a watershed provided by an embodiment of the present invention; Fig.10 It is a structural schematic diagram of a computer device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0010] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the method for evaluating the impact of human activities on total phosphorus pollution in a watershed proposed by the present invention, its specific implementation method, structure, characteristics and effects, in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form as described.

[0011] Among them, in the description of the embodiments of the present invention, unless otherwise specified, " / " means or, for example, A / B can mean A or B: "and / or" in the text is only a way to describe the association relationship of associated objects, indicating that there can be three relationships, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present invention, "multiple" refers to two or more than two.

[0012] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features.

[0013] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0014] Total phosphorus (TP) pollution in watersheds is an important issue in water environment management, which is directly related to water quality safety and ecosystem health. As a key nutrient element in water bodies, excessive reference total phosphorus concentrations can lead to eutrophication of water bodies, cause excessive growth of algae, and destroy the habitat of aquatic organisms, thus posing a threat to the sustainable use of water resources. There are many factors that affect the reference total phosphorus concentration, including both natural and human factors. In natural processes, soil erosion and the decomposition of aquatic plants and animals contribute to the phosphorus content in water bodies, but compared with human activities, such natural sources are usually slower and more limited. In contrast, the impact of human activities on the reference total phosphorus concentration is more direct and significant, especially agricultural fertilization, industrial emissions, livestock waste and urban sewage discharge, which greatly increase the phosphorus content in water bodies. Therefore, it is crucial to evaluate the impact of human activities on total phosphorus pollution in watersheds and propose effective prevention and control measures for protecting the water environment.

[0015] The total phosphorus pollution assessment techniques in related technologies mainly include traditional statistical methods, physical models, and spatial statistics and regression methods. Traditional statistical methods such as linear regression and correlation analysis are often used to study the relationship between human activities and total phosphorus pollution. Such methods rely on the linear relationship between factors and can provide quantitative assessments. However, when faced with complex nonlinear relationships between multiple factors, the explanatory power of traditional statistical methods is limited. On the other hand, physical models such as the Soil and Water Assessment Tool (SWAT) can predict the spatiotemporal distribution of total phosphorus pollution by simulating hydrological processes and the migration and diffusion of pollutants, combined with factors such as basin topography, climate, and land use. However, physical models usually require a large amount of field data support, and the calculations are complex and the parameter adjustment is cumbersome. In order to overcome the problem of insufficient spatial analysis, spatial statistics and regression methods such as Geographically Weighted Random Forest (GWRF) have been introduced into the assessment of total phosphorus pollution in the basin. GWR can take into account spatial heterogeneity, that is, the concentration of pollutants in different regions is affected by different human activities, which to a certain extent makes up for the shortcomings of regression models in spatial analysis. However, GWR still faces problems such as complex parameter settings and susceptibility to collinear parameters. In addition, most of the research in related technologies is based on a single data source, and the actual situation is often more complicated, so the assessment of the impact of total phosphorus pollution is not accurate enough. Moreover, due to the complex spatial interactions between the monitoring points in the basin, especially the influence between points in the same control unit, ignoring this point will result in the model being unable to accurately reflect the local pollution situation.

[0016] Based on this, the embodiment of the present invention provides an evaluation method for the impact of human activities on total phosphorus pollution in a watershed, combining and improving methods such as variance inflation factor (VIF), grey relation analysis (GRA), and GWRF to clarify the source and distribution of total phosphorus pollution in the water environment of the watershed. The embodiment of the present invention collects data on a variety of human activity elements, performs preprocessing, and uses variance inflation factor (VIF) to effectively solve the collinearity problem between influencing factors, screens out factors that have an impact on total phosphorus pollution and are independent of each other, to ensure the stability and accuracy of the analysis method.

[0017] At the same time, combined with the grey correlation analysis method, the candidate factors screened out by collinearity are further screened to extract target factors with stronger correlation with total phosphorus pollution concentration, ensuring the relevance and interpretability of the target factors input into the model, thereby improving the prediction ability.

[0018] In addition, the embodiment of the present invention introduces a geographically weighted random forest model, which can not only fully consider the differential impact of human activities in different regions on the reference total phosphorus concentration, but also combines the nonlinear processing capabilities of random forests, significantly improving the accuracy of total phosphorus pollution concentration prediction. The GWRF model can perform more accurate spatial heterogeneity analysis at the global and local levels. During the model training and prediction process, greater weights are given to monitoring points within the same control unit, and the interactions between monitoring points, especially the impact of monitoring points within the unit, are fully considered, thereby improving the prediction accuracy of local areas and enabling the model to better reflect the local characteristics of total phosphorus pollution. Finally, by analyzing and evaluating the results of the GWRF method, target factors that have a greater impact on total phosphorus pollution can be accurately identified from complex human activities, thereby providing a more scientific and effective basis for water environment management.

[0019] The following is a detailed description of a method for evaluating the impact of human activities on total phosphorus pollution in a watershed provided by the present invention in conjunction with the accompanying drawings. Figure 1 , which shows a schematic diagram of the implementation process of a method for evaluating the impact of human activities on total phosphorus pollution in a watershed provided by an embodiment of the present invention, the method comprising: 101, obtain the impact factors related to human activities in the basin.

[0020] Here, the influencing factors related to human activities in the basin may include: regional GDP, primary industry, secondary industry, tertiary industry, agricultural fertilizer application, population density, domestic sewage discharge, industrial wastewater discharge, land use type, etc.

[0021] In some possible implementations, first, the initial factors related to human activities in the watershed are obtained; wherein the data type of the initial factors is the same as the data type of the influencing factors. The initial factors can be obtained through public statistical data, remote sensing image data, and field measurements. Secondly, the initial factors are preprocessed to obtain processed factors; for example, after the data collection is completed, the initial factors are processed by missing value filling, outlier processing, and data standardization to obtain processed factors. Finally, based on the preset resolution, the processed factors are projected to obtain the influencing factors. For example, the preset resolution is 30×30; in this way, the processed factors are uniformly projected into a conical equal area projection (Albers_Conic_Equal_Area) with a uniform resolution of 30×30, thereby obtaining the influencing factors related to human activities in the watershed, so that the influencing factors can be used as input for subsequent steps.

[0022] In a specific example, taking Province A as an example, the data used in the embodiment of the present invention mainly include total phosphorus data of water quality monitoring at the control section, control unit data, and various human activity data (i.e., influencing factors related to human activities in the basin). The acquisition source and preprocessing process of the relevant data are as follows: First, obtain the total phosphorus data of water quality monitoring in the control section, control unit and zoning data of Province A.

[0023] Here, the total phosphorus data of water quality monitoring comes from the Planning Bureau, which records the annual average total phosphorus values ​​of 212 national and provincial controlled sections within the watershed of Province A from January to October 2023. The control unit and the regional division data of Province A come from the Planning Bureau, which is based on the analysis of the division of control units in the watershed spatial management system, with the control section as the node, and the spatial scope defined according to multiple factors such as the natural confluence characteristics of surface water, the distribution of pollution sources and administrative divisions.

[0024] Secondly, the initial factors related to human activities in Province A.

[0025] Here, the human factor statistics of Province A are derived from the 2023 Statistical Yearbook released by the Statistics Bureau of Province A. The influencing factors related to human activities that can be used include: Gross Regional Product (GRP), Primary Industry (PI), Secondary Industry (SI), Tertiary Industry (TI), Per capita GDP (PCGDP), Permanent Population (PP), Urban Population (UP), Rural Population (RP), Number of Construction Enterprises (NCE), Total Construction Output (TCO), Cultivated Land Area (CLA), Irrigated Land Area (ILA), Value Added of Agriculture, Forestry, Animal Husbandry and Fishery (AFAHVA), Agricultural Fertilizer Usage (AFU), Grain Sowing Area (GSA), Grain Output (GO), Oil Crop Output (OCO), Vegetable and Edible Mushroom Output (Vegetable and Edible Mushroom The data includes 28 types of data: annual output (VEMO), fruit output (FO), number of hogs slaughtered (SH), number of hogs on hand at the end of the year (YHI), number of sheep slaughtered (SS), number of poultry slaughtered (SP), number of industrial enterprises (NIE), industrial value-added growth (IVAG), industrial operating revenue (IOR), and total industrial economic profit (ITP). The data can analyze the impact of human activities such as economy, agriculture, and population in Province A on total phosphorus pollution from multiple dimensions. The embodiment of the present invention pre-processes the initial factors related to human activities in 28 watersheds in the statistical yearbook.

[0026] Finally, the influencing factors related to human activities in the watershed of Province A (i.e., human element raster data) are obtained.

[0027] Here, the land cover data comes from the Institute of Remote Sensing Information Processing. The dataset covers 9 types of land cover with a spatial resolution of 30 meters from 1990 to 2022, including cultivated land, forests, impervious surfaces, etc. For the land cover data in 2022, a binary classification method is used to screen out three types of land cover-cultivated land, forests and impervious surfaces. The specific method is to set the three types of land use, namely cultivated land, forests and impervious surfaces, to 1, and set other types of land cover to 0. Since adjacent areas of different land cover types have an overlapping impact on water quality, and various types of land cover do not share spatial boundaries, this feature cannot be fully reflected in water quality assessment. To solve this problem, the embodiment of the present invention uses the aggregation tool of ArcGIS 10.2 to aggregate the land cover types with a resolution of 30 meters to a scale of 1 kilometer through mean statistics to show the overlapping effect between different land cover types.

[0028] The population density data comes from the World Population Project. This dataset provides global population raster data with a resolution of 100 meters from 2000 to 2020. It is one of the most widely used population distribution datasets. This dataset not only has the advantages of high resolution and long-term series, but also has high accuracy.

[0029] The factory point of interest data comes from the Point of Interest (POI) dataset of the map. During the production process, factories often discharge wastewater and harmful substances into surrounding water bodies, such as rivers, lakes and oceans, which in turn affects the quality of the water environment. To evaluate this impact, the embodiment of the present invention first screens the factory's point of interest data and performs data cleaning (such as deduplication and screening). Then, using the kernel density analysis tool of ArcGIS 10.2, the bandwidth is set to 5 kilometers to calculate the kernel density distribution of the factory. This data not only shows the spatial distribution of factories, but also reflects the degree of factory agglomeration, providing strong support for water environment impact analysis.

[0030] In the embodiment of the present invention, all the impact factors obtained by preprocessing are converted into Albers_Conic_Equal_Area projection, and the pixels with a spatial resolution of 30*30 meters (m) are resampled using bilinear interpolation to finally obtain the impact factors. Figure 2 As shown, Figure 2 (a) to Figure 2(m) represent fertilizer use (AFU), sheep output (SS), poultry output (SP), secondary industry (SI), beef cattle output (SC), population density (POP), factory density (POI), primary industry (PI), oilseed output (OCO), factory density (IOR), impervious surface area (Impervious), fruit output (FO) and arable land (Cropland).

[0031] 102. Perform collinearity screening on the influencing factors to obtain candidate factors.

[0032] Here, the influencing factors are screened by analyzing the collinearity between the influencing factors to obtain candidate factors.

[0033] In some possible implementations, the above step 102 can be performed by Figure 3 The steps shown achieve: 301, determine the collinearity parameters between each of the influencing factors and other influencing factors.

[0034] Here, by calculating the variance inflation factor between each influencing factor and other influencing factors, and determining the variance inflation factor as the collinearity parameter, the collinearity parameter between each influencing factor and other influencing factors can be accurately calculated, so as to facilitate the screening of influencing factors.

[0035] In some possible implementations, the more than 30 influencing factors obtained after preprocessing are first processed using the VIF method to obtain candidate factors. Since there may be a strong collinearity relationship between different influencing factors in the input data, this will lead to instability and difficulty in interpretation of the prediction model. In order to solve this problem, the embodiment of the present invention first uses the VIF method to perform collinearity detection on the input variables (i.e., the input influencing factors). The basic idea of ​​VIF is to measure the multicollinearity between variables by evaluating whether a certain explanatory variable can be linearly explained by other explanatory variables. The variance inflation factor of the i-th variable (i.e., the i-th influencing factor) As shown in formula (1): (1); in, is the coefficient of determination obtained when the i-th influencing factor is used as the dependent variable and the other influencing factors are used as independent variables for regression. If the value is too high, it means that there is multicollinearity between the impact factors. Calculate the VIF value of each impact factor. The VIF value reflects the linear correlation between a certain impact factor and other impact factors. When the VIF value is high, it indicates that the impact factor has a strong correlation with other impact factors.

[0036] 302: Screen the influencing factors based on the collinearity parameters to obtain the candidate factors.

[0037] In some possible implementations, among the influencing factors, the influencing factors whose variance inflation factors are greater than a preset inflation threshold are eliminated to obtain the candidate factors. The preset inflation threshold may be a custom setting. For the influencing factors whose variance inflation factors exceed the preset inflation threshold, it is determined whether the influencing factors have a significant impact on the model prediction. If the VIF value of the influencing factor is too high and its contribution to the model is small, the influencing factor is eliminated to reduce the collinearity between the obtained candidate factors. In this way, the VIF method is used to exclude influencing factors with high collinearity from more than 30 influencing factors to obtain candidate factors. After multiple collinearity checks, the variance inflation factors (VIF) of the candidate influencing factors (PI, SI, PP, NCE, TCO, AFU, OCO, VEMO, FO, SC, SS, SP, NIE, IVAG, IOR, ITP, POP, POI, Cropland, Forest, Impervious) are as follows: Figure 4 As shown, in Figure 4 The horizontal axis represents the value of the variance inflation factor, and the vertical axis represents each influencing factor. In this way, by calculating the collinearity parameters between the influencing factors, the collinearity between the selected candidate factors is weaker, so that the candidate factors are more conducive to improving the stability of the geographically weighted random forest model.

[0038] 103. Based on the reference total phosphorus concentration and the candidate factor, determine the grey correlation degree between the candidate factor and the reference total phosphorus concentration.

[0039] Here, the reference total phosphorus concentration is the reference total phosphorus concentration of each monitoring point in the statistical basin. By comparing the reference total phosphorus concentration at each time in the reference total phosphorus concentration with the candidate factors at each time, the grey correlation between the candidate factors and the reference total phosphorus concentration is obtained.

[0040] In some possible implementations, the above step 103 can be performed by Figure 5 The steps shown achieve: 501. At any time, determine the difference between the reference total phosphorus concentration and the candidate factor to obtain a difference sequence.

[0041] Here, at each time, the difference between each reference total phosphorus concentration and each candidate factor in the reference total phosphorus concentration is calculated to obtain a difference sequence.

[0042] 502. Determine the grey correlation degree between the reference total phosphorus concentration and the candidate factor based on the maximum difference and the minimum difference in the difference sequence.

[0043] Here, after solving the collinearity problem, the GRA method is used to perform correlation analysis on the candidate factors to further narrow the range of candidate factors that affect the reference total phosphorus concentration. GRA is an analytical method that can handle the interaction between multiple factors in complex systems. By calculating the gray correlation between each candidate factor and the reference total phosphorus concentration, the candidate factor most correlated with the reference total phosphorus concentration is screened out. Among them, the gray correlation between the i-th candidate factor and the reference total phosphorus concentration is , as shown in formula (2): (2); in, is the value of the reference total phosphorus concentration at the kth moment, is the value of the i-th candidate factor at the k-th moment, ρ is the resolution coefficient (usually 0.5), and n is the number of moments. In this way, by calculating the difference between the reference total phosphorus concentration and the candidate factor at each moment, the grey correlation between the candidate factor and the reference total phosphorus concentration can be accurately analyzed.

[0044] In a specific example, taking Province A as an example, within the scope of Province A, the grey correlation degree corresponding to each candidate factor is as follows: Figure 6 As shown, in Figure 6 The horizontal axis represents the size of the gray correlation degree corresponding to each candidate factor, and the vertical axis represents each candidate factor. Combined with the analysis results of GRA, 13 main influencing factors (i.e. candidate factors) related to the reference total phosphorus concentration were finally screened out, namely: primary industry output value PI, secondary industry output value SI, agricultural fertilizer application amount AFU, sheep output SS, poultry output number SP, beef cattle output SC, industrial revenue IOR, fruit output FO, oil output OCO, cultivated land area (Cropland), population density POP, factory density POI, and impervious surface area Impervious.

[0045] 104. Screen candidate factors based on the grey relational degree to obtain target factors.

[0046] Here, candidate factors are screened according to the size of the grey correlation degree to obtain target factors that have a stronger impact on the reference total phosphorus concentration.

[0047] In some possible implementations, among the candidate factors, target candidate factors whose grey correlation is greater than a preset impact threshold are screened; and the target candidate factors are determined as the target factors. The preset impact threshold may be a custom value. For each candidate factor, the grey correlation value between it and each reference total phosphorus concentration in the reference total phosphorus concentration is calculated. The greater the correlation, the more significant the impact of the candidate factor on the reference total phosphorus concentration. According to the size of the grey correlation, the candidate factors that have an important impact on the reference total phosphorus concentration are further screened as input for subsequent steps. In this way, the subjectivity of candidate factor selection is avoided through the GRA method, ensuring that the model can focus on more critical candidate factors when predicting total phosphorus pollution.

[0048] 105. Use a geographically weighted random forest model to predict the impact of the target factor on the total phosphorus pollution in the watershed and obtain a prediction result.

[0049] Among them, the first weight of the monitoring points within the same control unit in the geographically weighted random forest model is greater than the second weight of the monitoring points outside the control unit.

[0050] In some possible implementations, the training process of the Geographically Weighted Random Forest model is as follows: The first step is to divide the watershed into multiple control units based on the hydrological characteristics, topography and water system distribution in the watershed.

[0051] Here, in order to finely analyze the impact of target factors of different human activities in the watershed on the reference total phosphorus concentration, the embodiment of the present invention introduces a geographically weighted random forest model. The core idea of ​​the GWRF model is to combine the spatial weighting mechanism with the decision tree integration model, so that the model can not only capture the spatial differences between different regions, but also handle the complex nonlinear relationship between the target factors and the reference total phosphorus concentration. Specifically, the GWRF model first assigns different weights to the input influencing factors according to the spatial position of the monitoring point; among them, the distribution of weights depends on the spatial distance between the monitoring points, the closer the distance, the greater the weight of the point, and the farther the distance, the gradually attenuated weight of the point. This weighting mechanism ensures that the model can capture spatial heterogeneity, that is, the reference total phosphorus concentration in different areas of the watershed may be affected by different target factors related to human activities. The basic formula of GWRF can be expressed as shown in formula (3): (3); in, is the predicted result of the impact of the target factor on the reference total phosphorus concentration in the ith geographical location, is the independent variable at the ith position (eigenvector, i.e., the eigenvector of the target factor at the ith position), is the longitude and latitude coordinates of the ith position, It is a regression function based on geographical location weighting and is constructed using a random forest model. is the error term, which represents the noise of the regression model.

[0052] The biggest advantage of the GWRF model is that it can handle spatial heterogeneity and nonlinear problems at the same time, ensuring that the model has higher prediction accuracy and explanatory power when facing complex total phosphorus pollution in watersheds. GWRF can capture the spatial differences between different regions and improve the adaptability of the model at different spatial scales. Compared with a single random forest model, GWRF effectively overcomes the defect of random forests ignoring the influence of spatial location when processing spatial data by introducing spatial weights. However, the problem with the GWRF model is that when facing local areas, it fails to fully consider the strong interactions between monitoring points within the control unit. This interaction is particularly evident in local areas within some watersheds, especially in areas such as control units that are divided based on hydrological characteristics and geographical boundaries, where monitoring points are often highly correlated. If these internal associations are ignored, the prediction accuracy of the model in local areas may decrease, making it difficult to accurately reflect complex local pollution characteristics.

[0053] In order to solve the problem that the GWRF model fails to fully consider the interaction between monitoring points when processing local areas, the embodiment of the present invention proposes a weight setting strategy for local model training and improves the GWRF model. The control section is a sampling section set up to monitor the water quality of a certain river basin or region. The control unit is a regional unit divided according to the hydrological characteristics, topography and water system distribution of the river basin with the control section as a node. The control section reflects the impact of human activities within the control unit on the water quality. The monitoring points within the same control unit have a strong correlation. Therefore, the embodiment of the present invention improves the GWRF model by dividing the river basin into multiple control units, thereby performing differentiated processing in model training.

[0054] In the second step, for each of the plurality of control units, a first weight of a monitoring point within the control unit and a second weight of a monitoring point outside the control unit are determined respectively.

[0055] Wherein, the first weight is greater than the second weight. In some possible implementations, first, for any monitoring point in each control unit, determine the spatial distance between any monitoring point and other monitoring points and the bandwidth of the control unit to which the other monitoring points belong; wherein the other monitoring points are in each control unit or outside each control unit. Here, for any monitoring point in a control unit, if another monitoring point is also in the control unit, then calculate the spatial distance between the two monitoring points and the bandwidth of the control unit; if the other monitoring point is outside the control unit, then calculate the spatial distance between the two monitoring points and the bandwidth of the control unit to which the other monitoring point is located.

[0056] Then, based on the spatial distance and the bandwidth, a first weight between the any monitoring point and other monitoring points in each control unit and a second weight between the any monitoring point and other monitoring points outside each control unit are determined.

[0057] Here, in the local model training, the embodiment of the present invention assigns a greater weight to the monitoring points within the same control unit, assigns a greater weight to the monitoring points within the same control unit based on the biquadratic kernel function, and assigns a smaller weight to the monitoring points outside the control unit, which can be achieved by setting different kernel bandwidth parameters for the internal and external points of the control unit. This weight adjustment strategy enables the model to more accurately reflect the interaction and influence of each monitoring point in the same control unit during local training, thereby improving the prediction accuracy of the local area. The first weight and the second weight can be calculated by formula (4): (4); in, is the weight between the i-th monitoring point and the j-th monitoring point in the control unit, is the spatial distance between two points, b is the kernel bandwidth parameter, which is used to control the speed of weight decay. is the bandwidth within the control unit, is the bandwidth outside the control unit, usually , to ensure that the points inside the unit have a greater influence and the weight decays more slowly; when When the corresponding bandwidth range is exceeded, the weight =0, that is, the monitoring point has no effect on the prediction results.

[0058] The third step is to train the geographically weighted random forest model to be trained based on the first weights and the corresponding monitoring points, and the second weights and the corresponding monitoring points, to obtain the geographically weighted random forest model.

[0059] Here, the first weight and the corresponding statistical data of the monitoring point, as well as the second weight and the corresponding statistical data of the monitoring point, are used as the input of the geographically weighted random forest model to be trained, so as to train the geographically weighted random forest model to be trained to obtain the geographically weighted random forest model. Through the segmented bandwidth processing method shown in formula (4), the spatial correlation between the monitoring points in the local area can be more accurately reflected, and the points inside and outside the control unit can be reasonably distinguished during the local model training, effectively improving the prediction accuracy of the local area. In this way, by improving the GWRF model to achieve refined estimation and analysis of the target factors of the reference total phosphorus concentration and related human activities in the basin, the main target factors related to human activities in each control unit can be accurately identified and analyzed.

[0060] In a specific example, taking Province A as an example, the embodiment of the present invention optimizes the GWRF model through local weight adjustment and applies it to various control units in Province A. Province A has complex terrain, covering a variety of water systems and river basins, and there are significant differences in the natural environment and human activities in the eastern and western regions. When training the local model, by setting different kernel bandwidth parameters, the embodiment of the present invention assigns greater weights to the monitoring points in the same control unit in Province A, which can more accurately reflect the impact of human activities in local areas on total phosphorus pollution. Such weight adjustment ensures that the GWRF model can better adapt to the complex river basin characteristics of Province A and improves the accuracy of total phosphorus pollution assessment.

[0061] In an embodiment of the present invention, after obtaining the influencing factors related to human activities in the watershed, the influencing factors are collinearly screened to obtain candidate factors, thereby effectively reducing the collinearity problem in the influencing factors related to human activities to reduce the collinearity between the candidate factors. Afterwards, by calculating the gray correlation between each candidate factor and the reference total phosphorus concentration, the target factor that has an important impact on the reference total phosphorus concentration can be screened out from the candidate factors. Finally, the geographically weighted random forest model is used to predict the impact of the target factor on the total phosphorus pollution in the watershed to obtain the prediction result; wherein, the first weight of the monitoring point in the same control unit in the geographically weighted random forest model is greater than the second weight of the monitoring point outside the control unit. In this way, since the geographically weighted random forest model gives a greater weight to the monitoring points in the same control unit, the interaction between the monitoring points, especially the impact between the monitoring points within the unit, is fully considered, thereby improving the prediction accuracy of the geographically weighted random forest model for local areas, and thus being able to more accurately evaluate the impact of the target factor on the total phosphorus pollution in the watershed.

[0062] In some embodiments, after predicting the impact of the target factor on the total phosphorus pollution in the watershed through the geographically weighted random forest model, the prediction results output by the geographically weighted random forest model can also be evaluated, which can be achieved by the following steps: First, obtain the statistical value of the impact of the target factor on the total phosphorus pollution in the watershed.

[0063] Here, the statistical value may be obtained by statistically analyzing each monitoring point in the watershed, for example, statistically analyzing the impact of the target factor on the total phosphorus pollution in the watershed at each monitoring point in the watershed.

[0064] Next, the difference between the statistical value and the predicted result is determined.

[0065] Here, the statistical value is subtracted from the predicted result output by the geographically weighted random forest model to obtain the difference between the statistical value and the predicted result.

[0066] Finally, based on the difference, the prediction result is evaluated to obtain an evaluation result.

[0067] Here, the root mean square error (RMSE) and the root mean square error-observation standard deviation ratio (RSR) are further calculated by the difference between the statistical value and the predicted result. The accuracy of the prediction result can be accurately evaluated by RMSE and RSR.

[0068] In some possible implementations, the coefficient of determination (R 2 ), root mean square error, and the ratio of the root mean square error observation standard deviation, which are usually used to evaluate the prediction performance of the geographically weighted random forest model. 2 It is used to measure the goodness of fit of the model. RMSE measures the error between the predicted value and the actual value. RSR is the ratio of RMSE standardized to the standard deviation of the observed value, which is used to further compare the size of the model error. As shown in formulas (5) to (7): (5); (6); (7); Where n is the number of data, is a statistical value, is the prediction result, and Represent the average values ​​of the prediction results and statistical values ​​respectively.

[0069] In a specific example, taking Province A as an example, the embodiment of the present invention optimizes the GWRF model through local weight adjustment and applies it to various control units in Province A. Province A has complex terrain, covering a variety of water systems and basins, and there are significant differences in the natural environment and human activities in the eastern and western regions. When training the local model, by setting different kernel bandwidth parameters, the embodiment of the present invention gives greater weight to the monitoring points in the same control unit in Province A, so that the impact of influencing factors related to human activities in the local area on total phosphorus pollution can be more accurately reflected. Such weight adjustment ensures that the GWRF model can better adapt to the complex basin characteristics of Province A and improves the accuracy of total phosphorus pollution assessment. Among them, the model evaluation results obtained by implementing a method for evaluating the impact of human activities on total phosphorus pollution in a basin provided by an embodiment of the present invention in Province A are as follows: RSR: 0.65065; R²: 0.5766577; RMSE: 0.02056. The evaluation results show that the model has high prediction accuracy and low error, and can better capture the key factors affecting the reference total phosphorus concentration, especially in the case of strong spatial heterogeneity.

[0070] The embodiment of the present invention visualizes the spatial distribution of the total phosphorus prediction results in each control unit (such as Figure 7 shown), from Figure 7 The visualization results of the target factors (PI, SI, AFU, OCO, SC, SS, SP, POI, Cropland, Impervious) shown clearly show that in the control unit in the eastern part of Province A, the cultivated land area has become the main target factor affecting the reference total phosphorus concentration. This phenomenon may be closely related to the dense cultivated land and frequent agricultural activities in the eastern region. In the control unit in the western part of Province A, oil crops have become the dominant influencing factor of the reference total phosphorus concentration, which is consistent with the widespread cultivation of oil crops and the large differences in land use patterns in the region. Further analysis of the importance of the impact of various human activity-related target factors on the reference total phosphorus concentration showed that the most important target factors in Province A are: cultivated land, agricultural fertilizer application, primary industry, oil production and poultry output. This conclusion provides strong support for relevant departments to formulate total phosphorus pollution prevention and control measures, especially in the eastern part of Province A, where the control of cultivated land and fertilizer use will become the key to effectively mitigating total phosphorus pollution. Among them, the ranking results of the importance of each target factor to total phosphorus pollution are as follows: Figure 8 As shown, from Figure 8 It can be seen that in Province A, the importance of each target factor to total phosphorus pollution is ranked from high to low as follows: Cropland, AFU, PI, OCO, SP, SS, SI, SC, IOR, POI, FO, POP, and Impervious.

[0071] The reference total phosphorus concentration distribution results predicted by the model show that the reference total phosphorus concentration in Province A presents a distribution characteristic of low concentration in the west and high concentration in the east. This phenomenon is highly similar to the distribution of cultivated land in Province A, indicating that there is a significant spatial correlation between cultivated land area and reference total phosphorus concentration. The cultivated land in the eastern region of Province A is concentrated, agricultural activities are frequent, and sources such as fertilization and agricultural wastewater may be the main reasons for the high reference total phosphorus concentration in this region. In the western region, due to the small cultivated land area and relatively weak agricultural activities, the reference total phosphorus concentration is generally low. Therefore, it can be reasonably inferred that within the scope of Province A, the cultivated land area and related agricultural activities are one of the main target factors affecting the distribution of reference total phosphorus concentration. This conclusion provides a scientific basis for water environment protection and pollution control in Province A, indicating that more stringent agricultural non-point source pollution control measures should be taken for areas with concentrated cultivated land in the east to effectively reduce the level of total phosphorus pollution.

[0072] In summary, the feasibility of this method can be fully demonstrated through the implementation and verification of the embodiment of the present invention in Province A. The model evaluation results show that indicators such as RSR, R² and RMSE indicate that the model has high prediction accuracy and low error, especially in the face of an environment with strong spatial heterogeneity, it can still accurately capture the target factors that affect the reference total phosphorus concentration. In addition, the spatial distribution results predicted by the model are highly consistent with the actual geographical characteristics and human activities, verifying the method's ability to effectively identify total phosphorus pollution sources in different regions. Therefore, the embodiment of the present invention provides reliable technical support for evaluating the impact of target factors related to human activities on total phosphorus pollution and formulating precise pollution prevention and control measures, and has a wide range of promotion and application value, especially in watersheds facing complex terrain and diverse human activities. It has strong applicability.

[0073] The embodiment of the present invention proposes a method for evaluating the impact of human activities on total phosphorus pollution in a watershed, aiming to solve the problems of multicollinearity among influencing factors related to human activities in the watershed, unscientific variable screening, insufficient processing of spatial heterogeneity, and so on.

[0074] First, by introducing the variance inflation factor method, the embodiment of the present invention effectively reduces the collinearity problem between the influencing factors related to human activities in the basin. By screening out candidate factors with less collinearity through the VIF method, the embodiment of the present invention ensures the independence of the input data, avoids overfitting, and improves the robustness and prediction accuracy of the geographically weighted random forest model.

[0075] Secondly, the embodiment of the present invention uses the grey correlation analysis method to further select the target factor with the strongest correlation with the reference total phosphorus concentration from the candidate factors preliminarily screened. The GRA method can systematically and quantitatively measure the correlation between the candidate factors and the reference total phosphorus concentration. This technical feature not only makes the geographically weighted random forest model have a high prediction accuracy, but also enhances the interpretability of the prediction results of the geographically weighted random forest model, which can provide a clearer pollution source analysis for watershed management and guide accurate pollution prevention and control measures. Then, the embodiment of the present invention proposes a local weight adjustment strategy to improve the geographically weighted random forest model. GWRF combines the advantages of spatial heterogeneity processing and nonlinear modeling, and can more accurately capture the differential effects of target factors related to human activities in different regions on total phosphorus pollution. However, when dealing with local areas, the GWRF model does not fully consider the strong interaction between monitoring points in the control unit, resulting in a lack of local prediction accuracy. Therefore, the embodiment of the present invention compensates for this deficiency by adjusting the kernel function. Since the control section monitoring points are a comprehensive reflection of the target factors and water quality related to human activities in the control unit, and there is often a strong interaction between the monitoring points in the control unit, ignoring this may lead to inaccurate predictions in local areas. Therefore, the embodiment of the present invention adjusts the kernel function, and by assigning greater weights to monitoring points within the same control unit, the processing capability of the geographically weighted random forest model for local areas is enhanced, thereby improving the prediction accuracy of local areas. This feature ensures that the geographically weighted random forest model has better performance when dealing with complex local pollution problems.

[0076] The embodiment of the present invention proposes a set of overall strategies to gradually focus on target factors related to key human activities. Combining VIF, GRA and the improved GWRF model with local weight adjustment, the embodiment of the present invention solves the problems of multicollinearity, unscientific variable screening, insufficient spatial heterogeneity treatment and other problems in related technologies, and realizes the refined prediction of total phosphorus pollution and the scientific evaluation of the main influencing factors. This method not only improves the stability and accuracy of the prediction, but also enhances the interpretability of the prediction results, provides more targeted and scientific decision-making support for watershed managers, and effectively improves the evaluation and governance efficiency of pollution in complex watersheds.

[0077] The embodiment of the present invention provides an evaluation device for the impact of human activities on total phosphorus pollution in a watershed. Fig. 9 , which shows a schematic diagram of the composition structure of a device for evaluating the impact of human activities on total phosphorus pollution in a watershed provided by an embodiment of the present invention. The device 900 includes: An acquisition module 901 is used to acquire influencing factors related to human activities in the watershed; A first screening module 902 is used to perform collinearity screening on the influencing factors to obtain candidate factors; A determination module 903 is used to determine the grey correlation degree between the candidate factor and the reference total phosphorus concentration based on the reference total phosphorus concentration and the candidate factor; A second screening module 904 is used to screen the candidate factors based on the grey relational degree to obtain a target factor; Prediction module 905 is used to use a geographically weighted random forest model to predict the impact of the target factor on the total phosphorus pollution in the watershed and obtain a prediction result; wherein, in the geographically weighted random forest model, the first weight of the monitoring point within the same control unit is greater than the second weight of the monitoring point outside the control unit.

[0078] In some possible implementations, the first screening module 902 is further used to determine the collinearity parameters between each of the influencing factors and other influencing factors; and screen the influencing factors based on the collinearity parameters to obtain the candidate factors.

[0079] In some possible implementations, the first screening module 902 is further used to determine the variance inflation factor between each influencing factor and other influencing factors; determine the variance inflation factor as the collinearity parameter; and eliminate the influencing factors whose variance inflation factor is greater than a preset inflation threshold from the influencing factors to obtain the candidate factors.

[0080] In some possible implementations, the determination module 903 is also used to determine the difference between the reference total phosphorus concentration and the candidate factor at any time to obtain a difference sequence; and determine the grey correlation degree between the reference total phosphorus concentration and the candidate factor based on the maximum difference and the minimum difference in the difference sequence.

[0081] In some possible implementations, the second screening module 904 is further configured to screen, from among the candidate factors, a target candidate factor whose grey correlation degree is greater than a preset influence degree threshold; and determine the target candidate factor as the target factor.

[0082] In some possible implementations, the prediction module 905 is also used to divide the watershed into multiple control units based on the hydrological characteristics, topography and water system distribution in the watershed; for each of the multiple control units, determine the first weight of the monitoring points in each control unit and the second weight of the monitoring points outside each control unit; wherein the first weight is greater than the second weight; based on the first weight and the corresponding monitoring point, and the second weight and the corresponding monitoring point, train the geographically weighted random forest model to be trained to obtain the geographically weighted random forest model.

[0083] In some possible implementations, the prediction module 905 is also used to determine, for any monitoring point within each control unit, the spatial distance between any monitoring point and other monitoring points and the bandwidth of the control unit to which the other monitoring points belong; wherein the other monitoring points are within each control unit or outside each control unit; based on the spatial distance and the bandwidth, determine the first weight between any monitoring point and other monitoring points within each control unit, and the second weight between any monitoring point and other monitoring points outside each control unit.

[0084] In some possible implementations, the acquisition module 901 is also used to obtain initial factors related to human activities in the watershed; perform data preprocessing on the initial factors to obtain processed factors; and perform area projection on the processed factors based on a preset resolution to obtain the influencing factors.

[0085] In some possible implementations, the prediction module 905 is also used to obtain a statistical value of the impact of the target factor on the total phosphorus pollution in the watershed; determine the difference between the statistical value and the prediction result; and evaluate the prediction result based on the difference to obtain an evaluation result.

[0086] Optionally, the transmission medium may be a wired link (such as but not limited to coaxial cable, optical fiber and digital subscriber line (DSL), etc.) or a wireless link (such as but not limited to wireless Fidelity (WIFI), Bluetooth and mobile device network, etc.). It should be noted that the device provided in the above embodiment is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above. In addition, the method embodiments provided in the above embodiments belong to the same concept. The specific implementation process is detailed in the method embodiments, which will not be repeated here.

[0087] Fig.10 is a schematic diagram of the structure of a computer device provided by an embodiment of the present invention. Fig.10 As shown, the computer device 1000 includes: a memory 1001, a processor 1002, and a computer program 1003 stored in the memory 1001 and running on the processor 1002, wherein when the processor 1002 executes the computer program 1003, the computer device can execute any of the aforementioned methods for assessing the impact of human activities on total phosphorus pollution in a watershed.

[0088] In addition, an embodiment of the present invention also protects a system, which may include a memory and a processor, wherein an executable program code is stored in the memory, and the processor is used to call and execute the executable program code to perform a method for evaluating the impact of human activities on total phosphorus pollution in a watershed provided by an embodiment of the present invention. This embodiment can divide the system into functional modules according to the above method example. For example, it can correspond to each functional module, or two or more functions can be integrated into one processing module. The above integrated module can be implemented in the form of hardware. It should be noted that the division of modules in this embodiment is schematic, which is only a logical function division, and there may be other division methods in actual implementation. It should be noted that all relevant contents of each step involved in the above method embodiment can be referred to the functional description of the corresponding functional module, which will not be repeated here.

[0089] It should be understood that the device provided in this embodiment is used to perform the above-mentioned method for evaluating the impact of human activities on total phosphorus pollution in a watershed, and thus can achieve the same effect as the above-mentioned implementation method. In the case of an integrated unit, the device may include a processing module and a storage module. Among them, when the device is applied to a device, the processing module can be used to control and manage the actions of the device. The storage module can be used to support the device to execute mutual program codes, etc. Among them, the processing module can be a processor or a controller, which can implement or execute various exemplary logic boxes, modules and circuits described in conjunction with the disclosure of the present invention. The processor can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of digital signal processing (DSP) and a microprocessor, etc., and the storage module can be a memory.

[0090] In addition, the device provided in the embodiment of the present invention may be a chip, a component or a module, and the chip may include a connected processor and a memory; wherein the memory is used to store instructions, and when the processor calls and executes the instructions, the chip may execute a method for evaluating the impact of human activities on total phosphorus pollution in a watershed provided in the above embodiment. This embodiment also provides a computer-readable storage medium, in which a computer program code is stored, and when the computer program code is run on a computer, the computer executes the above-mentioned related method steps to implement a method for evaluating the impact of human activities on total phosphorus pollution in a watershed provided in the above embodiment.

[0091] This embodiment also provides a computer program product, when the computer program product is run on a computer, the computer executes the above-mentioned related steps to implement a method for evaluating the impact of human activities on total phosphorus pollution in a watershed provided in the above embodiment. Among them, the device, computer-readable storage medium, computer program product or chip provided in this embodiment are all used to execute the corresponding method provided above, so the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method provided above, and will not be repeated here. Through the description of the above implementation mode, the technicians in the relevant field can understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In practical applications, the above-mentioned function allocation can be completed by different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In the embodiments provided by the present invention, it should be understood that the disclosed device and method can be implemented in other ways. For example, the device embodiment described above is only schematic, for example, the division of modules or units is only a logical function division, and there can be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, device or unit, which may be electrical, mechanical or other forms.

[0092] It should be noted that the sequence of the above-mentioned embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The process depicted in the accompanying drawings does not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or may be advantageous. The various embodiments in this specification are described in a progressive manner, and the same and similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments. The above content is only a specific implementation method of the present invention, but the protection scope of the present invention is not limited to this. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered within the protection scope of the present invention.

Claims

1. A method for assessing the impact of human activities on total phosphorus pollution in a watershed, characterized in that: The assessment methods for the impact of human activities on total phosphorus pollution in the watershed include: Obtain influencing factors related to human activities in the watershed; Performing collinearity screening on the influencing factors to obtain candidate factors; Based on the reference total phosphorus concentration and the candidate factor, determining the grey correlation degree between the candidate factor and the reference total phosphorus concentration; Screening the candidate factors based on the grey relational degree to obtain the target factor; A geographically weighted random forest model is used to predict the impact of the target factor on the total phosphorus pollution in the watershed to obtain a prediction result; wherein, in the geographically weighted random forest model, the first weight of the monitoring points within the same control unit is greater than the second weight of the monitoring points outside the control unit.

2. The method for evaluating the impact of human activities on total phosphorus pollution in a watershed according to claim 1, characterized in that: The influencing factors are subjected to collinearity screening to obtain candidate factors, including: Determine the collinearity parameters between each of the impact factors and other impact factors; The influencing factors are screened based on the collinearity parameters to obtain the candidate factors.

3. The method for evaluating the impact of human activities on total phosphorus pollution in a watershed according to claim 2, characterized in that: Determining the collinearity parameters between each of the influencing factors and other influencing factors includes: Determine the variance inflation factor between each of the impact factors and other impact factors; Determine the variance inflation factor as the collinearity parameter; Correspondingly, the influencing factors are screened based on the collinearity parameters to obtain the candidate factors, including: Among the influencing factors, the influencing factors whose variance inflation factors are greater than a preset inflation threshold are eliminated to obtain the candidate factors.

4. The method for evaluating the impact of human activities on total phosphorus pollution in a watershed according to claim 1, characterized in that: The step of determining the grey correlation between the candidate factor and the reference total phosphorus concentration based on the reference total phosphorus concentration and the candidate factor comprises: At any time, determining the difference between the reference total phosphorus concentration and the candidate factor to obtain a difference sequence; Based on the maximum difference and the minimum difference in the difference sequence, the grey correlation degree between the reference total phosphorus concentration and the candidate factor is determined.

5. The method for evaluating the impact of human activities on total phosphorus pollution in a watershed according to claim 1, characterized in that: The step of screening the candidate factors based on the grey relational degree to obtain the target factor includes: Among the candidate factors, target candidate factors whose grey correlation degree is greater than a preset influence degree threshold are selected; The target candidate factor is determined as the target factor.

6. The method for evaluating the impact of human activities on total phosphorus pollution in a watershed according to claim 1, characterized in that: The method further comprises: Based on the hydrological characteristics, topography and water system distribution in the watershed, the watershed is divided into a plurality of control units; For each of the plurality of control units, respectively determine a first weight of a monitoring point within each control unit and a second weight of a monitoring point outside each control unit; wherein the first weight is greater than the second weight; Based on the first weights and corresponding monitoring points, and the second weights and corresponding monitoring points, the geographically weighted random forest model to be trained is trained to obtain the geographically weighted random forest model.

7. The method for evaluating the impact of human activities on total phosphorus pollution in a watershed according to claim 6, characterized in that: The step of determining, for each of the plurality of control units, a first weight of a monitoring point within each control unit and a second weight of a monitoring point outside each control unit, respectively, comprises: For any monitoring point in each control unit, determine the spatial distance between any monitoring point and other monitoring points and the bandwidth of the control unit to which the other monitoring points belong; wherein the other monitoring points are in each control unit or outside each control unit; Based on the spatial distance and the bandwidth, a first weight between any monitoring point and other monitoring points in each control unit and a second weight between any monitoring point and other monitoring points outside each control unit are determined.

8. The method for evaluating the impact of human activities on total phosphorus pollution in a watershed according to claim 1, characterized in that: The influencing factors related to human activities in the watershed are obtained, including: Obtaining initial factors related to human activities in the watershed; Performing data preprocessing on the initial factors to obtain processed factors; Based on a preset resolution, area projection is performed on the processed factors to obtain the impact factors.

9. The method for evaluating the impact of human activities on total phosphorus pollution in a watershed according to claim 1, characterized in that: The method further comprises: using the geographically weighted random forest model to predict the impact of the target factor on the total phosphorus pollution in the watershed, and obtaining the prediction result. Obtaining the statistical value of the impact of the target factor on the total phosphorus pollution in the watershed; Determining a difference between the statistical value and the predicted result; Based on the difference, the prediction result is evaluated to obtain an evaluation result.

10. An evaluation device for the impact of human activities on total phosphorus pollution in a watershed, characterized in that: The device comprises: The acquisition module is used to obtain the influencing factors related to human activities in the watershed; A first screening module is used to perform collinearity screening on the influencing factors to obtain candidate factors; A determination module, used for determining the grey correlation degree between the candidate factor and the reference total phosphorus concentration based on the reference total phosphorus concentration and the candidate factor; A second screening module is used to screen the candidate factors based on the grey relational degree to obtain the target factor; A prediction module is used to use a geographically weighted random forest model to predict the impact of the target factor on the total phosphorus pollution in the watershed and obtain a prediction result; wherein, in the geographically weighted random forest model, the first weight of the monitoring points within the same control unit is greater than the second weight of the monitoring points outside the control unit.

Images