Heavy metal concentration prediction method combining collaborative Kriging and space-time geographic weighting
By combining synergistic Kriging and spatiotemporal geographic weighted regression models, the problem of spatiotemporal heterogeneity of traditional heavy metal pollution assessment methods on a large scale is solved, and high-precision soil heavy metal concentration prediction is achieved, which improves the accuracy and timeliness of pollution assessment.
Patent Information
- Application Number
- CN202510388713.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-31
- Publication Date
- 2025-08-15
AI Technical Summary
Traditional heavy metal pollution assessment methods are difficult to reflect the spatiotemporal heterogeneity of complex geographical environments within a large scale, and the existing machine learning models are insufficient in interpretability and regional applicability, resulting in insufficient accuracy and timeliness of pollution assessment.
Combined with synergistic Kriging and spatiotemporal geographic weighted regression models, we collect multi-source environmental data, use principal component analysis to simplify factors, build a dynamic regression model, and combine ArcGIS for Co-Kriging interpolation to accurately predict the soil heavy metal concentration.
The accuracy and timeliness of heavy metal pollution assessment are improved within a large scale, the problem of uneven time and space distribution of data is solved, and high-precision soil heavy metal concentration prediction is provided.
Smart Images
Figure CN120490435A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of environmental science and technology, and in particular relates to a method for predicting soil heavy metal concentration. Background Art
[0002] Heavy metal pollution has become a global soil environmental problem, particularly in areas with intensive agricultural and industrial activities. Its persistence, accumulation, and biotoxicity pose a serious threat to ecosystems and human health. Therefore, accurately predicting the large-scale distribution of heavy metal concentrations in soil is crucial for pollution prevention and control, land use management, and environmental remediation.
[0003] Traditional assessment methods rely primarily on field sampling and laboratory testing, comparing measured heavy metal concentrations with fixed geochemical background values. However, regional soils are influenced by multiple factors, including geological composition, land use, and anthropogenic activities, exhibiting significant spatiotemporal heterogeneity. Fixed background values therefore fail to fully reflect actual pollution levels. Furthermore, although a large amount of soil heavy metal data has been published in the literature, its uneven spatiotemporal distribution and data heterogeneity make direct integration and application challenging. These methods can, to a certain extent, reflect the spatial distribution characteristics of soil heavy metals, but they still have numerous limitations when dealing with complex geographic environments and large-scale data. In recent years, machine learning and geographically weighted spatiotemporal regression (GTWR) methods have demonstrated significant advantages in capturing the temporal and spatial variations in the relationship between soil heavy metal concentrations and environmental factors, but model interpretability and regional applicability still need to be improved.
[0004] To address the above issues, the present invention proposes a new method for predicting heavy metal concentrations that combines co-Kriging and spatiotemporal geographically weighted regression (GTWR). This method comprehensively utilizes co-Kriging interpolation to process the spatial correlation of multi-source environmental variables (such as soil type, climate factors, terrain characteristics, etc.), while combining the GTWR method to improve the adaptability of the prediction model to spatiotemporal changes. By constructing a dynamic regression model based on geographical weighting, this method can accurately predict soil heavy metal concentrations on a large scale, improving the accuracy and timeliness of regional pollution assessments. Summary of the Invention
[0005] In response to the evaluation limitations of the existing technology caused by traditional fixed background values and uneven temporal and spatial distribution of data, the present invention provides a heavy metal concentration prediction method that combines regional soil heterogeneity, so as to accurately predict soil heavy metal concentrations over a large scale and improve the accuracy and timeliness of regional pollution assessments.
[0006] The present invention provides a heavy metal concentration prediction method that combines co-kriging and spatiotemporal geographic weighting (GTWR). Co-kriging interpolation is used to process the spatial correlation of multi-source environmental variables (such as soil type, climate factors, and terrain characteristics). GTWR is also used to improve the adaptability of the prediction model to spatiotemporal changes. By constructing a dynamic regression model based on geographic weighting, soil heavy metal concentrations can be accurately predicted over a large scale, improving the accuracy and timeliness of regional pollution assessments. The specific steps include:
[0007] (1) Collect data on soil heavy metal concentrations from published literature, as well as data on factors closely related to changes in soil heavy metal concentrations, such as natural background, climatic conditions, agricultural activities, and industrial activities; the heavy metals include As, Cd, Cr, Cu, Hg, Ni, Pb, and Zn
[0008] (2) Using principal component analysis, we can effectively integrate and simplify the natural geographical and socioeconomic factors that affect soil heavy metal concentrations;
[0009] (3) Establish the relationship between soil heavy metal concentrations and different influencing factors based on spatiotemporal geographically weighted regression model;
[0010] (4) Construct a GTWR+Co-Kriging comprehensive evaluation model and evaluate the model's goodness of fit and prediction accuracy;
[0011] (5) Select the optimal solution and use the ArcGIS geostatistical analysis module to perform Co-Kriging interpolation to predict the concentrations of different heavy metals.
[0012] In step (1), the data collected from published literature on soil heavy metal concentrations, natural background, climate conditions, agricultural activities, industrial activities and other factors closely related to changes in soil heavy metal concentrations specifically include:
[0013] The published soil heavy metal concentration data include the following: (1) soil samples should be from the topsoil of 0-20 cm; (2) the study should include basic information such as heavy metal concentration, number of sampling points, sampling location, sampling year, etc.; (3) the processing and analysis of soil samples should be strictly carried out in accordance with the quality assurance and control system; (4) studies with special cases of heavy pollution such as landfills should be excluded; (5) samples should be analyzed and processed scientifically and rigorously using analytical methods (Table 1);
[0014] Table 1 Determination methods of heavy metal elements in soil
[0015]
[0016] Data on factors closely related to changes in soil heavy metal concentrations, such as the natural background, climatic conditions, agricultural activities, and industrial activities, include average annual precipitation, average annual temperature, normalized difference vegetation index, soil pH, soil organic matter content, soil total nitrogen content, soil total phosphorus content, soil total potassium content, nighttime light, land use type, altitude, slope, lithology, gross domestic product, value added of the primary industry, value added of the secondary industry, use of agricultural fertilizers, total sown area of crops, number of industrial enterprises above a designated size, and actual paved road area at the end of the year;
[0017] The heavy metals include As, Cd, Cr, Cu, Hg, Ni, Pb and Zn.
[0018] In step (2), the principal component analysis method is used to effectively integrate and simplify the natural geographical and socioeconomic factors that affect the concentration of heavy metals in soil, specifically including:
[0019] Principal component analysis (PCA) is used to reduce the number of variables in physical geographic and socioeconomic datasets, extracting a few important principal components that represent the majority of the variable information. The principal component scores obtained after PCA transformation are multiplied by their spatial weights to generate a series of new predictor variables.
[0020] In step (3), the relationship between soil heavy metal concentration and different influencing factors is established based on the spatiotemporal geographic weighted regression model, specifically including:
[0021] The natural geographical and socioeconomic factors extracted by the PCA method are respectively incorporated into the explanatory variables of the GTWR model. The relevant formula of the GTWR model is as follows:
[0022]
[0023] Among them, u i and v i are the longitude and latitude of the i-th site, t i is the time corresponding to the site, y i is the dependent variable value corresponding to the i-th site, X ik is the kth explanatory variable value corresponding to the i-th site, ε i is the model error term, β0(u i ,v i ,t i ) is the regression intercept corresponding to the i-th site, β k (u i ,v i ,t i ) is the regression coefficient corresponding to the kth explanatory variable at the i-th site. The regression coefficient is calculated as follows:
[0024]
[0025] Among them, W(u i ,v i ,t i ) is the spatiotemporal weight matrix; CR k,i is the contribution of the kth explanatory variable to the i-th site (%), CR k is the overall contribution (%) of the k-th explanatory variable.
[0026] In step (4), the GTWR+Co-Kriging comprehensive evaluation model is constructed and the goodness of fit and prediction accuracy of the model are evaluated, specifically including:
[0027] Based on the results of the GTWR model, the three covariates with the largest contribution are selected as input information and included in the Co-Kriging model. The calculation formula for the predicted value of the main variable to be estimated is as follows:
[0028]
[0029] Among them, Z(x) is the result of the main variable being affected by the auxiliary variable, Z(x i ) is the main variable, T(x j ) is an auxiliary variable, λ i and λ j are the weight values associated with Z and T respectively.
[0030] Parameters such as R-square, standardized mean, and standardized root mean square were used to evaluate the goodness of fit and prediction accuracy of the GTWR model;
[0031] In step (5), the optimal solution is selected and Co-Kriging interpolation is performed using the ArcGIS geostatistical analysis module to predict the concentrations of different heavy metals, specifically including:
[0032] Three scenarios were selected for concentration prediction, with one (CK1), two (CK2), and three (CK3) predictors contributing the most to the impact of the eight heavy metals. The optimal scenario was selected and Co-Kriging interpolation was performed using the ArcGIS Geostatistical Analysis module to predict the concentrations of different heavy metals.
[0033] The present invention also provides a computer-readable storage medium, which specifically includes a method program for predicting heavy metal concentrations by combining collaborative kriging and spatiotemporal geographic weighting (GTWR). When the method program for predicting heavy metal concentrations by combining collaborative kriging and spatiotemporal geographic weighting (GTWR) is executed by a processor, the steps of the method for predicting heavy metal concentrations by combining collaborative kriging and spatiotemporal geographic weighting (GTWR) as described in any one of the above items are implemented.
[0034] The present invention provides a method for predicting heavy metal concentrations by combining co-kriging and spatiotemporal geographic weighting (GTWR), including: collecting data on soil heavy metal concentrations from published literature, natural background, climatic conditions, agricultural activities, industrial activities, and other factors closely related to changes in soil heavy metal concentrations; using the principal component analysis method to effectively integrate and simplify the natural geographical and socioeconomic factors affecting soil heavy metal concentrations; establishing the relationship between soil heavy metal concentrations and different influencing factors based on a spatiotemporal geographic weighted regression model; constructing a GTWR+Co-Kriging comprehensive evaluation model, and evaluating the model's goodness of fit and prediction accuracy; selecting the best solution, using the ArcGIS geostatistical analysis module to perform Co-Kriging interpolation, and predicting different heavy metal concentrations. Compared with the prior art, the present invention overcomes the limitations of the prior art that rely on field sampling and laboratory analysis, as well as the uneven spatiotemporal distribution of public data obtained through literature retrieval, and provides a method for predicting heavy metal concentrations by combining co-kriging and spatiotemporal geographic weighting (GTWR), which can accurately predict soil heavy metal concentrations over a large scale and improve the accuracy and timeliness of regional pollution assessment. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] Figure 1 This is a flow chart of the heavy metal concentration prediction method combining co-kriging and spatiotemporal geographic weighting in the present invention.
[0036] Figure 2 Schematic diagram of the sampling locations and sampling quantities of soil heavy metal concentration data published in Chinese literature.
[0037] Figure 3 Schematic diagram of the contribution of different influencing factors to the concentration of heavy metals in Chinese soils.
[0038] Figure 4 Schematic diagram of the comparison results of prediction performance of different models.
[0039] Figure 5 This is a schematic diagram of the prediction results of the spatial distribution of heavy metal concentrations in China's soils based on the present invention. DETAILED DESCRIPTION
[0040] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, in the absence of conflict, the embodiments of the present application and the features therein can be combined with each other.
[0041] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.
[0042] Figure 1 A flow chart of a method for predicting heavy metal concentrations by combining co-kriging and spatiotemporal geographic weighting (GTWR) according to one embodiment of the present invention;
[0043] S102: Collect data on soil heavy metal concentrations from published literature, as well as data on factors closely related to changes in soil heavy metal concentrations, such as natural background, climatic conditions, agricultural activities, and industrial activities;
[0044] S104, using principal component analysis, effectively synthesized and simplified the natural geographical and socioeconomic factors that affect soil heavy metal concentrations;
[0045] S106, establish the relationship between soil heavy metal concentrations and different influencing factors based on spatiotemporal geographically weighted regression model;
[0046] S108, construct the GTWR+Co-Kriging comprehensive evaluation model and evaluate the model's goodness of fit and prediction accuracy.
[0047] S110, select the optimal solution, use the ArcGIS geostatistical analysis module to perform Co-Kriging interpolation, and predict different heavy metal concentrations.
[0048] It should be noted that the present invention provides a method for predicting heavy metal concentrations by combining co-Kriging and spatiotemporal geographic weighting (GTWR). By collecting data on soil heavy metal concentrations from published literature, natural background, climatic conditions, agricultural activities, industrial activities and other factors closely related to changes in soil heavy metal concentrations; using the principal component analysis method, the natural geographical and socioeconomic factors affecting soil heavy metal concentrations are effectively integrated and simplified; based on the spatiotemporal geographic weighted regression model, the relationship between soil heavy metal concentrations and different influencing factors is established; a GTWR+Co-Kriging comprehensive evaluation model is constructed, and the model's goodness of fit and prediction accuracy are evaluated; the optimal solution is selected, and the ArcGIS geostatistical analysis module is used for Co-Kriging interpolation to predict different heavy metal concentrations. This solves the problem that high-density sampling is costly and time-consuming, and the data obtained only represent the conditions at a specific point in time, making it difficult to reflect the long-term trend of heavy metal concentrations. It also takes into account the spatiotemporal attributes of the non-uniform spatiotemporal distribution of the collected published data, significantly improving the prediction accuracy of soil heavy metal concentrations.
[0049] Furthermore, in a preferred embodiment of the present invention, the collection of data on soil heavy metal concentrations from published literature, natural background, climate conditions, agricultural activities, industrial activities, and other factors closely related to changes in soil heavy metal concentrations specifically includes:
[0050] The published soil heavy metal concentration data include the following: 1) soil samples should be from the topsoil layer (0-20 cm); 2) the study should include basic information such as heavy metal concentration, number of sampling points, sampling location, and sampling year; 3) soil sample processing and analysis must be strictly carried out in accordance with the quality assurance and control system; 4) studies involving special cases of heavy pollution, such as landfills, should be excluded; 5) samples should be processed and analyzed using scientific and rigorous methods (Table 1);
[0051] Table 1 Determination methods of heavy metal elements in soil
[0052]
[0053] The data on factors closely related to changes in soil heavy metal concentrations, such as natural background, climatic conditions, agricultural activities, and industrial activities, include average annual precipitation, average annual temperature, normalized difference vegetation index, soil pH, soil organic matter content, soil total nitrogen content, soil total phosphorus content, soil total potassium content, nighttime light, land use type, altitude, slope, lithology, gross domestic product, value added of the primary industry, value added of the secondary industry, use of agricultural fertilizers, total sown area of crops, number of industrial enterprises above a designated size, and actual paved road area at the end of the year;
[0054] The heavy metals include As, Cd, Cr, Cu, Hg, Ni, Pb and Zn.
[0055] It should be noted that the soil heavy metal concentration data collected from published literature are sourced from databases such as Web of Science (https: / / www.webofscience.com / ), Science Direct (https: / / www.sciencedirect.com), and China National Knowledge Infrastructure (CNKI) (http: / / www.cnki.net). Physical geographic data are primarily sourced from the National Earth System Science Data Center (https: / / www.geodata.cn) and the Resources and Environmental Science and Data Center of the Chinese Academy of Sciences (https: / / www.resdc.cn). Socioeconomic data are primarily sourced from publicly available information from the National Bureau of Statistics, including the National Bureau of Statistics database (https: / / data.stats.gov.cn) and statistical yearbooks at all levels (2000-2022): China Statistical Yearbook, China City Statistical Yearbook, China Urban Construction Statistical Yearbook, China Rural Statistical Yearbook, China County Statistical Yearbook, Provincial Statistical Yearbooks, and Municipal Statistical Yearbooks.
[0056] Furthermore, in a preferred embodiment of the present invention, the principal component analysis method effectively integrates and simplifies the natural geographical and socioeconomic factors that affect the concentration of heavy metals in soil, specifically including:
[0057] Principal component analysis (PCA) is used to reduce the number of variables in physical geographic and socioeconomic datasets, extracting a few important principal components that represent the majority of the variable information. The principal component scores obtained after PCA transformation are multiplied by their spatial weights to generate a series of new predictor variables.
[0058] It should be noted that in principal component analysis, it is necessary to ensure that most of the information in the variables is captured. Therefore, it is recommended that the variance explained and cumulative contribution should be greater than 75%. This not only reduces the complexity of the model, but also avoids multicollinearity problems and improves the stability and explanatory power of the prediction model.
[0059] Furthermore, in a preferred embodiment of the present invention, the relationship between soil heavy metal concentration and different influencing factors is established based on the spatiotemporal geographical weighted regression model, specifically including:
[0060] The natural geographical and socioeconomic factors extracted by the PCA method are respectively incorporated into the explanatory variables of the GTWR model. The relevant formula of the GTWR model is as follows:
[0061]
[0062] u i and v i are the longitude and latitude of the i-th site, t i is the time corresponding to the site, y i is the dependent variable value corresponding to the i-th site, X ik is the kth explanatory variable value corresponding to the i-th site, ε i is the model error term, β0(u i ,v i ,t i ) is the regression intercept corresponding to the i-th site, β k (u i ,v i ,t i ) is the regression coefficient corresponding to the kth explanatory variable at the i-th site. The regression coefficient is calculated as follows:
[0063]
[0064] W(u i ,v i ,t i ) is the spatiotemporal weight matrix. CR k,iis the contribution of the kth explanatory variable to the i-th site (%), CR k is the overall contribution (%) of the k-th explanatory variable.
[0065] It should be noted that the GTWR model is implemented by applying the "Geographically and Temporally Weighted Regression" external plug-in of ArcGIS 10.5.
[0066] Furthermore, in a preferred embodiment of the present invention, the construction of the GTWR+Co-Kriging comprehensive evaluation model and the evaluation of the model's goodness of fit and prediction accuracy specifically include:
[0067] Based on the results of the GTWR model, the three covariates with the largest contribution are selected as input information and included in the Co-Kriging model. The calculation formula for the predicted value of the main variable to be estimated is as follows:
[0068]
[0069] Among them, Z(x) is the result of the main variable being affected by the auxiliary variable, Z(x i ) is the main variable, T(x j ) is an auxiliary variable, λ i and λ j are the weight values associated with Z and T respectively.
[0070] Parameters such as R-square, standardized mean, and standardized root mean square were used to evaluate the goodness of fit and prediction accuracy of the GTWR model;
[0071] It should be noted that co-kriging can select the three variables with the largest contribution rate, so three comprehensive evaluation models can be constructed. Therefore, the optimal model is finally selected by comprehensively considering parameters such as R square, standard mean and standard root mean square.
[0072] Furthermore, in a preferred embodiment of the present invention, the optimal solution is selected, and Co-Kriging interpolation is performed using the ArcGIS geostatistical analysis module to predict the concentrations of different heavy metals, specifically including:
[0073] Three scenarios were selected, one (CK1), two (CK2), and three (CK3) predictors that contributed the most to the eight heavy metals. The optimal scenario was selected and Co-Kriging interpolation was performed using the ArcGIS Geostatistical Analysis module to predict the concentrations of different heavy metals.
[0074] It should be noted that the quality of the model prediction results is judged based on the relevant parameters of the model. When the standard mean value is close to 0 and the standard root mean square prediction error is close to 1, the model fit is good. CK1: Select one of the most contributing factors to participate in the prediction of co-kriging; CK2: Select two of the most contributing factors to participate in the prediction of co-kriging; CK3: Select three of the most contributing factors to participate in the prediction of co-kriging
[0075] In addition, in another embodiment provided by the present invention, taking heavy metals in Chinese soil as an example, a method for predicting heavy metal concentrations by combining co-kriging and spatiotemporal geographic weighting (GTWR) provided by the present invention is described, and the specific analysis steps are as follows:
[0076] The study collected the data on soil heavy metal concentrations in China from published literature. The study finally included more than 1,700 studies in 31 of the 34 provinces in my country, with a total of 280,000 sampling points. Due to the lack of relevant research, the concentrations of heavy metals in soil, including As, Cd, Cr, Cu, Hg, Ni, Pb and Zn, were analyzed. Figure 2 The schematic diagram of the sampling locations and sampling quantities of soil heavy metal concentration data published in China is shown;
[0077] Furthermore, data on factors closely related to changes in soil heavy metal concentrations, such as natural background, climatic conditions, agricultural activities, and industrial activities, were collected, including annual average precipitation, annual average temperature, normalized difference vegetation index, soil pH, soil organic matter content, soil total nitrogen content, soil total phosphorus content, soil total potassium content, nighttime light, land use type, altitude, slope, lithology, gross domestic product, added value of the primary industry, added value of the secondary industry, use of agricultural fertilizers, total sown area of crops, number of industrial enterprises above designated size, and actual paved road area at the end of the year, as shown in Table 2 below:
[0078] Table 2 Physical geography and socioeconomic data
[0079]
[0080]
[0081] Furthermore, principal component analysis (PCA) was used to effectively integrate and simplify the natural geographic and socioeconomic factors influencing soil heavy metal concentrations. The first eight principal components (PC1-PC8) cumulatively explained 80.17% of the total variance, significantly reflecting the main information and structure in the dataset. Specifically, PC1 is closely related to economic development and urbanization, representing the impact of human economic activities on the soil environment. PC2 is significantly correlated with soil pH and vegetation status (as reflected by the NDVI indicator), revealing the interaction between soil chemical properties and ecological quality. PC3 focuses on the intensity of agricultural activities, especially the use of pesticides and fertilizers. PC4 reflects the nutrient status of soil, such as organic matter and nitrogen content. PC5 is related to the intensity of human activities and their impact on land use patterns. PC6 reveals the potential influence of topography on soil heavy metal concentrations. PC7 and PC8 indicate the phosphorus and potassium content and soil parent material properties, respectively, which may have a significant impact on the behavior and distribution of heavy metals in soil.
[0082] Furthermore, the relationship between soil heavy metal concentration and different influencing factors was established based on the spatiotemporal geographic weighted regression model, and the contribution of different influencing factors to heavy metal concentration was obtained. Figure 3 As shown, soil Cu and As concentrations are primarily affected by economic development and urbanization levels. Furthermore, agricultural activities significantly contribute to the accumulation of Pb and Zn in soil. Human activity intensity and land use patterns are significantly correlated with soil Hg and Ni concentrations. Topography significantly influences soil Cd concentrations.
[0083] Furthermore, a GTWR+Co-Kriging comprehensive evaluation model is constructed, such as Figure 4 As shown, for the heavy metals Cd, Cu, Pb, and Zn, the CK2 scheme produced the best concentration prediction results, while the CK3 scheme produced the best concentration prediction results for As, Cr, Hg, and Ni. By selecting two or three key factors as model inputs, the main environmental gradients affecting heavy metal concentrations can be effectively captured, while ensuring model simplicity and maximizing the interrelationships between these factors and their combined impact on heavy metal concentrations.
[0084] Finally, based on the optimal model solution, the ArcGIS geostatistical analysis module was used to perform Co-Kriging interpolation to predict the concentrations of different heavy metals.
[0085] The predicted results of different heavy metal concentrations are as follows Figure 5As shown, Cd concentrations ranged from 0.07 to 10.29 mg / kg, with high concentrations concentrated in southwestern provinces such as Sichuan, Yunnan, Guizhou, and Guangxi. Cr concentrations ranged from 29.98 to 136.01 mg / kg, showing a relatively even distribution. High concentrations were primarily found in southwestern and northwestern my country. Cu concentrations peaked at 211.37 mg / kg, occurring in the Inner Mongolia Autonomous Region. Hg concentrations ranged from 0.04 to 12.13 mg / kg, with high concentrations concentrated in provinces such as the Xinjiang Uyghur Autonomous Region, the Inner Mongolia Autonomous Region, and the Guangxi Zhuang Autonomous Region. Ni concentrations were also relatively evenly distributed, with the highest concentration (142.67 mg / kg) occurring in Anhui Province. Pb and Zn concentrations ranged from 15.47 to 294.27 mg / kg and 6.74 to 404.64 mg / kg, respectively, showing a wide range and uneven distribution. High concentrations were primarily concentrated in provinces such as Sichuan, Hunan, and the Guangxi Zhuang Autonomous Region.
[0086] Based on the above analysis, we can know that the method for heavy metal concentration prediction that combines co-kriging and spatiotemporal geographic weighting (GTWR) can accurately predict the concentration of heavy metals in soil, introducing a new perspective for better utilizing public data for large-scale pollutant research and providing reliable technical support for soil pollution prevention and control and environmental management.
[0087] The second aspect of the present invention provides a computer-readable storage medium, characterized in that the computer-readable storage medium includes a method program for predicting heavy metal concentrations by combining co-kriging and spatiotemporal geographic weighting (GTWR). When the method program for predicting heavy metal concentrations by combining co-kriging and spatiotemporal geographic weighting (GTWR) is executed by a processor, the steps of the method for predicting heavy metal concentrations by combining co-kriging and spatiotemporal geographic weighting (GTWR) as described above are implemented.
[0088] In the several embodiments provided herein, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the effective integration and simplification of natural, socioeconomic factors is merely one method for index screening and utilization. In actual implementation, other methods may be employed, such as relationship analysis or machine learning Borute, or some features may be ignored or not executed.
[0089] Those skilled in the art will appreciate that all or part of the steps of the above-mentioned method embodiments may be implemented by hardware associated with program instructions, and the aforementioned program may be stored in a computer-readable storage medium. When the program is executed, the program executes the steps of the above-mentioned method embodiments. The aforementioned storage medium includes various media that can store program codes, such as mobile storage devices, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.
[0090] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.
[0091] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.
Claims
1. A heavy metal concentration prediction method that combines co-kriging and spatiotemporal geographic weighting. It comprehensively utilizes co-kriging interpolation to process the spatial correlation of multi-source environmental variables, including soil type, climate factors, and terrain characteristics, and combines the GTWR method to improve the adaptability of the prediction model to spatiotemporal changes. By constructing a dynamic regression model based on geographic weighting, it accurately predicts soil heavy metal concentrations over a large scale, improving the accuracy and timeliness of regional pollution assessment. The characteristics are: The specific steps include: (1) Collect published data on soil heavy metal concentrations, natural background, climate conditions, agricultural activities, industrial activities, and factors closely related to changes in soil heavy metal concentrations; the heavy metals included As, Cd, Cr, Cu, Hg, Ni, Pb, and Zn; (2) Using principal component analysis, we can effectively integrate and simplify the natural geographical and socioeconomic factors that affect soil heavy metal concentrations; (3) Establish the relationship between soil heavy metal concentrations and different influencing factors based on spatiotemporal geographically weighted regression model; (4) Construct a GTWR+Co-Kriging comprehensive evaluation model and evaluate the model's goodness of fit and prediction accuracy; (5) Select the optimal solution and use the ArcGIS geostatistical analysis module to perform Co-Kriging interpolation to predict the concentrations of different heavy metals.
2. The method for predicting heavy metal concentration according to claim 1, wherein: In step (1): The published soil heavy metal concentration data require the following: (1) soil samples should be from the topsoil layer of 0-20 cm; (2) the data should include basic information such as heavy metal concentration, number of sampling points, sampling location, and sampling year; (3) the processing and analysis of soil samples should be strictly carried out in accordance with the quality assurance and control system; (4) the study area should be excluded from special scenes with heavy pollution such as landfills; (5) the samples should be scientifically and rigorously analyzed and processed using the analytical methods shown in Table 1; Table 1 Determination of heavy metal elements in soil The data on factors closely related to changes in soil heavy metal concentrations, including natural background, climatic conditions, agricultural activities, industrial activities, include average annual precipitation, average annual temperature, normalized difference vegetation index, soil pH, soil organic matter content, soil total nitrogen content, soil total phosphorus content, soil total potassium content, night light, land use type, altitude, slope, lithology, gross domestic product, value added of the primary industry, value added of the secondary industry, use of agricultural chemical fertilizers, total sown area of crops, number of industrial enterprises above designated size, and actual paved road area at the end of the year.
3. The method for predicting heavy metal concentration according to claim 2, wherein: In step (2), the principal component analysis method is used to reduce the number of variables in the physical geographic and socioeconomic data sets, and several important principal components are extracted to represent most of the variable information; the principal component scores obtained after PCA transformation are multiplied by their weights in spatial distribution to generate a series of new predictor variables.
4. The method for predicting heavy metal concentration according to claim 3, wherein: Step (3) specifically includes: The natural geographical and socioeconomic factors extracted by the PCA method are respectively incorporated into the explanatory variables of the GTWR model. The relevant formula of the GTWR model is as follows: Among them, u i and v i are the longitude and latitude of the i-th site, t i is the time corresponding to the site, y i is the dependent variable value corresponding to the i-th site, X ik is the kth explanatory variable value corresponding to the i-th site, ε i is the model error term, β0(u i ,v i ,t i ) is the regression intercept corresponding to the i-th site, β k (u i ,v i ,t i ) is the regression coefficient corresponding to the kth explanatory variable of the i-th site; the calculation formula of the regression coefficient is as follows: Among them, W(u i ,v i ,t i ) is the spatiotemporal weight matrix; CR k,i is the contribution of the kth explanatory variable to the i-th site (%), CR k is the overall contribution (%) of the k-th explanatory variable.
5. The method for predicting heavy metal concentration according to claim 4, wherein: Step (4) specifically includes: Based on the results of the GTWR model, the three covariates with the largest contribution are selected as input information and included in the Co-Kriging model. The formula for calculating the predicted value of the main variable to be estimated is as follows: Among them, Z(x) is the result of the main variable being affected by the auxiliary variable, Z(x i ) is the main variable, T(x j ) is an auxiliary variable, λ i and λ j are the weight values associated with Z and T respectively; R-square, standardized mean and standardized root mean square parameters were used to evaluate the goodness of fit and prediction accuracy of the GTWR model.
6. The method for predicting heavy metal concentration according to claim 5, characterized in that: In step (5), one (CK1), two (CK2), and three (CK3) predictor variables that contribute the most to the impact of the eight heavy metals are selected as three schemes for concentration prediction; the optimal scheme is selected, and the ArcGIS geostatistical analysis module is used to perform Co-Kriging interpolation to predict the concentrations of different heavy metals.
7. A computer-readable storage medium, characterized in that A program comprising a method for predicting heavy metal concentrations that combines collaborative kriging and spatiotemporal geographic weighting. When the program for predicting heavy metal concentrations that combines collaborative kriging and spatiotemporal geographic weighting is executed by a processor, the steps of the method for predicting heavy metal concentrations that combines collaborative kriging and spatiotemporal geographic weighting as described in any one of claims 1 to 6 are implemented.