CN112986538A

CN112986538A - Large-area soil heavy metal detection and space-time distribution characteristic analysis method and system

Info

Publication number: CN112986538A
Application number: CN202110487626.XA
Authority: CN
Inventors: 邱罗; 毛先成; 刘启波; 刘占坤
Original assignee: Central South University
Current assignee: Central South University
Priority date: 2021-05-06
Filing date: 2021-05-06
Publication date: 2021-06-18
Also published as: CN114443982B; CN114443982A

Abstract

The invention discloses a method and a system for detecting heavy metals in soil in a large area and analyzing space-time distribution characteristics, wherein the method comprises the following steps: 1) constructing an analysis model for spatial distribution of a plurality of heavy metal elements in a large area; 2) determining an analysis object; 3) sampling and detecting urban surface soil in each gridding area; 4) and substituting the obtained sampling composite detection data into an analysis model, analyzing the heavy metal content and the geographic information of the soil, and drawing a total content distribution diagram of the heavy metal elements. The system provided by the invention comprises a remote server, a plurality of front-end machines and terminal equipment which are connected and communicated with each other through a network. According to the invention, the distribution characteristics of four elements of As, Cd, Hg and Pb in a large area are analyzed by constructing a new mathematical model, screening a typical detection object and comprehensively utilizing sampling point data and geochemical survey data, the area coverage of an analysis result is complete, the accuracy is high, and the automation degree of the analysis process is high.

Description

Large-area soil heavy metal detection and space-time distribution characteristic analysis method and system

Technical Field

The invention relates to the technical field of soil heavy metal detection and geographic information analysis, in particular to a method and a system for detecting heavy metal in soil in a large area and analyzing space-time distribution characteristics.

Background

With the rapid development of urbanization in China, more and more urban groups closely related to regions, economy and culture appear successively. The urban group is the highest spatial organization form of the city developing to the mature stage, and refers to an urban group which generally takes more than 1 super-large city as a core, takes more than 3 large cities as a constituent unit, is formed by relying on developed infrastructure networks such as traffic communication and the like, has compact spatial organization and close economic connection, and finally realizes high city sharing and high integration. The urban group is a huge, multi-core and multi-level urban group formed by gathering a plurality of super cities and big cities which are distributed in a centralized manner on regions, and is a union of metropolitan areas. Including national level Jingjin Ji City group, Chang triangular City group, Guangdong hong ao City group, Yu forming City group, City group in each province, Guangdong Yanhui, Hunan Changan Tan, etc. According to the regional development theory, investigation and research are carried out on the aspects of geographic positions, resource conditions, industrial distribution, policy systems, development modes and the like of all urban groups, so that a basis can be provided for the urban group and regional integrated development policy system and the like. However, due to various technical difficulties, no report has been found for soil heavy metal pollution detection and spatial and temporal distribution characteristic analysis for large areas such as urban communities.

The soil is the most valuable natural resource for human survival, has the functions of adsorbing, buffering and purifying environmental pollutants such as heavy metals, but the soil heavy metal pollution can cause direct or indirect harm to the normal survival and development of human beings and organisms, and the research on the space and time distribution of the heavy metals in the soil and the trend analysis and prediction are important foundations for formulating large-area ecological environment protection policies and technical schemes such as urban groups. The media reports that heavy metal pollution and health damage incidents occur in some domestic cities since the 20 th century, but in the past, people have conducted heavy metal pollution detection and heavy metal spatial distribution investigation on specific objects such as farmlands only in small areas such as villages and small towns, but have not conducted investigation and research on large areas across cities such as cities, and particularly have not conducted investigation and research on the time dimension of large areas. In the prior art, in the process of sampling, detecting and analyzing soil heavy metals, due to the small sampling point area, limited sample quantity and non-uniform data format, the method can not be directly applied to the research of large areas such as urban groups and the like; the existing research methods often cannot completely cover the space distribution and time change conditions of the whole large area, so that the representativeness, the accuracy and the trend of an analysis result are insufficient, and further analysis evaluation on the soil environment quality of the large area of an urban group and a targeted prevention and treatment scheme cannot be made.

Meanwhile, most of the current analysis, evaluation and result output of the soil environment quality are manually finished by field experts, and the professional technologies mastered by the field experts are utilized to analyze and judge in a manual mode, so that the defects of low data processing capacity, low analysis efficiency, certain subjectivity of analysis results and the like exist, new technologies such as machine learning and the like cannot be applied to related fields, the automation degree of the analysis process is low, and the process repeatability and the method verifiability are poor.

Disclosure of Invention

Aiming at the defects in the prior art, the invention aims to provide a large-area soil heavy metal detection and space-time distribution characteristic analysis method taking an urban group As a research object, construct a new method and a new system, analyze the time and space characteristics of distribution of four elements of As, Cd, Hg and Pb in a large area by constructing a new mathematical model, screening a typical detection object and comprehensively utilizing sampling point data, multi-target geochemical investigation and remote sensing data, ensure that the area coverage of an analysis result is complete and the accuracy is high, and further analyze and evaluate the soil environment quality of the urban group.

The invention also aims to provide a computer analysis system for implementing the method, which realizes the automation of collection, input, operation and result output of soil heavy metal detection and space-time distribution characteristic analysis data of a large area through the cooperation of network communication, computer software and hardware and a detection instrument, improves the data analysis processing capacity, the processing efficiency and the accuracy and objectivity of analysis results, and avoids the interference of human factors on the data and result output.

In order to achieve the purpose, the technical scheme adopted by the invention is as follows:

a large-area soil heavy metal detection and space-time distribution characteristic analysis method comprises the following steps:

1) constructing an analysis model for spatial distribution of various heavy metal elements in large areas such As urban groups, analyzing the contents of Hg and Pb in soil by adopting a multivariate stepwise regression model in geospatial regression, and analyzing the contents of As and Cd in soil by adopting a least square regression model;

11) selecting 80% of large-area sampling composite detection data for establishing a geographic spatial distribution analysis regression model, and simulating the spatial distribution of heavy metals by using the correlation between soil variables and other influence factors;

in the step-by-Step Multiple Linear Regression (SMLR) model in step 1), the step-by-step regression is a partial regression equation composed of factors which are selected from n independent variables of the multiple linear regression and play an important role in the dependent variable y. Stepwise regression requires that x be checked one by one at each step of the calculation, ensuring that the final regression equation contains all and only those independent variables x that have significant effect on the dependent variable y; the stepwise regression process comprises two basic steps, namely removing the non-significant variables from the regression model, and introducing new variables into the regression model and checking the variables one by one; the model formula is (3-1):

the model formula of Partial Least Squares Regression (PLSR) is (3-2):

in the formula, independent variables are factors influencing heavy metals in soil: soil pH, slope, grade, elevation, NDVI, defined as variables respectivelyx ₁～x ₅The two dumb variables of land utilization mode agricultural land and non-utilized land are respectively defined asx ₆、x ₇Dependent variable ofy ₁～y ₄Respectively represents the contents of As, Cd, Hg and Pb.

12) Selecting 20% of large-area sampling composite detection data for model precision detection, and detecting the accuracy and precision of a model prediction result by adopting cross validation and trend analysis, wherein the calculation formulas of an absolute average error MAE, an average relative error MRE and a root mean square error RMSE are as follows:

whereinnAs to the number of samples,M _kis as followskThe actual measured value of each sample point is measured,P _kis a predicted value.

2) According to the analysis model, determining typical analysis objects of the heavy metals in the soil in the large area As, Cd, Hg and Pb elements in the soil;

3) gridding a large urban group area, sampling and detecting urban surface soil in each gridding area according to a set proportion and a set place, and obtaining sampling composite detection data of As, Cd, Hg and Pb element contents, sampling time and GIS grid correlation in the urban group soil; the method specifically comprises the following steps:

31) carrying out GIS meshing on a large area of an urban group, collecting soil samples in each grid, and recording the GIS grid information of the sample collection time and the collection place;

32) taking the collected sample back to a laboratory, naturally drying the sample, removing plant residues and broken stones, and grinding the sample to 100 meshes by using an agate pot body;

33) pretreating a soil sample: carrying out digestion by using a nitric acid-perchloric acid-hydrofluoric acid mixed solution;

34) detecting the content of Cd and Pb elements by adopting a plasma emission spectrometer (ICP-OES), and detecting the content of As and Hg elements by adopting an atomic fluorescence Analyzer (AFS);

35) respectively associating the content of the metal elements detected by each sample with GIS grid information of the acquisition time and the acquisition place of the sample to obtain GIS-based sampling composite detection data of the metal elements in each soil sample;

36) acquiring multi-target geochemical survey and remote sensing data from the outside, conducting standardized processing on the data, importing the data into a multi-target survey database module of a remote server, and calling the data together with composite detection data obtained by sampling points; and simultaneously, according to the GIS grids of the multi-target geochemical survey and the remote sensing data, planning the GIS grids of the sampling points again, so that the sampled composite detection data obtained repeatedly before and after the sampling has comparability.

4) Substituting the obtained sampling composite detection data into a stepwise regression model and a least square regression model, analyzing the heavy metal content and the geographic information of the soil, analyzing the statistical characteristics and the spatial variability of As, Cd, Hg and Pb elements in the surface soil of the urban group, and drawing a total content distribution diagram of 4 heavy metal elements in the surface soil of the whole large area by adopting an ArcGIS and GS + spatial analysis module;

including using the compound detected data of sampling, analysis heavy metal element content and land use, pH value, elevation, slope are to, NDVI vegetation index factor correlation carries out the source and the migration mode analysis of heavy metal, specifically includes:

41) analyzing the effect of land use/cover change (LUCC) on heavy metal distribution;

42) analyzing the influence of the pH value on the heavy metal content of the soil;

43) analyzing the influence of natural geographic factors on the heavy metal content of the soil;

44) and analyzing the influence of vegetation coverage on the heavy metal content of the soil.

5) Based on sampling composite detection data, evaluating the soil environment quality of a large area of an urban group by adopting a single-factor index and internal Merlot comprehensive pollution index method, and outputting a trend result of soil pollution space characteristic analysis in the area;

6) the method comprises the following steps of constructing a time dimension analysis mathematical model, adopting a BP neural network model and sampling composite detection data to carry out prediction analysis on the change of the soil heavy metal element content time dimension, and specifically comprising the following steps:

61) determining model structure and parameters

The constructed time dimension analysis mathematical model is a BP neural network model, and the neuron number of the input layer and the output layer of the learning algorithm in the BP neural network model is set as follows: inputting 9 neurons in the layer, wherein the 9 neurons correspond to 9 influence factors; the output neuron is 4, corresponding to the content of 4 heavy metal elements; the model is provided with 4 hidden layers, and the number of hidden layer neurons is 3;

62) data pre-processing

Preprocessing the sampling composite detection data obtained in the steps 1) to 4) by adopting a normalization method according to different dimensions, and limiting the data in a [0, 1] interval to form standardized data; the calculation formula of the normalization method is (5-4):

in the formula:

respectively representing the maximum value and the minimum value of each group of factor variables;

respectively normalizing the pre-normalization value and the normalized value of each group of factor variables;

63) performing network training

Setting network initial parameters of the BP neural network model, wherein the maximum training time is 5000, the network learning rate is 0.05, and the target root mean square error is 0.53 multiplied by 10^-3(ii) a Learning method built in by calling BP neural network learning algorithm toolTraining a sample, and obtaining a BP neural network for subsequent analysis and prediction after training;

64) performing network model validation

Verifying the trained BP neural network model through a known sample; if the prediction result is closer to the actual value, the actual analysis prediction can be carried out, otherwise, the step (63) is repeated until the prediction result is closer to the actual value;

65) and inputting the standard data to be analyzed into the trained and verified BP neural network model, performing predictive analysis, obtaining the time dimension distribution characteristics of the heavy metals in the soil in the large area, and outputting an analysis result.

A large-area soil heavy metal detection and space-time distribution characteristic analysis system for implementing the method is a distributed computer system with a B/S framework, and specifically comprises a remote server, a plurality of front-end machines and terminal equipment which are connected and communicated with each other through a network;

the remote server is internally provided with:

the system comprises a main control unit, a GIS gridding management module, a data processing module, a spatial distribution characteristic analysis module, an ArcGIS spatial analysis module, a GS + spatial analysis module and a pollution condition evaluation module;

an I/O module, an operation module and a storage module are arranged in the main control unit;

a GIS positioning management module, a sampling management module, a sample detection management module and a data acquisition module are arranged in the front-end computer;

the terminal equipment is a GIS data acquisition terminal, a handheld GPS locator, a plasma emission spectrometer and an atomic fluorescence analyzer;

the front-end computer is connected with a GIS data acquisition terminal, a handheld GPS locator, a plasma emission spectrometer and an atomic fluorescence analyzer of the terminal equipment through a data interface, and the detection data of the terminal equipment is acquired through a data acquisition module.

And the remote server is also internally provided with a soil heavy metal pollution condition evaluation module which is used for evaluating the soil heavy metal pollution condition based on the spatial distribution characteristic analysis data and the conclusion.

The remote server is also provided with a multi-target survey database module for managing data based on multi-target geochemical survey and remote sensing, and the data is called by a main control unit and other modules of the analysis system.

The remote server is also provided with a time dimension analysis module, and the change of the soil heavy metal element content time dimension is subjected to prediction analysis by adopting a BP neural network model and sampling composite detection data.

Compared with the prior art, the invention has the following advantages:

(1) according to the method and the system for detecting the heavy metal in the soil in the large area and analyzing the space-time distribution characteristics, the space and time dimension investigation analysis based on the GIS is carried out on the soil environment quality of the large area represented by the urban group according to the requirement of preventing and controlling the soil pollution in the large area such as the urban group, and the distribution characteristics of the space and time dimension are modeled and analyzed by researching the data acquisition and the arrangement of the heavy metal in the soil of the target area, so that a foundation is laid for further carrying out the environment quality evaluation of the soil heavy metal pollution, particularly the pollution evaluation and the health risk evaluation. According to the method and the system, a self-built mathematical model and a computer system are adopted, sampling point detection data are utilized, externally obtained multi-target geochemical investigation and remote sensing data are combined, and the spatial distribution characteristics of the heavy metal elements are analyzed by a multiple stepwise regression model and a least square regression model to obtain accurate results of the spatial distribution characteristics of the As, Cd, Hg and Pb elements. The analysis result shows that compared with the traditional kriging interpolation method based on sampling point data, the regression model has higher analysis precision on the heavy metal spatial distribution. The invention overcomes the defects of small sampling point area, limited sample quantity and non-uniform data format in the sampling detection and analysis process of the soil heavy metal in the prior art, and can be applied to the research of large areas such as urban groups and the like after the sampling detection and the external multi-target geochemical investigation and remote sensing data are subjected to standardized processing; meanwhile, the provided method and system can completely cover the space distribution of the whole large area and the time change of a long span, improve the representativeness, the accuracy and the trend of the analysis result, and can further analyze and evaluate the soil environment quality of the large area of the urban mass and formulate a targeted prevention and treatment scheme based on the result.

(2) According to the method and the system for detecting the heavy metal in the large-area soil and analyzing the space-time distribution characteristics, in the geospatial distribution simulation analysis, the spatial distribution of the heavy metal is predicted by utilizing the correlation between the soil variable and other influence factors, and a stepwise multiple linear regression model and a partial least square regression model are adopted. The soil environment quality analysis, evaluation and result output provided by the invention do not need manual participation after the modeling and model training are completed, the system can automatically complete analysis, operation and result output according to a built-in model, the data processing capability is strong, the analysis efficiency is high, the analysis result is objective, and new technologies such as machine learning can be applied to the relevant fields such as soil heavy metal analysis, so that the automation degree of the analysis process is improved, the process repeatability and the method verifiability are strong, and the iterative upgrade capability is strong.

(3) The invention overcomes the defects that most of the existing researches use the semi-variance function of geostatistics, but the function fitting and the theoretical model selection are greatly influenced by subjective factors, and the accuracy of the result is not high, and aiming at the obvious spatial heterogeneity of the spatial distribution of the heavy metal, a mode of combining soil sampling analysis acquisition and external data is adopted, so that abundant spatial change information of the heavy metal in the soil is obtained in multiple ways, the number of sampling points is increased, and the method is more suitable for the reality of the spatial variation degree.

(4) According to the method, a large database and a mathematical analysis model for researching the heavy metals are constructed by combining short-term sampling point test data with long-term multi-target survey data, the spatial distribution change and the change trend of the soil heavy metal migration mode are analyzed in a time dimension, and a visual result is output. The method provides a large-time span data based on the soil heavy metal spatial distribution, and researches are carried out by using artificial intelligence theories and methods such as a BP network model and the like, so that the problems of high difficulty in researching the large-area soil heavy metal change trend and few related achievements in a large-span time range are solved.

(5) In order to solve the problems of common uncertainty and low accuracy of conclusion in health risk assessment risk evaluation, the invention firstly solves the accuracy of obtaining and analyzing a large amount of soil heavy metal space-time distribution characteristic data, constructs a reasonable computer network system and establishes complete data, a method and a verification system. The multi-element stepwise regression and least square regression model constructed by the method analyzes the spatial distribution of each heavy metal element, takes factors such As pH value, elevation, gradient, slope direction and NDVI of land As auxiliary variables to analyze the distribution characteristics of four elements such As As, Cd, Hg and Pb, can fully consider the influence of environmental factors on the spatial variables, and well reappear the detailed information of the spatial and temporal changes of the heavy metal in the soil in a complex environment; compared with a traditional kriging interpolation method system based on sampling data space autocorrelation, the method has higher precision on soil heavy metal time and space distribution analysis and trend analysis results, and is wider in applicable range.

To more clearly illustrate the structural features and effects of the present invention, the following detailed description is given with reference to the accompanying drawings and specific embodiments.

Drawings

FIG. 1 is a schematic diagram of a topological structure of a large-area soil heavy metal detection and space-time distribution characteristic analysis system according to an embodiment of the present invention;

FIG. 2 is a schematic flow chart of a large-area soil heavy metal detection and spatial-temporal distribution characteristic analysis system module structure and an analysis method according to embodiment 1 of the present invention;

FIG. 3 is a schematic diagram illustrating the results of analysis of the spatial distribution of As elements in surface soil of a large area according to an embodiment of the present invention;

FIG. 4 is a schematic diagram showing the analysis result of the spatial distribution of Cd in surface soil of a large area according to the present invention;

FIG. 5 is a diagram showing the analysis result of the space distribution of Hg element in the surface soil of a large area;

FIG. 6 is a diagram showing the analysis result of the spatial distribution of Pb element in surface soil of a large area;

FIGS. 7(a) -7 (d) are schematic diagrams illustrating the evaluation results of the environmental quality of the soil single element according to the embodiment of the present invention, wherein:

FIG. 7(a) is a diagram showing the results of single factor evaluation of As element;

FIG. 7(b) is a diagram showing the evaluation result of single factor of Cd element;

FIG. 7(c) is a graph showing the evaluation results of Hg element by a single factor;

FIG. 7(d) is a diagram showing the results of single factor evaluation of Pb element;

FIG. 8 is a bar graph of the contamination levels of various elements in accordance with an embodiment of the present invention;

FIG. 9 is a diagram showing the evaluation results of the Mello comprehensive pollution index;

FIG. 10 is a schematic diagram of the results of comprehensive evaluation of soil environmental quality according to an embodiment of the present invention;

FIG. 11 is a schematic flow chart showing the modular structure and analysis method of the large-area soil heavy metal detection and spatial-temporal distribution feature analysis system in accordance with embodiment 3 of the present invention;

FIG. 12 is a schematic diagram of a BP network model construction process according to an embodiment of the present invention;

fig. 13(a) -13 (d) are schematic diagrams illustrating results of predicting heavy metal content in soils in the changsha city, where:

FIG. 13(a) is a diagram showing the predicted result of As element content;

FIG. 13(b) is a diagram showing the result of predicting Cd content;

FIG. 13(c) is a graph showing the predicted Hg content;

FIG. 13(d) is a diagram showing the predicted result of the Pb content;

fig. 14(a) -14 (d) are schematic diagrams of results of predicting heavy metal content in soil of the continental city, in which:

FIG. 14(a) is a diagram showing the predicted result of As element content;

FIG. 14(b) is a diagram showing the result of predicting Cd content;

FIG. 14(c) is a graph showing the predicted Hg content;

FIG. 14(d) is a diagram showing the predicted result of the Pb content;

fig. 15(a) -15 (d) are schematic diagrams of results of predicting the heavy metal content in the soil in Hunan Tan City, where:

FIG. 15(a) is a diagram showing the predicted result of As element content;

FIG. 15(b) is a diagram showing the result of predicting Cd content;

FIG. 15(c) is a graph showing the predicted Hg content;

FIG. 15(d) is a diagram showing the result of predicting the Pb content.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Example (b):

referring to fig. 1 to 15(d), the method for detecting heavy metals in soil in a large area and analyzing spatial-temporal distribution characteristics provided by the embodiment of the invention comprises the following steps:

1) constructing an analysis model for spatial distribution of multiple heavy metal elements in large areas such as urban groups and the like, programming the model and storing the model in a module corresponding to a remote server in advance; specifically, a multivariate stepwise regression model in geospatial regression is adopted to analyze the content of Hg and Pb in the soil, and a least square regression model is adopted to analyze the content of As and Cd in the soil;

11) selecting 80% of large-area sampling composite detection data, importing the data into a remote server for establishing a geographic space distribution analysis regression model, and simulating the space distribution of heavy metals by utilizing the correlation between soil variables and other influence factors;

the model formula of Partial Least Squares Regression (PLSR) is (3-2):

3) carrying out GIS gridding management on a large urban group area through a remote server, sampling and detecting surface soil (including a surface layer, a subsurface layer and a deep layer) in each gridding area according to a set proportion and a set place, adopting 1000m multiplied by 1000m grid area division in the large area As a sampling grid unit according to GIS information, and obtaining sampling composite detection data related to the content of As, Cd, Hg and Pb elements in the urban group soil, sampling time and GIS grids through a front-end computer; the method specifically comprises the following steps:

35) the content of the metal elements detected by each sample is respectively associated with the GIS grid information of the acquisition time and the acquisition place of the sample by a front-end computer to obtain GIS-based sampling composite detection data of the metal elements in each soil sample;

36) acquiring multi-target geochemical survey and remote sensing data from the outside, conducting standardized processing on the data, importing the data into a multi-target survey database module of a remote server, and calling the data together with composite detection data obtained by sampling points; simultaneously, according to the GIS grids of the multi-target geochemical survey and the remote sensing data, GIS grid planning of sampling points is carried out again, so that sampled composite detection data obtained repeatedly before and after are comparable;

4) at a remote server, substituting the obtained sampling composite detection data into a stepwise regression model and a least square regression model, analyzing the heavy metal content and the geographic information of the soil, analyzing the statistical characteristics and the spatial variability of As, Cd, Hg and Pb elements in surface soil of an urban group, and drawing a total content distribution diagram of 4 heavy metal elements in the surface soil of the whole large area by adopting an ArcGIS and GS + spatial analysis module;

the method specifically comprises the following steps of analyzing the correlation between the content of heavy metal elements and land utilization, pH value, elevation, gradient, slope direction and NDVI vegetation index factors by using sampled composite detection data, and analyzing the source and migration mode of the heavy metals:

5) On the remote server side, evaluating the soil environment quality of a large area of an urban group by adopting a single-factor index and inner-Mello comprehensive pollution index method based on sampling composite detection data, and outputting a trend result of soil pollution space characteristic analysis in the area;

6) the method comprises the following steps of constructing a time dimension analysis mathematical model, programming the mathematical model, inputting the mathematical model into a time dimension analysis module corresponding to a remote server, and carrying out prediction analysis on the change of the soil heavy metal element content time dimension by adopting a BP neural network model and sampling composite detection data, and specifically comprises the following steps:

61) determining model structure and parameters

62) data pre-processing

in the formula:

63) performing network training

Setting network initial parameters of the BP neural network model, wherein the maximum training time is 5000, the network learning rate is 0.05, and the target root mean square error is 0.53 multiplied by 10^-3(ii) a Training a sample by calling a learning method built in a BP neural network learning algorithm tool, and obtaining a BP neural network for subsequent analysis and prediction after training; the training can be carried out by adopting another computer which is networked with the remote server so as to accelerate the learning speed, and the obtained BP neural network model is led into the remote server after the training is finished;

64) performing network model validation

Verifying the trained BP neural network model through a known sample at a remote server end; if the prediction result is closer to the actual value, the actual analysis prediction can be carried out, otherwise, the step (63) is repeated until the prediction result is closer to the actual value;

65) and inputting the standard data to be analyzed into the trained and verified BP neural network model at a remote server end, performing predictive analysis, obtaining the time dimension distribution characteristics of the heavy metals in the soil in the large area, and outputting an analysis result.

A large-area soil heavy metal detection and space-time distribution characteristic analysis system for implementing the method is a distributed computer system with a B/S framework, and specifically comprises a remote server, a plurality of front-end machines and terminal equipment which are connected and communicated with each other through a network; the remote server may select the appropriate program server to configure.

The remote server is internally provided with:

the system comprises a main control unit, a GIS gridding management module, a data processing module, a spatial distribution characteristic analysis module, an ArcGIS spatial analysis module, a GS + spatial analysis module and a pollution condition evaluation module; each module is internally provided with a corresponding mathematical model, an analysis program or a management program;

a GIS positioning management module, a sampling management module, a sample detection management module and a data acquisition module are arranged in the front-end computer; the front-end computer can specifically select a computer, a PLC main control computer and other equipment.

The terminal equipment is a GIS data acquisition terminal, a handheld GPS locator, a plasma emission spectrometer and an atomic fluorescence analyzer; the GIS data acquisition terminal and the handheld GPS locator are used for acquiring GIS information and grid information of a sample when the sample is acquired; the plasma emission spectrometer and the atomic fluorescence analyzer are used for detecting chemical components of a sample; the specific types of the terminal and the instrument can be selected according to specific requirements.

Specific example 1:

according to the embodiment of the invention, a large area of a long pond city group in Hunan province of Hunan province is taken as a research object, Changsha is a central city of the long pond city group, province of Hunan province, and Tankan and Hunan pond are city group sub-central cities which are important industrial bases and traffic main roads in Hunan province. Y-shaped long strain pond three cities serving as core area of urban group along Xiangjiang "Distributed in a shape, and the distance between every two is about 40 km. From the view of natural geographic conditions, the three cities have the same geological background, so that the reasonability of the overall evaluation of the research area is better ensured; secondly, the landform and the developed water system in the research area play a certain role in the diffusion of elements; the industrial and agricultural development and positioning in the third Long Tan, the third City, highlights the influence of human action on the spatial distribution of heavy metals. The geographical positions of the long plant pond urban groups are located in the middle east of Hunan province, the downstream of Hunan river and the west of long flat basin, covering three main cities of Changsha, Tanzhou and Hunan pond and development planning areas thereof, and the total area is 2920 km²Accounting for 13.3 percent of the total land area of the whole province. The geographic coordinates of the study area are: e112 ° 29 '42 "-113 ° 17' 22", N27 ° 42 '05 "-28 ° 24' 57". In the aspect of soil characteristics, long plant pond soil is mainly distributed on a complete new system flushing layer of one-grade and two-grade lands at two sides of Hunan river, the upper part is sandy clay and clay sandy soil, and the lower part is sandy pebbles and sand. Affected by the action of a water system, the soil layer is weak in adhesion and is very easy to be taken away by running water. The soil in the research area is classified according to the matrix of the soil, and mainly comprises new generation sediments, chalky red sandstone, marbled sand shale, granite, slate, limestone and the like. If red soil and rice soil are classified according to the traditional soil, the red soil and the rice soil account for about 80 percent of the total soil area; the rest are moisture soil, purple soil, lime soil and other soils, and are suitable for the growth of various crops and forest trees.

The method for detecting heavy metals in large-area soil and analyzing spatial distribution characteristics, provided by the embodiment, comprises the following steps:

1) constructing an analysis model for spatial distribution of multiple heavy metal elements in a large area of a long quan urban group, analyzing the content of Hg and Pb in soil by adopting a multivariate stepwise regression model in geospatial regression, and analyzing the content of As and Cd in soil by adopting a least square regression model;

11) selecting 80% of the composite detection data of the large-area sample for establishing a geographic spatial distribution analysis regression model, and simulating the spatial distribution of the heavy metals by utilizing the correlation between the soil variables and other influence factors;

the Stepwise Multiple Linear Regression (SMLR) model has a model formula of (3-1):

the Stepwise Multiple Linear Regression (SMLR) model, stepwise regression, is to select the factors that play important roles in dependent variable y from n independent variables of the multiple linear regression to form a partial regression equation. Stepwise regression requires that x be checked one by one at each step of the calculation, ensuring that the final regression equation contains all and only those independent variables x that contribute significantly to the dependent variable y. The stepwise regression process comprises two basic steps, one is to remove the insignificant variables from the regression model, and the other is to introduce new variables into the regression model and to test the variables one by one.

The model formula of the Partial Least Squares Regression (PLSR) is (3-2):

in the formulas (3-1) and (3-2), independent variables are factors influencing the heavy metal of soil: soil pH, slope, grade, elevation, NDVI, defined as variables respectivelyx ₁～x ₅The two dumb variables of land utilization mode agricultural land and non-utilized land are respectively defined asx ₆、x ₇Dependent variable ofy ₁～y ₄Respectively represents the contents of As, Cd, Hg and Pb.

The Partial Least Squares Regression (PLSR) provided by the invention is a hidden variable method for explaining two spaces to model a covariance structure. The partial least squares are suitable for the condition that space samples are few and multiple correlations exist among variables, and integrate the advantages of principal component analysis, typical correlation analysis and linear regression analysis. In the partial least squares regression modeling, the parameters are respectively in independent variablesx And dependent variabley Extract of the above-mentioned plantst1 Andu1，t1 andu1 must be able to present data table argumentsx And dependent variabley And the degree of correlation between the two is maximized. When in uset1 Andu1after extraction, the method is carried outx To pairt1 And is returned toy To pairu1 If the equation reaches satisfactory precision, the algorithm is terminated; otherwise, the residual information is used for carrying out the second round of component extraction until the set precision is reached.

12) Selecting 20% of the composite detection data of the large-area sample for model precision detection, adopting cross validation and trend analysis to commonly detect the accuracy and precision of the model prediction result, and calculating formulas of the absolute average error MAE, the average relative error MRE and the root mean square error RMSE are as follows:

3) performing GIS gridding division on a large area of a long quan city group by adopting a WGS-84 (World Geodetic System 1984) World Geodetic coordinate System, and sampling and detecting city surface soil in each gridding area according to a set grid proportion and a set place (1000 m x 1000m is a grid according to a 1:20 ten thousand proportion), so As to obtain sampling composite detection data associated with GIS grids in the city group soil, such As content of As, Cd, Hg and Pb elements and collection time;

31) carrying out GIS meshing on a large area of an urban group, collecting soil samples in each grid, and collecting the samples by adopting a five-point plum blossom collection method;

35) and correlating the content of the metal elements detected by each sample with GIS data of the grid where the sample is located to obtain composite detection data of the content of the metal elements in each soil sample based on GIS.

36) Importing externally acquired multi-target geochemical survey and remote sensing data into a multi-target survey database module of a remote server, and calling the multi-target survey and remote sensing data together with composite detection data obtained by sampling points; the method comprises geochemical survey data of 1:20 million areas in Hunan province, soil heavy metal survey data in 1986 and 2005, ecological geochemical survey data of Changtang Tan urban area in 2002-2006, and the like; and simultaneously, carrying out GIS grid planning of sampling points according to the GIS grids of the multi-target geochemical survey and the remote sensing data, and carrying out sampling and detection by using the same GIS grids.

The external soil element content data adopted by the embodiment of the invention is data from GIS gridding collection, the obtained soil samples comprise 655 surface soil samples and 129 deep soil samples, the samples are collected by a gridding method, and 1 sampling point is arranged every square kilometer.

4) And substituting the obtained sampling composite detection data into a stepwise regression model and a least square regression model, analyzing the heavy metal content and the geographic information of the soil, analyzing the statistical characteristics and the spatial variability of the As, Cd, Hg and Pb elements in the surface soil of the urban group, and drawing the total content distribution diagram of the 4 heavy metal elements in the surface soil of the whole large area by adopting an ArcGIS and GS + spatial analysis module.

FIGS. 3 to 6 are schematic diagrams respectively showing the results of analysis of the spatial distribution of As, Cd, Hg and Pb elements in surface soil of a large area in accordance with the embodiment of the present invention.

The step 4) specifically comprises the following steps: and analyzing the source and migration mode of the heavy metal by analyzing the correlation of the content of the heavy metal elements and factors such as land utilization, pH value, elevation, gradient, slope direction, NDVI vegetation index and the like. The sources and migration modes of heavy metals are complex and changeable, and the spatial distribution of the heavy metals is influenced by various factors. Land utilization and pH value are important factors for controlling soil heavy metal accumulation and spatial distribution, and natural geographic factors and vegetation coverage also play a certain role in the distribution pattern of heavy metal content. Because the factors have a certain mutual relationship, the invention mainly relates to 6 influence factors of soil pH value, land utilization, gradient, slope direction, elevation and vegetation coverage.

41) The effect of land use/cover change (LUCC) on heavy metal distribution was analyzed,

42) analyzing the influence of the pH value on the heavy metal content of the soil,

43) analyzing the influence of natural geographic factors on the heavy metal content of the soil,

A large-area soil heavy metal detection and spatial distribution characteristic analysis system for implementing the method is a distributed computer system with a B/S framework, and specifically comprises a remote server, a plurality of front-end machines and terminal equipment which are connected and communicated with each other through a network;

the remote server is internally provided with:

the system comprises a main control unit, a GIS gridding management module, a data processing module, a spatial distribution characteristic analysis module, an ArcGIS spatial analysis module and a GS + spatial analysis module;

The remote server is also provided with a multi-target survey database module for managing externally acquired data based on multi-target geochemical survey and remote sensing, and the data is called by a main control unit and other modules of the analysis system so as to enlarge the data volume and the time span of the data.

In the embodiment, for the remote sensing data, preprocessing such as band synthesis and geometric correction is performed on the Landsat remote sensing data by adopting ENVI5.3 and ArcGISI 10.2.2 software, and then the normalized vegetation index is further extracted on the basis. The NDVI value is obtained through a band operation formula, and the value is-1. And performing format conversion on the administrative division data, superposing each influence factor distribution map, and drawing a soil heavy metal distribution map by using an ArcGIS space analysis module.

According to the embodiment of the invention, after the sampling point data and the external data are merged, the coordinates adopt a WGS-84 coordinate system, the surface soil sampling point data is input into Arcmap, and through Gaussian Kruger projection and the Western' an 80 coordinate system, a remote server can generate a shp-format surface soil heavy metal sampling point distribution diagram.

The statistical analysis of the sampled data in this embodiment is to analyze and count all the data by using SPSS 22.0 and Excel 2016. Descriptive statistics include maximum, minimum, mean, standard deviation, coefficient of variation, etc.; the data correlation adopts pearson correlation analysis; the difference analysis adopts single-factor variance analysis, and the significance level is p less than 0.05; and the influence factor analysis adopts a principal component analysis method.

Specific example 2:

the method and the system for detecting heavy metals in large-area soil and analyzing the space-time distribution characteristics are basically the same as those in the embodiment 1, and are different in that a soil heavy metal pollution condition evaluation (namely soil environment quality evaluation) module is further arranged in the remote server and used for evaluating the soil heavy metal pollution condition based on the space distribution characteristic analysis data and the analysis result.

The soil environment quality evaluation is actually a comprehensive evaluation of the ecological geochemistry, and is generally performed on a regional background value, and comprises soil native environment quality evaluation, soil environment pollution evaluation, soil resource quality evaluation and the like. The research is used for carrying out environmental quality evaluation on soil heavy metal element distribution characteristics caused by land utilization and natural reasons so as to analyze the influence of soil heavy metal distribution in an area on the soil quality, and can be further used for analyzing the relation between the soil environmental safety and the crowd health.

The space-time distribution characteristic analysis method further comprises the following steps:

5) and evaluating the soil environment quality of the long quan urban group by adopting a single factor index and an inner Meiluo comprehensive pollution index to obtain the soil pollution characteristics in the large area.

The single-factor pollution index is divided into five grades according to GB15618-2018 soil environment quality standard and calculation results of formula (2-9), and is listed in Table 1.

In the formula:

: the index of single-factor pollution is,

i: the amount of contaminants in the soil is such that,

Ci：measured concentration

Si：Screening values given in soil environmental quality standards.

TABLE 1 soil environmental quality grading

The results of the evaluation of the single factor index in this example are shown in Table 2, FIGS. 7(a) to 7(d), and FIG. 8.

TABLE 2 Single factor evaluation results

The evaluation result of the element single-factor pollution index shows that As pollution is mainly in the Changsha and distributed in a narrow and long shape along two banks of Xiangjiang river, and the pollution grade is IV-grade moderate pollution. Part of the continents are also moderately polluted, and meanwhile, a large area of light pollution is distributed in the northern part of Hunan river. As elements have strong self-purification effect on water, As is in a segmented enrichment state in the Xiangjiang river basin.

The large area is located in a Cd abnormal zone of a Xiangjiang river basin, and the total area is 2000 km²Wide distribution and high abnormal strength. The Cd heavy pollution area is concentrated in a long plant pond urban area, wherein the abnormal intensity of a plant continent is highest, and the distribution area is largest; cd of the Changsha is abnormally distributed at two banks along Hunan river, and has an obvious concentration center; the Cd abnormality of Hunan Tan is mainly in chemical enterprises along the river and surrounding areas, and there is a single-point abnormality around manganese ore in the north and around popular rivers in the south.

The Hg element pollution is mainly in the northwest direction of the Tazhou city, and light pollution areas are scattered and distributed in Hunan ponds and Changsha.

Pb element pollution is mainly in the northwest direction of the Tazhou, the whole pollution area is relatively small, single-point abnormity occurs near a vehicle factory, and single-point abnormity also exists near a Hunan Tan chemical enterprise, which shows that the Pb abnormity is greatly related to industrial pollution.

The method for evaluating the result by adopting the inner Merlot comprehensive index specifically comprises the following steps:

the single-factor pollution index method can only reflect the pollution of a single element to the environment, and cannot show the comprehensive effect of all pollutants, so the internal Merolow (Nemerow) comprehensive pollution index method is introduced to comprehensively evaluate 4 heavy metal elements in the soil. The calculation formula is as follows:

in the formula:

P _Nis the comprehensive pollution index of the inner Metro solution,P _istandard single factor pollution index = heavy goldBelongs to the measured value of elements/the standard value of soil environment quality,nas to the number of samples,P _i ² _maxis a heavy metal elementi Maximum value of the contamination index.

And (3) grading the comprehensive soil pollution degree according to the calculation result of the formula (2-10), wherein the comprehensive soil quality evaluation grading standard is defined according to the GB15618-2018 standard and the division method of DZ T0295-2016 land quality geochemistry evaluation standard (see Table 3).

TABLE 3 soil quality comprehensive evaluation grading Standard

The soil environment quality of the long quan urban group is comprehensively evaluated by adopting the inner Mello comprehensive pollution index, and the output visualization results are shown in fig. 9 and 10. As can be seen from fig. 9, the comprehensive pollution condition of the large area is 6.49% of the clean area, the ratio of moderate pollution to severe pollution reaches 26.84%, the ratio of severe pollution exceeds moderate pollution, the ratio of light pollution to light pollution is 29.51% and 37.16%, respectively, and the pollution degrees are ranked as follows: mild > severe > moderate > clean zone.

As can be seen from FIG. 10, the soil heavy metals are enriched along Xiangjiang river basin and show obvious tendency. The pollution conditions of three main cities of long quan are more serious than those of other areas, the severe pollution grades are distributed most widely, and the moderate and severe pollution areas are distributed in a block shape along the river basin of Hunan river. Compared with three cities, the pollution condition is slight sand, and the enrichment characteristic is that the sand is distributed in a long strip shape along Hunan river; the heavy metal distribution of Hunan pond has no obvious concentration center, and the single-point abnormity is influenced by the distribution of heavy industrial enterprises; the most serious pollution condition is the continent, the heavy metal pollution in the region is particularly serious, the pollution area is distributed most widely, and the continuity is best, which is irrelevant to the situation that the continent is a heavy industrial city.

The results show that: as is in a segmented enrichment state in the Yangtze river basin, and Cd is distributed in a strip shape in the south-north direction along the plant continent-Hunan pond-Changsha-Wangcheng; hg. The Pb pollution is light as a whole, and the pollution is heavy in the northwest direction of the continents; the 4-element comprehensive pollution presents obvious tendency along the Xiangjiang river basin, and the moderate and severe pollution areas are distributed in a block shape along the Xiangjiang river basin; the three urban pollution levels are ranked as continent > Hunan Tan > Changsha.

Specific example 3:

referring to fig. 11 to 12, the method and system for detecting heavy metal in soil in a large area and analyzing characteristics of space-time distribution provided in the embodiments of the present invention are substantially the same as those in embodiments 1 to 3, and are different in that the remote server further includes a time dimension analysis module for performing prediction analysis on changes in the time dimension of the content of heavy metal elements in soil by using a BP neural network model and sampled composite detection data.

The formation and development of soil are complex, and the soil has regularity and is influenced by human activities. In order to maintain the ecological environment, the soil environment quality is generally required to be known, which means that the soil must be subjected to geochemical investigation and research, and the work is time-consuming and labor-consuming, and particularly, the large-area investigation is more difficult from the economic and time aspects. Therefore, when the soil environment is evaluated, the invention provides the method for analyzing the enrichment migration rule of the pollutants in the soil according to the existing historical data so as to predict the content and the evolution trend of various pollutants in the soil in a period of time in the future, thereby researching and formulating the targeted prevention and treatment measures.

6) and (3) constructing a time dimension analysis mathematical model, and predicting and analyzing the change of the heavy metal element content of the soil in time dimension by adopting a BP neural network model and sampling composite detection data. The factor selection of this embodiment is specifically: selecting annual rainfallx ₁Total value of domestic productionx ₂Industrial gross output valuex ₃Total populationx ₄Total agricultural yieldx ₅Total amount of harmful waste waterx ₆Total amount of exhaust gas dischargedx ₇Amount of harmful solid waste producedx ₈Area of green land at the end of nine yearsx ₉The data of 9 large influence factors in 1986-2017 are collected and collated by taking the data as main factors influencing the change of heavy metals and through the statistics yearbook of Hunan province and Long Tan san City. Based on the obtained continuous 20-year time sequence data of the heavy metal content of the long pond three city soil and 9 influence factors, determining a function relation between the heavy metal content and multiple factors by adopting a BP neural network learning algorithm, and analyzing and calculating the average value of the heavy metal content of the long pond three city soil in 2006-2017 according to the statistical yearbook data of the influence factors.

Because factors influencing the content change of the heavy metal elements are more, the accumulation and purification processes of the heavy metals exist in the soil at the same time, and the evolution law of the heavy metals cannot be accurately described by a common regression method, the heavy metal content of the soil is predicted by selecting a BP network model in the embodiment of the invention. The BP network is also called an error back propagation neural network and consists of an input layer, an output layer and a hidden layer, the model can classify any complex mode through self-training and learning, and the optimal estimation value is calculated when the input value is given. The method specifically comprises the following steps:

61) determining model structure and parameters

62) data pre-processing

in the formula:

the pre-normalized and post-normalized values for each set of factor variables are separately determined.

63) Performing network training

Setting network initial parameters of the BP neural network model, wherein the maximum training time is 5000, the network learning rate is 0.05, and the target root mean square error is 0.53 multiplied by 10^-3(ii) a Training a sample by calling a learning method built in a BP neural network learning algorithm tool, and obtaining a BP neural network for subsequent analysis and prediction after training;

64) performing network model validation

Verifying the trained BP neural network model through a known sample; if the prediction result is closer to the actual value, the actual analysis prediction can be carried out, otherwise, the step (63) is repeated until the prediction result is closer to the actual value; the training fitting effect of this embodiment is shown in fig. 13(a) -15 (d), and the prediction result is closer to the actual value, so that it can be used for actual analysis and prediction.

65) Inputting the standard data to be analyzed into the trained and verified BP neural network model, performing predictive analysis, obtaining the time dimension distribution characteristics of the heavy metals in the soil in the large area, and outputting the analysis result, which is shown in Table 4.

Specifically, based on 9 factor data in 2006-2017, the heavy metal content of the soil in the three cities of long quan is predicted by using a BP model. Table 4 lists the average content of heavy metal elements in each market in 2017 and the change rate between 2005 and 2017. According to the predicted value, the average content of heavy metal elements in each region is slowly increased in 2005-2017 years, and even a negative increase phenomenon occurs in individual years. The growth of 4 heavy metal elements in the whole plant continent is larger than that of Changsha and Hunan puddle, and the growth rates of different elements show that the increase of Pb, Cd and Hg is relatively more. The results show that since 2005, ecological environmental awareness of the long pond urban group is comprehensively improved, government strictly controls the discharge of industrial three wastes, and measures such as shut-down, transformation and relocation of large-scale pollution enterprises in the old industrial area of the shoal clean pond and ecological pollution control in various cities are implemented to ensure that the heavy metal content of the whole long pond is weakened in comparison with the prior growth trend; however, in recent years, the number of motor vehicles is greatly increased due to the improvement of the living standard of people, and the emission of automobile exhaust is also a pollution source with increased heavy metal content.

Table 4 mean value change/(mg-kg) of heavy metal element content in long pond soil 2005-2017^-1)

The method analyzes the distribution characteristics of four elements of As, Cd, Hg and Pb in a large area of an urban group by constructing a new mathematical model, screening a typical detection object, comprehensively utilizing sampling point data and multi-target geochemical survey and remote sensing data, has complete area coverage of an analysis result, high accuracy and high automation degree of an analysis process, does not need manual intervention, and can further analyze and evaluate the soil environment quality of the urban group, develop the trend and predict and analyze the heavy metal pollution health risk.

According to the method and the system for detecting heavy metals in the large-area soil and analyzing the space-time distribution characteristics, which are provided by the invention, the self-built mathematical model and the computer system are adopted to carry out GIS-based spatial and time dimension investigation and analysis on the environmental quality of the large-area soil represented by the urban group aiming at the requirement of preventing and treating soil pollution in the large areas such as the urban group, and compared with the traditional kriging interpolation method based on sampling point data, the regression model has higher precision in analyzing the space distribution of the heavy metals. The invention overcomes the defects of small sampling point area, limited sample quantity and non-uniform data format in the sampling detection and analysis process of the soil heavy metal in the prior art, and can be applied to the research of large areas such as urban groups and the like after the sampling detection and the external multi-target geochemical investigation and remote sensing data are subjected to standardized processing; meanwhile, the provided method and system can completely cover the space distribution and the long-span time change of the whole large area, improve the representativeness, the accuracy and the trend of the analysis result, further perform analysis and evaluation on the soil environment quality, the change trend and the like of the large area of the urban mass based on the result, and lay a foundation for making a targeted prevention and control scheme.

While the present invention has been described in detail and with reference to the embodiments thereof as illustrated in the accompanying drawings, it will be apparent to one skilled in the art that various changes and modifications can be made therein. Therefore, certain details of the embodiments are not to be interpreted as limiting, and the scope of the invention is to be determined by the appended claims.

Claims

1. A large-area soil heavy metal detection and space-time distribution characteristic analysis method is characterized by comprising the following steps:

1) constructing an analysis model for spatial distribution of a plurality of heavy metal elements in a large area, analyzing the contents of Hg and Pb in soil by adopting a multivariate stepwise regression model in geospatial regression, and analyzing the contents of As and Cd in soil by adopting a least square regression model;

3) gridding a large urban group area, sampling and detecting urban surface soil in each gridding area according to a set proportion and a set place, and obtaining the content of As, Cd, Hg and Pb elements in the urban group soil and sampling composite detection data associated with sampling time and a GIS grid;

2. The method for large-area soil heavy metal detection and spatial-temporal distribution feature analysis according to claim 1, further comprising the steps of:

5) based on the sampling composite detection data, a single-factor index and inner-merozoite comprehensive pollution index method is adopted to evaluate the soil environment quality of the large area, and trend results of soil pollution space characteristic analysis in the area are output.

3. The method for large-area soil heavy metal detection and spatial-temporal distribution feature analysis according to claim 1, further comprising the steps of:

6) and (3) constructing a time dimension analysis mathematical model, and predicting and analyzing the change of the heavy metal element content of the soil in time dimension by adopting a BP neural network model and sampling composite detection data.

4. The method for detecting heavy metals in large-area soil and analyzing characteristics of space-time distribution according to claim 3, wherein the step 6) specifically comprises the following steps:

61) determining model structure and parameters

62) data pre-processing

Preprocessing the sampling composite detection data obtained in the steps 1) to 4) by adopting a normalization method according to different dimensions, and limiting the data in a [0, 1] interval to form standardized data; the calculation formula of the normalization method is as follows:

in the formula:

63) performing network training

64) performing network model validation

5. The method for large-area soil heavy metal detection and spatial-temporal distribution feature analysis according to claim 1, wherein the step 4) further comprises:

and analyzing the correlation between the content of the heavy metal elements and land utilization, pH value, elevation, gradient, slope direction and NDVI vegetation index factors by using the sampling composite detection data, and analyzing the source and migration mode of the heavy metals.

6. The large-area soil heavy metal detection and space-time distribution feature analysis method according to claim 5, characterized by comprising the following steps:

7. The method for detecting heavy metals in soil in large area and analyzing characteristics of spatial distribution according to claim 1, wherein the step 1) further comprises the following steps:

in the step-by-Step Multiple Linear Regression (SMLR) model in the step 1), the step-by-step regression is a partial regression equation formed by selecting factors which play an important role in the dependent variable y from n independent variables of the multiple linear regression;

stepwise regression requires that x be checked one by one at each step of the calculation, ensuring that the final regression equation contains all and only those independent variables x that have significant effect on the dependent variable y; the stepwise regression process comprises two basic steps, namely removing the non-significant variables from the regression model, and introducing new variables into the regression model and checking the variables one by one; the model formula is (3-1):

the model formula of Partial Least Squares Regression (PLSR) is (3-2):

8. The method for large-area soil heavy metal detection and spatial-temporal distribution feature analysis according to claim 7, further comprising:

9. The method for detecting heavy metals in large-area soil and analyzing characteristics of space-time distribution according to claim 1, wherein the step 3) specifically comprises the following steps:

35) and respectively associating the content of the metal elements detected by each sample with the GIS grid information of the acquisition time and the acquisition place of the sample to obtain GIS-based sampling composite detection data of the metal elements in each soil sample.

10. The method for detecting heavy metals in soil in large area and analyzing characteristics of space-time distribution according to claim 9, wherein the step 3) specifically comprises the following steps:

step 36): acquiring multi-target geochemical survey and remote sensing data from the outside, conducting standardized processing on the data, importing the data into a multi-target survey database module of a remote server, and calling the data together with composite detection data obtained by sampling points; and simultaneously, according to the GIS grids of the multi-target geochemical survey and the remote sensing data, planning the GIS grids of the sampling points again, so that the sampled composite detection data obtained repeatedly before and after the sampling has comparability.

11. A large-area soil heavy metal detection and space-time distribution characteristic analysis system for implementing the method of any one of claims 1 to 10, which is a distributed computer system with a B/S architecture, and specifically comprises a remote server, a plurality of front-end machines and terminal equipment which are connected and communicated with each other through a network;

the remote server is internally provided with:

12. The large-area soil heavy metal detection and spatial-temporal distribution characteristic analysis system according to claim 11, wherein the remote server is further provided with a soil heavy metal pollution condition evaluation module for evaluating the soil heavy metal pollution condition based on the spatial distribution characteristic analysis data and the conclusion.

13. The large-area soil heavy metal detection and space-time distribution characteristic analysis system according to claim 11, wherein the remote server is further provided with a multi-target survey database module for managing data based on multi-target geochemical survey and remote sensing for the main control unit and other modules of the analysis system to call.

14. The large-area soil heavy metal detection and spatial-temporal distribution feature analysis system according to claim 11, wherein the remote server is further provided with a time dimension analysis module, and the time dimension change of the heavy metal element content in the soil is subjected to prediction analysis by adopting a BP neural network model and sampling composite detection data.