A method, system and medium for classified control of heavy metal pollution in cultivated land

Abstract

The present invention provides a method, system and medium for classified control of heavy metal pollution in cultivated land, relating to the technical field of environmental protection, including: obtaining remote sensing image data of a strictly controlled area for heavy metal pollution, and dividing the strictly controlled area into multiple plots with different environmental conditions by using the object-oriented method; constructing multiple suitability constraint conditions for the control measures in the strictly controlled area by using the distance to the town, the distance to the industrial area, the slope, the production capacity, the control measures and the area variable; taking the expenditure cost, the cultivated land area and the safety utilization rate of various control measures in the strictly controlled area as the constraint conditions of the control measures in the strictly controlled area, and combining with the suitability constraint conditions to obtain the classified control measures with the minimum total cost in the strictly controlled area for heavy metal pollution. The above method helps to realize the zonal and block management of cultivated land polluted by heavy metals.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental protection, and particularly relates to a method, a system and a medium for classified control of heavy metal pollution in cultivated land. Background Art

[0002] Controlling heavy metal pollution in cultivated land is of great significance for protecting the stability of the ecosystem, ensuring the quality and safety of agricultural products and maintaining human health, and helps to achieve the sustainable utilization of land resources and promote sustainable development.

[0003] At present, for plots contaminated by heavy metals, environmental supervision agencies usually adopt on-site inspection methods, and the control measures for different classifications of heavy metal pollution are different. For example, in the safe utilization area, agronomic control measures such as replacing with low-accumulation varieties and adjusting soil acidity and alkalinity are implemented; in the strict control area, the planting of edible crops is prohibited, and measures such as adjusting the planting structure and returning farmland to forests or grasslands are implemented.

[0004] However, the on-site inspection method is costly and inefficient, and it is impossible to conduct large-area and long-term supervision and management; the soil conditions, surrounding environments, etc. within the same partition vary greatly, and it is relatively general about what control measures should be taken for each specific plot, lacking practicality. Summary of the Invention

[0005] The present invention provides a method, a system and a medium for classified control of heavy metal pollution in cultivated land. The above method helps to achieve the management of heavy metal polluted cultivated land by regions and blocks.

[0006] The present invention provides a method for classified control of heavy metal pollution in cultivated land, including:

[0007] Obtaining remote sensing image data of the strict control area of heavy metal pollution, and dividing the strict control area of heavy metal pollution by the object-oriented method to obtain each plot of the heavy metal pollution partition;

[0008] Obtaining the crop types and rotation patterns on each plot in the strict control area, and determining the control measures for each type of strict control area, including changing the land use type, adjusting the planting structure and ecological conversion of farmland;

[0009] Taking the distance to the town, the distance to the industrial area, the slope and the production capacity as influencing factors of the control measures for each type of strict control area to construct multiple scenarios, taking the controllable area of implementing each control measure in each scenario as the area variable, and constructing multiple suitability constraints for the control measures of the strict control area according to the scenarios, the control measures and the area variable;

[0010] A multi-objective linear programming model was constructed, and the expenditure costs of various control measures in strict control areas were taken as the objective function of the multi-objective linear programming model. The cultivated land area and safe utilization rate were taken as constraints for the control measures in strict control areas. Combined with multiple suitability constraints, the multi-objective linear programming model was used to obtain classified control measures that minimized the total cost of strict control areas of heavy metal pollution under various scenarios.

[0011] Preferably, the method of dividing the heavy metal pollution strict control area by using an object-oriented method comprises the following steps:

[0012] Perform multi-scale segmentation on remote sensing image data and generate image object layers at different segmentation scales, wherein the characteristics of the image objects after multi-scale segmentation include: spectrum, texture and geometry;

[0013] Select the feature space and assign different weights to each feature to obtain the classification results of the land objects;

[0014] According to the classification results of the ground objects, the hierarchical structure corresponding to the ground objects is obtained, and the image object layers are connected according to the hierarchical structure corresponding to the ground objects to obtain the various plots of heavy metal pollution zones.

[0015] Preferably, the multi-objective linear programming model is constructed, and the expenditure of various types of control measures in strict control areas is used as the objective function of the multi-objective linear programming model, including:

[0016] Construct a multi-objective linear programming model:

[0017] The expenditures of various control measures in strictly controlled areas are taken as the objective functions of the multi-objective linear programming model. The expenditures of control measures in strictly controlled areas include the expenditures of changing land use types, adjusting planting structures and ecological land withdrawal, which are respectively set as the objective functions F 1 ( X ), F 2 ( X )and F 3 ( X ),but:

[0018] MinF 1 ( X )=∑ C 1i × X a ;

[0019] MinF 2 ( X )=∑ C 2i ×X b ;

[0020] MinF 3 ( X ) = ∑ C 3i × X c ;

[0021] In the formula: C 1i is the cost per unit area of taking the control measure of changing land use type in the i th scenario, X a is the area of contaminated cultivated land taking the control measure of changing land use type in the i th scenario, where a = 3 i - 2; C 2i is the cost per unit area of taking the control measure of adjusting planting structure in the i th scenario, X b is the area of contaminated cultivated land taking the measure of adjusting planting structure in the i th scenario, where b = 3 i - 1; C 3i is the cost per unit area of taking the control measure of ecological conversion of cropland to forest or grassland in the i th scenario, X c is the area of contaminated cultivated land taking the control measure of ecological conversion of cropland to forest or grassland in the i th scenario, where c = 3 i ;

[0022] The weight coefficients of the three objective functions are all 1 / 3. The multi-objective function is converted into a single-objective function:

[0023] ;

[0024] Construct the cultivated land area constraint condition, and the cultivated land area constraint condition is that the sum of the cultivated land areas taking various control measures is equal to the total cultivated land area in the strict control area;

[0025] Construct the safety utilization rate constraint condition, and the safety utilization rate constraint condition is that the safety utilization rate of contaminated cultivated land reaches the preset standard or above;

[0026] According to the constraint conditions and the suitability constraint conditions, obtain the classification control measures with the minimum total cost.

[0027] Preferably, it further includes: applying a decision tree model, taking the expenditure costs of various control measures in strictly controlled areas as the objective function of the decision tree model, taking the cultivated land area and safety utilization rate as the constraint conditions for the control measures in strictly controlled areas, and combining multiple suitability constraint conditions to obtain the classification control measures with the minimum total cost for the heavy metal pollution strictly controlled areas in each scenario through a multi-objective linear programming model.

[0028] Preferably, the obtaining of the crop types and rotation patterns on each plot in the strictly controlled area includes:

[0029] Collecting data of multiple sample points of remote sensing image data, and annotating the land cover types and rotation patterns for each sample point data;

[0030] Obtaining the spectral feature data of each sample point data, and calculating the index feature data based on the spectral feature data;

[0031] Using the extracted feature data and the annotation data of the sample points to train a random forest classification model, and using the trained random forest classification model to obtain the crop types and rotation patterns on each plot in the safe utilization area, and thereby determining the various control measures for changing the land use type and adjusting the planting structure in the strictly controlled area.

[0032] Preferably, the bands of the spectral feature data include: B, G, R, NIR, SWIR1, and SWIR2, and the index feature data includes: NDVI, EVI, NDWI, LSWI, NDSI, NDSVI, NDTI, WDVI, and NDFI9.

[0033] The present invention also provides a classification control system for cultivated land heavy metal pollution, which is used to implement the classification control method for cultivated land heavy metal pollution as described in any one of the above. The classification control system for cultivated land heavy metal pollution includes:

[0034] A plot division module, which is used to obtain remote sensing image data of the heavy metal pollution strictly controlled area, and divide the strictly controlled area into multiple plots with different environmental conditions by using the object-oriented method;

[0035] A plot analysis module, which is used to obtain the crop types and rotation patterns on each plot in the strictly controlled area;

[0036] A constraint condition construction module, which is used to take the distance to the town, the distance to the industrial area, the slope, and the production capacity as the influencing factors for the control measures in the strictly controlled area, construct multiple scenarios, take the controllable area of implementing various control measures in each scenario as the area variable in the constraint equation, and construct multiple suitability constraint conditions for the control measures in the strictly controlled area according to the scenarios, control measures, and area variables;

[0037] The classification control module is used to take the expenditure costs, cultivated land areas, and safety utilization rates of various strictly controlled area control measures as the constraint conditions of the strictly controlled area control measures, and combine the suitability constraint conditions to obtain the classification control measures with the minimum total cost for the heavy metal pollution strictly controlled area.

[0038] The present invention also provides a computer-readable storage medium, on which a data processing program is stored. When the data processing program is executed by a processor, the steps of the classification control method for cultivated land heavy metal pollution described in any one of the above are implemented.

[0039] Compared with the prior art, the present invention divides different sub-areas of the research area into individual plots, obtains the crop types and rotation patterns of each plot, establishes a multi-objective linear programming model, and combines the classification control measures with the minimum cost of polluted cultivated land in the classification control measures. The zoning and classification control methods can effectively achieve a higher degree of adaptability in the supervision and management of plots with different conditions, the overall control is more feasible, and the classification control measures effectively improve the control efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 is a flowchart of the classification control method for cultivated land heavy metal pollution in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0042] Embodiment 1

[0043] A classification control method for cultivated land heavy metal pollution according to the present invention has a specific process as Figure 1 shown, including the following steps:

[0044] Step S101, obtaining remote sensing image data of the heavy metal pollution strictly controlled area, and dividing the strictly controlled area into multiple plots with different environmental conditions by using the object-oriented method.

[0045] This step is used to divide each sub-area of heavy metal pollution into individual plots so that the soil conditions and surrounding environment within the same plot are consistent.

[0046] By using the object-oriented method, the plot boundaries of each sub-area of heavy metal pollution are extracted. The object-oriented extraction of remote sensing information includes two processes: image segmentation and image classification. The former generates image objects and provides a basis for the latter.

[0047] The pollution conditions of the cultivated land soil and crops in the strictly controlled areas are relatively serious, and it is usually impossible to continue growing edible crops. The land use type can be changed, the planting structure can be adjusted, or ecological conversion of farmland can be carried out according to the specific conditions of the polluted cultivated land.

[0048] Of course, the above control measures depend on the specific pollution situation and relevant policies. The control measures themselves are not the core technical features provided by this application. Therefore, the specific possible methods adopted will not affect the evaluation of the creativity of this application.

[0049] Obtaining each plot in the heavy metal pollution zone by using the object-oriented method includes:

[0050] Performing multi-scale segmentation on the remote sensing image data to generate an image object layer at different segmentation scales. Among them, the features of the image objects after multi-scale segmentation include: spectrum, texture, and geometry;

[0051] Selecting a feature space and assigning different weights to each feature to obtain a land cover classification result;

[0052] Obtaining a hierarchical structure corresponding to the land cover according to the land cover classification result, and connecting the image object layer according to the hierarchical structure corresponding to the land cover to obtain each plot in the heavy metal pollution zone.

[0053] Through the above steps, land cover classification is performed on the remote sensing image data, and each plot in the heavy metal pollution zone is obtained by using the land cover classification result. The object-oriented method is elaborated in detail below.

[0054] Multi-scale segmentation: Use the multi-scale segmentation algorithm to segment the remote sensing image to generate an image object layer. This process segments the image into regions of different sizes, and each region is regarded as an object. Segmenting at different scales can better capture the spatial information of the land cover, thereby improving the accuracy of classification. Among them, each image object contains its features in terms of spectrum, texture, and geometry.

[0055] Feature extraction and weight assignment: For each image object, the corresponding features need to be extracted from its spectrum, texture, and geometric features.

[0056] According to the importance of the features, different weights are assigned to different features. For example, the spectral feature may have the greatest impact on land cover classification, so a higher weight may be assigned, while the texture and geometric features may be assigned lower weights.

[0057] Land classification uses the extracted features and weights to classify image objects. Various classification algorithms can be used, such as Support Vector Machine (SVM), Random Forest, etc. According to the extracted features, the image objects are classified and categorized into different land types, such as buildings, vegetation, water bodies, etc.

[0058] Obtain the corresponding hierarchical structure of land features. Based on the land classification results, a hierarchical structure corresponding to the land features can be established. This hierarchical structure can be a tree structure, reflecting the hierarchical relationship between different land types. For example, buildings can be divided into subcategories such as residential, commercial, and industrial, and vegetation can be divided into subcategories such as forests, grasslands, and farmlands, and so on. This hierarchical structure helps to organize and manage the land classification results and provides more abundant information to describe the spatial distribution and relationship of land features.

[0059] Connect the image object layer. Based on the land classification results and the corresponding hierarchical structure of land features, the image object layer is connected according to this hierarchical structure. In this way, each plot in the heavy metal pollution zoning can be obtained. Each plot can be classified as a specific land type and has its characteristic and location information, that is, the soil conditions and the surrounding environment within the same plot are consistent.

[0060] In step S102, the distances from towns, industrial areas, slopes, and production capacities are used as influencing factors for the control measures in the strictly controlled areas. Multiple scenarios are constructed. The controllable areas for implementing various control measures in each scenario are used as area variables in the constraint equations. Multiple suitability constraint conditions for the control measures in the strictly controlled areas are constructed based on the scenarios, control measures, and area variables.

[0061] In step S103, the expenditure costs, cultivated land areas, and safety utilization rates of various control measures in the strictly controlled areas are used as constraint conditions for the control measures in the strictly controlled areas. Combining with the suitability constraint conditions, the classification control measures with the minimum total cost in the heavy metal pollution strictly controlled areas are obtained. The above two steps are used to obtain the classification control measures with the minimum total cost for each plot in the heavy metal pollution zoning.

[0062] The above solution involves a multi-objective linear programming model, which is a mathematical optimization model used to solve decision-making problems considering multiple objectives simultaneously. In multi-objective linear programming, there are multiple objective functions, and each objective function needs to be minimized or maximized.

[0063] The multi-objective linear programming model has two or more objective functions, and all objective functions and constraints are linear. The multi-objective linear programming model is used to determine the classification control measures and their regions and areas for the cultivated land in the strict control area. There are many ways to control the polluted cultivated land in the strict control area. The costs of different control measures are different, and the impact on the cultivated land area, production capacity, etc. is also different. Therefore, the multi-objective linear programming model is used to weigh the costs of various control measures. Under the requirements of the constraints such as cultivated land area and safe utilization rate, the measures with the least total cost of cultivated land control in the strict control area are obtained.

[0064] Firstly, the distance to towns, distance to industrial areas, slope and production capacity are taken as influencing factors of control measures in strict control areas. Multiple scenarios are constructed, and the controllable area for implementing various control measures in each scenario is used as the area variable in the constraint equation. Based on the scenarios, control measures and area variables, multiple suitability constraints for control measures in strict control areas are constructed.

[0065] Furthermore, the expenditures, cultivated land area and safe utilization rate of various control measures in strict control areas are taken as constraints of control measures in strict control areas. Combined with the suitability constraints, classified control measures that minimize the total cost of strict control areas of heavy metal pollution are obtained, including:

[0066] The expenditures of the control measures in the strict control area include: the expenditures of changing the land use type, adjusting the planting structure and ecological land withdrawal, which are set as the objective functions F 1 ( X ), F 2 ( X )and F 3 ( X ),but:

[0067] MinF 1 ( X )=∑ C 1i × X a ;

[0068] MinF 2 ( X )=∑ C 2i × X b ;

[0069] MinF 3 ( X )=∑ C 3i ×X c .

[0070] Wherein: C 1i is the cost per unit area of taking control measures for changing land use types in the i th scenario, X a is the area of contaminated cultivated land where control measures for changing land use types are taken in the i th scenario. Among them, a = 3 i - 2; C 2i is the cost per unit area of taking control measures for adjusting planting structure in the i th scenario, X b is the area of contaminated cultivated land where measures for adjusting planting structure are taken in the i th scenario. Among them, b = 3 i - 1; C 3i is the cost per unit area of taking ecological conversion of cropland control measures in the i th scenario, X c is the area of contaminated cultivated land where ecological conversion of cropland control measures are taken in the i th scenario. Among them, c = 3 i .

[0071] The weight coefficients of the three objective functions are all 1 / 3, and the multi-objective function is converted into a single-objective function:

[0072] .

[0073] The cultivated land area constraint condition is that the sum of the cultivated land areas where various control measures are taken is equal to the total cultivated land area in the strict control area.

[0074] The safety utilization rate constraint condition is that the safety utilization rate of contaminated cultivated land reaches the preset standard or above.

[0075] According to the constraint conditions and suitability constraint conditions, classification control measures with the minimum total cost are obtained.

[0076] In the above steps, the constraints on cultivated land area, safety utilization rate, and cultivated land productivity depend on the actual situation and are not the focus of the present invention.

[0077] When the cultivated land is in the safe utilization area for heavy metal pollution, it is necessary to obtain the crop types and rotation patterns on each plot in the strict control area.

[0078] Obtaining the crop types and rotation patterns on each plot in the strictly controlled area includes:

[0079] Collecting data of multiple sample points of remote sensing image data, and annotating the land cover types and rotation patterns for each sample point data;

[0080] Obtaining the spectral feature data of each sample point data, and calculating the index feature data based on the spectral feature data;

[0081] Using the extracted feature data and the annotation data of the sample points to train a random forest classification model, and using the trained random forest classification model to obtain the crop types and rotation patterns on each plot.

[0082] Since crops all grow on cultivated land and the texture differences are not significant, the image data after fusing multiple features such as spectral features and index features is used for the automatic classification of crop types in the study area.

[0083] Among them, the bands of the spectral feature data include: B, G, R, NIR, SWIR1, and SWIR2, and the index feature data includes: NDVI, EVI, NDWI, LSWI, NDSI, NDSVI, NDTI, WDVI, NDFI9.

[0084] The specific implementation steps for obtaining the crop types and rotation patterns on each plot are shown in the following embodiments.

[0085] The second embodiment of this application provides an example of classified control of heavy metal pollution in cultivated land in the strictly controlled area using the method of the first embodiment.

[0086] The study area is located in the hilly area of the southeast coast of China. Generally, the plots are relatively fragmented, and the cultivated land is generally scattered among the crisscrossing roads, river networks, and residential areas. Based on the high-spatial-resolution Google Earth image data, this embodiment uses the object-oriented method in the E-Cognition software to extract the plot boundaries in typical locations such as severely heavy metal polluted areas, areas with complex planting structures, and areas around cities.

[0087] In the GEE platform, the random forest algorithm is used to identify the crop types and rotation patterns on cultivated land. Based on the Sentinel-2 and Landsat-7 / 8 image data, the random forest algorithm is used, combined with the characteristics of crop types and classification sample points, to obtain the spatial distribution of crop types and rotation patterns in the study area, and to verify the classification accuracy. Specifically as follows:

[0088] Since crops all grow on cultivated land and have little texture difference, the image data after fusing multiple features such as spectral features and index features are used for the automatic classification of crop types in the study area. The spectral features include the spectral features of 6 bands: B, G, R, NIR, short-wave infrared 1 (SWIR1) and short-wave infrared 2 (SWIR2). Based on the 6 bands, 9 index features are calculated: NDVI, EVI, NDWI, land surface water index (LSWI), normalized snow cover index (NDSI), normalized difference senescent vegetation index (NDSVI), normalized difference tillage index (NDTI), weighted difference vegetation index (WDVI), and normalized difference flood index (NDFI).

[0089] According to relevant statistical data, the main crops in the study area are rice, rapeseed, and wheat. Since rapeseed and wheat are overwintering crops growing on dry land and the harvest time is in the first half of the year, in this embodiment, the remote sensing images in the first half of the year are selected to divide the dry land crops into three categories: rapeseed, wheat, and others. Since rice grows in paddy fields and there are single-season rice and double-season rice, in this embodiment, the remote sensing images of the whole year are selected to divide the paddy field crops into five categories: early rice, middle rice, late rice, double-season rice, and others. Due to the existence of crop rotation in the study area, in this embodiment, seven major planting structures are further obtained according to the geographical location and growth period of the crops: rice-rape rotation, rice-wheat rotation, single-season rice, double-season rice, rapeseed, wheat, and other crops. Based on the crop sample points collected on the spot and combined with the high-resolution images of Sentinel-2 and Google Earth, a total of 705 sample point data are collected, including 109 rapeseed samples, 104 wheat samples, 65 early rice samples, 119 middle rice samples, 59 late rice samples, and 98 double-season rice samples. The various samples are randomly divided into training samples and validation samples according to a ratio of about 7:3.

[0090] As shown in Table 1, the overall accuracy (OA) of the dry land crop classification in the study area is 86.05%, and the Kappa coefficient is 0.7873, indicating a high degree of consistency and the classification result reaching a relatively high accuracy. Among them, the user accuracy of rapeseed is the highest, reaching 90.91%, and the mapping accuracy of wheat is the highest, reaching 89.66%.

[0091] Table 1 Statistical table of mapping accuracy of dry land crops

[0092]

[0093] As shown in Table 2, the overall accuracy (OA) of the paddy field crop classification in the study area is 82.68%, and the Kappa coefficient is 0.7803, indicating a high degree of consistency and the classification result reaching a relatively high accuracy. Among them, the user accuracy of late rice is the highest, reaching 88.89%, and the mapping accuracy of middle rice is the highest, reaching 86.11%.

[0094] Table 2 Statistical Table of Mapping Accuracy of Paddy Field Crops

[0095]

[0096] In this embodiment, a multi-objective linear programming model is used to determine the classified control measures, their areas and regions for the cultivated land in the strictly controlled areas. There are various control methods for the polluted cultivated land in the strictly controlled areas. The costs of different control measures vary, and their impacts on cultivated land area, production capacity, etc. are also different. Therefore, a multi-objective linear programming model is adopted to balance the costs of various control measures, and under the requirements of meeting the constraints such as cultivated land area and safety utilization rate, a measure plan with the least total control cost for the cultivated land in the strictly controlled areas is obtained.

[0097] The pollution conditions of the cultivated land soil and crops in the strictly controlled areas are relatively serious, and it is usually impossible to continue growing edible crops. The land use type can be changed, the planting structure can be adjusted, or ecological conversion of farmland can be carried out according to the specific conditions of the polluted cultivated land. If the land use type of the polluted cultivated land is changed to construction land, land compensation fees, resettlement subsidies, compensation fees for young crops and ground attachments shall be compensated to individual farmers or rural village-level collective economic organizations that originally owned the cultivated land, and the new land users need to pay the fees for the use of newly increased construction land. The land compensation fees and resettlement subsidies are affected by location and production capacity; the compensation fees for young crops and ground attachments are mainly affected by production capacity; the fees for the use of newly increased construction land are mainly affected by location. If the planting structure of the polluted cultivated land is adjusted to grow non-edible cash crops such as mulberry, flowers, and nursery stocks, crop replanting subsidies and field infrastructure management and protection fees shall be compensated to individual farmers or rural village-level collective economic organizations that originally owned the cultivated land. The crop replanting subsidies are affected by location and production capacity; the field infrastructure management and protection fees are affected by the undulation of the terrain and production capacity. If ecological conversion of farmland is carried out on the polluted cultivated land, converting farmland to forest or grassland, construction and maintenance fees for converting farmland to forest or grassland (including construction costs and operating costs), garden planting and technical and economic inputs (including garden construction costs, land rent, seedling costs, labor costs, fertilizers, pesticides, irrigation equipment, plastic films and other material costs) shall be compensated to individual farmers or rural village-level collective economic organizations that originally owned the cultivated land. The construction and maintenance fees for converting farmland to forest or grassland are affected by location and the undulation of the terrain; the garden construction costs are mainly affected by the degree of undulation of the terrain; the land rent and labor costs are mainly affected by location; the material costs such as fertilizers, pesticides, irrigation equipment, and plastic films are mainly affected by production capacity.

[0098] All three types of control measures will reduce the planting area of ​​grain crops and cash crops, thereby reducing grain production capacity. In accordance with the principle of "returning as much as possible, replenishing as much as possible" to ensure the production capacity and safety of cultivated land, the quality of unused, inefficiently used, idle, damaged, degraded non-cultivated land, inefficiently used, and residual forest land, with a focus on medium and low-yield fields, can be improved in areas where polluted cultivated land is located. The cost of quality improvement will be affected by the location of the cultivated area, terrain undulations and production capacity.

[0099] According to the influencing factors of measures in the strict control area, 16 scenarios are set to analyze the different scenarios of polluted cultivated land in the strict control area under the constraints of various influencing factors and the possible control measures, as shown in the table below. The polluted cultivated land area corresponding to each scenario is used as the controllable area variable for various measures in the constraint equation, as shown in the table below. Combined with the cost of taking control measures, the constraints of cultivated land area, safe utilization rate and production capacity are further analyzed, and 48 decision variables are selected, including all measures that may be taken for all polluted cultivated land in the strict control area, as shown in Table 3.

[0100] Table 3 Control measures for different scenarios of polluted cultivated land in strict control areas under the constraints of various influencing factors

[0101]

[0102] Since excessive use of certain control measures may lead to the loss of other types of farmland control in the strictly controlled area, it is necessary to weigh different control measures when taking control measures for polluted farmland, such as the balance between adjusting the planting structure and ecological land withdrawal. The costs of changing the land use type, adjusting the planting structure, and ecological land withdrawal in the strictly controlled area are set as the objective functions. F 1 ( X ), F 2 ( X )and F 3 ( X ), then:

[0103] MinF 1 ( X )=∑ C 1i × X a ;

[0104] MinF 2 ( X )=∑ C 2i × X b ;

[0105] MinF3 ( X ) = ∑ C 3i × X c ;

[0106] Where: C 1i is the cost per unit area of taking the control measure of changing land use type in the i th scenario, X a is the polluted cultivated land area of taking the control measure of changing land use type in the i th scenario, where, a = 3 i - 2; C 2i is the cost per unit area of taking the control measure of adjusting planting structure in the i th scenario, X b is the polluted cultivated land area of taking the measure of adjusting planting structure in the i th scenario, where, b = 3 i - 1; C 3i is the cost per unit area of taking the control measure of ecological conversion of cropland to forest or grassland in the i th scenario, X c is the polluted cultivated land area of taking the control measure of ecological conversion of cropland to forest or grassland in the i th scenario, where, c = 3 i .

[0107] The weights of the three objective functions are summed up to 1 by linear weighting. Since the probabilities of taking these three types of control measures for polluted cultivated land are the same, the weight coefficients are all 1 / 3. Accordingly, the multi-objective function can be converted into a single-objective function for solution, then:

[0108] MinF ( X ) = MinF 1 ( X ) + MinF 2 ( X ) + MinF 3 ( X ).

[0109] Based on the constraints of cultivated land area, safety utilization rate, and cultivated land productivity in the first embodiment, combined with the suitability constraints of urban distance, industrial zone distance, slope, and healthy productivity status when formulating various cultivated land control measures, 18 constraint equations are set, as shown in Table 4.

[0110] Table 4 Statistical Table of Constraint Equations

[0111]

[0112] The total controlled area of cultivated land in the strictly controlled area is 952.77 hm 2 , and its heavy metal pollution and planting structure are relatively complex. In this embodiment, according to information such as the distance from the town, the distance from the industrial zone, the slope, and the healthy productivity status, a multi-objective linear programming model is used to classify and control the cultivated land in the strictly controlled area.

[0113] When using the Excel Solver function to obtain the decision variable values at the optimal solution of the objective function, as shown in Table 5. For the polluted cultivated land in Scenarios 2, 4, 6, 7, and 8, the control measure of changing the land use type is adopted. For the polluted cultivated land in Scenarios 1, 3, 9, 11, 13, and 15, the control measure of adjusting the planting structure is adopted. For the polluted cultivated land in Scenarios 5, 10, 12, 14, and 16, the control measure of ecological conversion of cropland to forest or grassland is adopted. Combining the conditions of the distance from the town and the industrial zone, the terrain, and the healthy productivity conditions in each scenario, it can be obtained that: the polluted cultivated land around the town and the industrial zone is classified as the type of changing the land use type, and referring to the approval process for the conversion of agricultural land to construction land, it can apply for approval to change to construction land to support the local construction land demand; the polluted cultivated land located in the grain production functional area, the important agricultural product production protection area, the plain area, and with better healthy productivity conditions is classified as the type of adjusting the planting structure, and the planting structure is adjusted according to local conditions, and non-edible cash crops such as fiber crops (such as cotton, hemp, mulberry), flower seedlings, and energy crops (such as Lindera glauca, Panicum virgatum) are planted to achieve efficient utilization; the polluted cultivated land located in the low hill and gentle slope area and with poor healthy productivity conditions is classified as the type of ecological conversion of cropland to forest or grassland, and after applying for approval according to regulations, it implements the conversion of cropland to forest or grassland and changes to ecological land to promote ecological environmental protection.

[0114] Table 5 Decision Variable Values at the Optimal Solution of the Objective Function

[0115]

[0116] According to the control measures taken for the polluted cultivated land in each scenario in the strictly controlled area, the cultivated land areas of the same control measure type are respectively summed up, as shown in Table 6. For the polluted cultivated land in the strictly controlled area, the area that needs to adopt the control measure of adjusting the planting structure is the largest, up to 517.37 hm 2, accounting for 54.30% of the cultivated land area in the strictly controlled area, exceeding half, and is scattered throughout the study area, especially in the central region. The area of polluted cultivated land that requires control measures to change the land use type is the smallest, which is 168.72 hm 2 , accounting for 17.71%, is concentrated in the urban area in the central part of the study area. The area of polluted cultivated land that requires ecological conversion measures is 266.67 hm 2 , accounting for 27.99% of the cultivated land area in the strictly controlled area, is concentrated in the higher terrain area in the west of the central part of the study area.

[0117] Table 6 Areas and Proportions of Each Control Measure Type

[0118]

[0119] The heavy metal pollution zoning also includes priority protection areas and safe utilization areas. Among them, the cultivated land in the priority protection areas is generally cleaner and healthier. The crops in the area are all in a pollution-free state, and most of the soil is in a pollution-free state, with a small amount of slight pollution. For the cultivated land that is pollution-free for both soil and crops, it is directly included in the permanent basic farmland for continuous protection, such as adopting green production technologies to further improve the productivity and health of the cultivated land. For the cultivated land with slight soil pollution and pollution-free crops, appropriate agronomic control measures such as soil testing and formulated fertilization are taken to reduce the heavy metal pollution degree of the soil, and at the same time, technologies such as green prevention and control and cultivated land conservation are assisted to improve the cultivated land quality; the heavy metal pollution and planting structure in the safe utilization areas are complex. According to information such as crop pollution levels, soil pollution levels, soil pH, types of planted crops, and main soil pollutants, the cultivated land in the safe utilization areas can be classified and controlled. Specific control measures are shown below.

[0120] In the method for classified control of cultivated land heavy metal pollution in the first embodiment, steps S102 and S103 can also adopt the following technical solution: using a decision tree model to obtain the classified control measures for each plot in the strictly controlled area of heavy metal pollution.

[0121] A decision tree model is a method of classifying data based on knowledge to form a tree structure, which is highly similar to a flowchart in structure and has advantages such as strong readability and fast classification speed. The top node of the decision tree is the root node of the tree, the branches represent test outputs, the leaf nodes represent categories, and the non-leaf nodes represent attribute tests. A series of classification rules are generated through learning and induction. Therefore, each path from the root node to the leaf node is a classification rule.

[0122] The third embodiment of the present invention provides a method for obtaining the classified control measures for each plot in the safe utilization area of heavy metal pollution by using a decision tree.

[0123] Use the decision tree model to determine the classification control measures, their areas and regions for the cultivated land in the safe utilization area. There are significant differences in crop pollution, soil pollution, soil pH, crop types, and main soil pollutants of the polluted cultivated land in the safe utilization area. The corresponding control measures are also quite different. The cultivated land in the safe utilization area can adopt the following measures according to the specific pollution situation. If there is heavy metal pollution in the crops, the currently planted varieties can be replaced with low-accumulation varieties of the crops. If there is heavy metal pollution in the soil, measures can be taken according to the cultivated land type and its main pollutant type. For paddy fields polluted by Cd, Hg, Pb, and As, the water regulation method can be adopted; for paddy fields polluted by Cd and Pb, the foliar control method can be adopted; for paddy fields polluted by Cd and Hg, the organic material regulation method can be adopted; for farmlands polluted by Cd, Hg, and Pb, the in-situ passivation method can be adopted; for farmlands polluted by Cd, Zn, and As, the phytoextraction method can be adopted. For the acid-base problem of the soil, the lime or gypsum regulation method can be adopted. The optimized fertilization method is usually used in combination with other measures and is applicable to all soils and crops polluted by heavy metals. In addition, measures such as deep plowing, microbial remediation, and straw removal from the field can be used in combination according to the actual situation of soil heavy metal pollution. Some classification rules of the decision tree are shown in Table 7.

[0124] Table 7 Some classification rules of the decision tree

[0125]

[0126] The total control area of the cultivated land in the safe utilization area is 4650.28 hm 2 , and a total of seven control technologies based on the idea of the "VIP + n" integrated technology are adopted, as shown in Table 8. V (variety) refers to the replacement of low-accumulation varieties; I (irrigation) refers to the optimization of water and fertilizer regulation; P (pH) refers to the adjustment of soil acidity and alkalinity; n refers to other composite measures such as foliar control. The "VIP" technology includes two specific measures, and the area of cultivated land pollution control using this technology is the largest, which is 2218.56 hm 2 , with a proportion as high as 47.70%, almost occupying half of the cultivated land area in the safe utilization area, and is significantly concentrated in the central region. The control area of the "lime / gypsum regulation + optimized fertilization" measure is 1638.26 hm 2 , with a proportion of 35.23%, which is the highest among the control areas of specific measures and is also widely distributed in the study area, scattered in the northern region. There are 92.14 hm 2 of neutral soil that requires the replacement of low-accumulation varieties and the optimization of water and fertilizer regulation, and another 7.27 hm 2 of neutral soil that requires water regulation, foliar control, and organic material regulation. The control areas of the "IP + n" technology and the "P + n" technology are not much different, about 250 hm 2, mainly distributed in the northern part of the central region. The controlled area using only other composite measures is relatively small, among which 127.78 hm 2 of arable land only needs optimized fertilization, and 34.57 hm 2 of arable land only needs to plant hyperaccumulator plants, and 3.90 hm 2 of arable land needs to carry out optimized fertilization and in-situ passivation.

[0127] Table 8 Control Technologies

[0128]

[0129] The method of the classified control measures for the priority protection area can refer to the above technical solution and will not be elaborated here.

[0130] The fourth embodiment of the present invention provides a classified control system for heavy metal pollution in arable land. The classified control system for heavy metal pollution in arable land includes:

[0131] A plot division module, configured to obtain remote sensing image data of the strict control area for heavy metal pollution, and divide the strict control area into multiple plots with different environmental conditions by using the object-oriented method;

[0132] A plot analysis module, configured to obtain the crop types and rotation patterns on each plot in the strict control area;

[0133] A constraint condition construction module, configured to use the distance to the town, the distance to the industrial area, the slope, and the production capacity as the influencing factors of the control measures in the strict control area, construct multiple scenarios, use the controllable area of implementing various control measures in each scenario as the area variable in the constraint equation, and construct multiple suitability constraint conditions for the control measures in the strict control area according to the scenarios, control measures, and area variables;

[0134] A classified control module, configured to use the expenditure cost, arable land area, and safety utilization rate of various control measures in the strict control area as the constraint conditions of the control measures in the strict control area, and combine with the suitability constraint conditions to obtain the classified control measures with the minimum total cost in the strict control area for heavy metal pollution.

[0135] The fifth embodiment of the present invention provides a computer-readable storage medium, on which a data processing program is stored. When the data processing program is executed by a processor, it implements the steps of the classified control method for heavy metal pollution in arable land as described in any one of the above.

[0136] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for classified control of heavy metal pollution in cultivated land, characterized in that: include: Obtain remote sensing image data of the heavy metal pollution strict control area, divide the heavy metal pollution strict control area using object-oriented method, and obtain each plot of heavy metal pollution zone; Obtain the crop types and rotation patterns on each plot in the strictly controlled area, and determine the control measures for various strictly controlled areas, including changing land use types, adjusting planting structure and ecological land withdrawal; Taking the distance from town, distance from industrial area, slope and production capacity as the influencing factors of various control measures in strict control areas to construct multiple scenarios, the controllable area for implementing various control measures in each scenario is taken as the area variable. Based on the scenarios, control measures and area variables, the controllable area in each scenario of the control measures in the strict control area is constructed, and the controllable area in each scenario is taken as the suitability constraint condition; A multi-objective linear programming model was constructed, and the expenditure costs of various control measures in strict control areas were taken as the objective function of the multi-objective linear programming model. The cultivated land area and safe utilization rate were taken as constraints for the control measures in strict control areas. Combined with multiple suitability constraints, the multi-objective linear programming model was used to obtain classified control measures that minimized the total cost of strict control areas of heavy metal pollution under various scenarios.

2. The method for classified control of heavy metal pollution in cultivated land according to claim 1, characterized in that: The method of dividing the heavy metal pollution strict control area by using the object-oriented method includes the following steps: Perform multi-scale segmentation on remote sensing image data and generate image object layers at different segmentation scales, wherein the characteristics of the image objects after multi-scale segmentation include: spectrum, texture and geometry; Select the feature space and assign different weights to each feature to obtain the classification results of the land objects; According to the classification results of the ground objects, the hierarchical structure corresponding to the ground objects is obtained, and the image object layers are connected according to the hierarchical structure corresponding to the ground objects to obtain the various plots of heavy metal pollution zones.

3. The method for classified control of heavy metal pollution in cultivated land according to claim 1, characterized in that: The multi-objective linear programming model is constructed, and the expenditure of various control measures in the strict control area is used as the objective function of the multi-objective linear programming model, including: Construct a multi-objective linear programming model: The expenditures of various control measures in strictly controlled areas are taken as the objective functions of the multi-objective linear programming model. The expenditures of control measures in strictly controlled areas include the government expenditures when changing land use types, adjusting planting structures and ecological land withdrawal. They are set as objective functions F1(X), F2(X) and F3(X) respectively, then: MinF1(X)=∑C 1i ×X a ; MinF2(X)=∑C 2i ×X b ; MinF3(X)=∑C 3i ×X c ; Where: C 1i is the cost per unit area of ​​the land use type change control measures in the i-th scenario, X a is the area of ​​polluted cultivated land that takes land use type change control measures under the i-th scenario, where a=3i-2; C 2i is the cost per unit area of ​​the measures to adjust the planting structure under the ith scenario, X b is the area of ​​polluted cultivated land that takes measures to adjust the planting structure under the i-th scenario, where b = 3i-1; C 3i is the cost per unit area of ​​ecological land reclamation control measures under the i-th scenario, X c is the area of ​​polluted farmland that adopts ecological land withdrawal control measures under the i-th scenario, where c = 3i; The weight coefficients of the three objective functions are all 1 / 3, converting the multi-objective function into a single objective function: Construct the cultivated land area constraint condition, which is that the sum of cultivated land areas with various control measures is equal to the total cultivated land area in the strict control area; Construct a safe utilization rate constraint condition, which requires that the safe utilization rate of polluted cultivated land reaches or exceeds the preset standard; According to the constraints and suitability constraints, the classified control measures with the minimum total cost are obtained.

4. The method for classified control of heavy metal pollution in cultivated land according to claim 1, characterized in that: Also includes: Using the decision tree model, the expenditure costs of various control measures in strict control areas are taken as the objective function of the decision tree model, and the cultivated land area and safe utilization rate are taken as constraints for the control measures in strict control areas. Combined with multiple suitability constraints, a multi-objective linear programming model is used to obtain classified control measures that minimize the total cost of strict control areas of heavy metal pollution under various scenarios.

5. The method for classified control of heavy metal pollution in cultivated land according to claim 1, characterized in that: The acquisition of crop types and rotation patterns on each plot in the strictly controlled area includes: Collect multiple sample point data of remote sensing image data, and annotate the ground object type and crop rotation mode for each sample point data; Obtain spectral characteristic data of each sample point data, and calculate exponential characteristic data according to the spectral characteristic data; The extracted feature data and the labeled data of the sample points are used to train a random forest classification model. The trained random forest classification model is used to obtain the crop types and rotation patterns on each plot in the safe utilization area, and thus determine the various strict control measures in the control area to change the land use type and adjust the planting structure.

6. The method for classified control of heavy metal pollution in cultivated land according to claim 5, characterized in that: The bands of the spectral characteristic data include: B, G, R, NIR, SWIR1 and SWIR2, and the index characteristic data include: NDVI, EVI, NDWI, LSWI, NDSI, NDSVI, NDTI, WDVI and NDFI9.

7. A classification and control system for heavy metal pollution in cultivated land, characterized in that: Used to implement the method for classified control of heavy metal pollution in cultivated land as described in any one of claims 1 to 6, the classified control system for heavy metal pollution in cultivated land comprises: The land parcel division module is used to obtain remote sensing image data of the heavy metal pollution strict control area, divide the heavy metal pollution strict control area using an object-oriented method, and obtain the various land parcels of the heavy metal pollution zone; The plot analysis module is used to obtain the crop types and rotation patterns on each plot in the strictly controlled area, and determine the control measures for various strictly controlled areas, including changing the land use type, adjusting the planting structure and ecological land withdrawal; Construct a constraint condition module, which is used to construct multiple scenarios by taking town distance, industrial zone distance, slope and production capacity as influencing factors of various control measures in strict control areas, taking the controllable area for implementing various control measures in each scenario as the area variable, constructing the controllable area in each scenario of control measures in strict control areas based on the scenario, control measures and area variables, and taking the controllable area in each scenario as the suitability constraint condition; The classification control module is used to take the expenditure of various control measures in strict control areas as the objective function of the multi-objective linear programming model, and take the cultivated land area and safe utilization rate as the constraints of the control measures in the strict control areas. Combined with multiple suitability constraints, the multi-objective linear programming model is used to obtain the classification control measures that minimize the total cost of the strict control areas of heavy metal pollution under various scenarios.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a data processing program, and when the data processing program is executed by the processor, the steps of the method for classification and control of heavy metal pollution in cultivated land as described in any one of claims 1 to 6 are implemented.

Images