A method and system for dividing a soil pollution area

Abstract

This invention discloses a method and system for delineating soil contaminated areas. The method includes: generating a first pollution feature layer by fusing historical pollution data and remote sensing spectral features, and generating a second pollution concentration layer based on an anisotropic pollution diffusion model; analyzing whether pixel spectral features indicate pollution using a trained discriminant model, and marking pixels as core, suspicious, or clean pixels based on concentration exceeding the standard; determining the diffusion correlation between suspicious areas and adjacent core areas by calculating boundary migration feature similarity; merging highly correlated areas according to pollution diffusion fusion rules based on the correlation results to form pollution diffusion areas; and finally optimizing, aggregating, and classifying all contaminated areas based on preset rules to generate a final soil contaminated area delineation map with risk levels. This invention achieves automated and high-precision identification and delineation of soil contaminated areas.

Description

Technical Field

[0001] This invention belongs to the field of soil pollution monitoring technology, and in particular relates to a method and system for delineating soil pollution areas. Background Technology

[0002] Accurate and efficient delineation of soil contamination areas is a prerequisite for conducting pollution risk assessments, developing remediation strategies, and carrying out land planning and management. Traditional soil pollution surveys mainly rely on dense sampling sites, on-site sampling, and laboratory analysis. This method is costly, time-consuming, and has limited spatial representativeness, making it difficult to achieve rapid and accurate delineation over a large area.

[0003] In recent years, remote sensing technology, especially hyperspectral remote sensing, has made it possible to monitor soil pollution non-contactly and over a large scale. However, existing technical solutions often have the following shortcomings: 1) They usually rely solely on remote sensing spectral information, ignoring the value of multi-source information such as historical pollution records, geological and hydrological conditions, and discrete point measured data, resulting in a single basis for classification; 2) They do not adequately consider the spatial correlation between pollution sources and sinks, or the migration and diffusion patterns of pollutants in the environmental medium, leading to a discrepancy between the classification boundaries and the actual spatial distribution continuity of pollutants; 3) They lack effective and refined processing mechanisms for areas where remote sensing information and measured data are inconsistent (i.e., uncertain areas), often resorting to simple classification or ignoring them, affecting the overall classification accuracy; 4) The classification process is not highly automated or intelligent, making it difficult to adapt to the needs of large-scale, dynamically updated applications.

[0004] Therefore, there is an urgent need for a soil pollution zoning method that can deeply integrate multi-source heterogeneous data, fully consider the spatial migration characteristics of pollutants, and handle uncertain regions, so as to improve the scientificity, accuracy and practicality of the zoning results. Summary of the Invention

[0005] This invention aims to overcome the shortcomings of existing technologies and provide a method and system for delineating soil pollution areas. This method integrates spectral remote sensing, historical pollution records and field monitoring data, and introduces a pollutant diffusion model based on physical mechanisms and intelligent discrimination rules to achieve automated and high-precision delineation of soil pollution areas.

[0006] In a first aspect, the present invention provides a method for delineating soil contaminated areas, comprising:

[0007] According to the preset data processing rules, the spectral remote sensing data of pollution sources, historical pollution site data, and soil pollutant concentration data of preset monitoring points in the pollution area to be divided are processed to obtain the first pollution feature layer and the second pollution concentration layer.

[0008] By using a preset pollutant spectral response discrimination model, it is determined whether the spectral absorption characteristics of at least one pixel in the first pollution feature layer indicate pollution, and whether the pollutant concentration at the position corresponding to the at least one pixel in the second pollution concentration layer exceeds a preset pollution threshold.

[0009] Based on the judgment result, at least one pixel is marked as a pollution core pixel, a pollution suspected pixel, or a clean pixel, and at least two adjacent pollution core pixels are spatially clustered to obtain at least one pollution core region, and at least two adjacent pollution suspected pixels are spatially clustered to obtain at least one pollution suspected region.

[0010] Determine whether the similarity of the first boundary migration features between the first suspected pollution area and the adjacent first pollution core area is greater than a preset suspected diffusion threshold, wherein the first suspected pollution area is any pollution core area among at least one pollution core area.

[0011] If the pollution level is greater than the suspected diffusion threshold, the first suspected pollution area and the first pollution core area are merged based on the preset pollution diffusion fusion rules to obtain the pollution diffusion area.

[0012] Based on preset pollution risk patch merging rules, at least one pollution core area and at least one pollution diffusion area are aggregated to generate a soil pollution area delineation map of the pollution area to be delineated.

[0013] Secondly, the present invention provides a soil pollution area delineation system, comprising:

[0014] The processing module is configured to process the spectral remote sensing data of pollution sources, historical polluted land data, and soil pollutant concentration data of preset monitoring points in the polluted area to be divided according to preset data processing rules, so as to obtain a first pollution feature layer and a second pollution concentration layer.

[0015] The first judgment module is configured to determine, through a preset pollutant spectral response discrimination model, whether the spectral absorption characteristics of at least one pixel in the first pollution feature layer indicate pollution, and whether the pollutant concentration at the position corresponding to the at least one pixel in the second pollution concentration layer exceeds a preset pollution threshold.

[0016] The clustering module is configured to mark at least one pixel as a polluted core pixel, a suspected polluted pixel, or a clean pixel based on the judgment result, and to perform spatial clustering on at least two adjacent polluted core pixels to obtain at least one polluted core region, and to perform spatial clustering on at least two adjacent suspected polluted pixels to obtain at least one suspected polluted region.

[0017] The second judgment module is configured to judge whether the similarity of the first boundary migration feature between the first suspected pollution area and the adjacent first pollution core area is greater than a preset suspected diffusion threshold, wherein the first suspected pollution area is any pollution core area among at least one pollution core area.

[0018] The fusion module is configured to, if the pollution level is greater than the suspected diffusion threshold, merge the first suspected pollution area with the first core pollution area based on a preset pollution diffusion fusion rule to obtain the pollution diffusion area.

[0019] The generation module is configured to aggregate at least one pollution core area and at least one pollution diffusion area based on a preset pollution risk patch merging rule, and generate a soil pollution area delineation map of the pollution area to be delineated.

[0020] Thirdly, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the soil contaminated area delineation method of any embodiment of the present invention.

[0021] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program is executed by a processor, the processor performs the steps of the soil contaminated area delineation method of any embodiment of the present invention.

[0022] The method and system for delineating soil pollution areas in this application first establishes a modeling foundation based on deep fusion of multi-source data and physical mechanisms. By nonlinearly fusing historical pollution impact fields with hyperspectral features, an enhanced first pollution feature layer is generated, effectively improving the detection capability of weak pollution signals. Simultaneously, an anisotropic dominant diffusion direction field is constructed based on environmental elements, driving spatial interpolation to generate a second concentration layer that conforms to the migration patterns of pollutants, laying a reliable data foundation for accurate analysis. In the core analysis stage, the method innovatively introduces a regional correlation quantitative assessment model based on physical mechanisms. By comprehensively calculating directional consistency, diffusion probability, and boundary feature similarity, the diffusion correlation between suspected pollution areas and known core areas is scientifically determined, achieving a leap from empirical speculation to quantitative calculation. Furthermore, a hierarchical intelligent fusion and aggregation strategy is adopted. Based on diffusion model verification and anisotropic morphological operations, the reasonable fusion of associated areas is achieved. An iterative optimization algorithm is used to globally aggregate all pollution areas, ultimately forming spatially continuous and internally homogeneous pollution patches. The entire process is completed automatically based on quantitative rules, and can calculate the average concentration and risk level for each patch. Finally, it outputs a decision support thematic map that integrates spatial range, pollution intensity and risk information. It effectively overcomes the shortcomings of traditional methods, such as reliance on manual labor, limited accuracy, low efficiency and insufficient practicality of results, and provides an efficient and reliable technical tool for precise soil environmental management. Attached Figure Description

[0023] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 A flowchart illustrating a method for delineating soil contaminated areas according to an embodiment of the present invention;

[0025] Figure 2 This is a structural block diagram of a soil pollution area delineation system provided in an embodiment of the present invention;

[0026] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Please see Figure 1 The diagram shows a flowchart of a method for delineating soil contamination areas according to this application.

[0029] like Figure 1 As shown, the method for delineating soil pollution areas specifically includes the following steps:

[0030] Step S101: According to the preset data processing rules, the spectral remote sensing data of pollution sources, historical pollution site data, and soil pollutant concentration data of preset monitoring points in the pollution area to be divided are processed to obtain the first pollution feature layer and the second pollution concentration layer.

[0031] In this step, we acquire spectral remote sensing data of pollution sources, historical polluted site data, and soil pollutant concentration data at preset monitoring points for the area to be delineated as a polluted area.

[0032] Spatially overlay and weightedly fuse the spectral absorption features of at least one pixel in the historical polluted site data and the pollution source spectral remote sensing data to generate a first pollution feature layer.

[0033] The soil pollutant concentration data is interpolated into a second pollution concentration layer with the same spatial resolution as the first pollution feature layer, based on a preset anisotropic pollution diffusion model.

[0034] Specifically, the process of spatially overlaying and weighted fusion of the spectral absorption features of at least one pixel in the historical polluted site data and the pollution source spectral remote sensing data to generate a first pollution feature layer includes:

[0035] Obtain the geographic center coordinates, pollution intensity level, and historical activity period of a specific site from the historical polluted site data;

[0036] Based on a certain pollution intensity level and a certain number of historical activity years, a certain persistent impact weight factor for a certain land parcel is calculated, and a certain dynamic impact field centered on a certain geographic center coordinate is constructed according to the certain persistent impact weight factor. That is, after traversing all land parcels, all dynamic impact fields of all land parcels are obtained.

[0037] The spectral remote sensing data of the pollution source is preprocessed to extract the spectral absorption depth of each pixel in the preset pollutant characteristic band.

[0038] The superposition value of all dynamic influence fields at a certain pixel location is calculated as a comprehensive historical pollution impact index for that pixel.

[0039] The comprehensive historical pollution impact index and the spectral absorption depth of a pixel are input into a preset nonlinear response function to calculate the enhanced spectral characteristic value of the pixel. The expression for the nonlinear response function is as follows:

[0040] ,

[0041] In the formula, For pixels Enhanced spectral characteristic values, For pixels At the spectral absorption depth of the preset pollutant characteristic band, For the maximum enhancement coefficient, To control the curvature parameters for enhancing speed, For pixels The comprehensive historical pollution impact index, It is a natural exponential function;

[0042] Traverse all pixels, normalize the calculated enhanced spectral feature values, and arrange them according to the spatial coordinates of each pixel in the pollution source spectral remote sensing data to generate the first pollution feature layer.

[0043] In a specific embodiment, the specific implementation process for generating the first pollution feature layer is as follows:

[0044] First, the system acquires multi-source basic data of the polluted area to be delineated.

[0045] The core of generating the first pollution feature layer lies in organically integrating prior spatial knowledge of historical pollution into the interpretation of current remote sensing spectral information. Specifically, for each plot recorded in the historical pollution plot archive, the system extracts the geographic center coordinates of the j-th plot. Archived original values ​​of pollution intensity levels (For example, classifying pollution incidents into levels 1-5 based on their scale or the multiple of exceedance) and the year in which the pollution incident occurred. .

[0046] Calculate the persistence impact weighting factor for each land parcel. This persistent impact weighting factor comprehensively reflects the potential impact intensity of the land parcel at its current location. Its calculation follows physical laws and consists of two parts: a static baseline and dynamic attenuation. ,in: Based on pollution intensity level The resulting static baseline weights. A pre-defined mapping relationship is established within the system (e.g., ...). (or a lookup table), where a is the proportionality coefficient, and a>0 represents the static baseline weight corresponding to each unit increase in pollution intensity level. The amount to be added is intended to linearly map the discrete original pollution intensity levels to a continuous weighted range, where b is the intercept, representing the weighted value when the original pollution intensity level... When the value is zero (or the lowest level), the basic static weight assigned is b. b can be zero or a small positive value, which is used to set the baseline for weight calculation.

[0047] It is a dynamic factor that decays exponentially over time. For the current year, Let e ​​be the decay coefficient determined based on the average half-life or degradation rate of the target pollutant in the environment, and let e be a natural constant. This dynamic factor simulates the natural attenuation of pollution impacts over time.

[0048] Furthermore, a dynamic influence field is constructed for each plot of land. This is based on the persistent influence weighting factor calculated above. The maximum influence intensity of the plot at its center point is determined by the geographical center of the plot. Taking the origin as the origin, its spatial influence intensity follows a law of attenuation with distance. For any pixel within the region... The impact value of the land parcel on it The calculation is as follows:

[0049] ,

[0050] In the formula, d j(x,y) For pixels The Euclidean distance to the center of the j-th plot, where R is the preset global influence radius used to control the rate of spatial decay.

[0051] Subsequently, the comprehensive historical pollution impact index of the pixel is calculated. For each pixel, the impact of all historically polluted sites on it is summed to obtain its comprehensive historical pollution impact index. The comprehensive historical pollution impact index quantitatively characterizes the background value of potential pollution risk accumulated at the pixel location due to all historical pollution events.

[0052] Meanwhile, the acquired spectral remote sensing data of pollution sources undergoes rigorous preprocessing, including radiometric calibration and atmospheric correction, to ensure data quality. Then, for the known characteristic absorption bands of target pollutants (such as specific heavy metals and petroleum hydrocarbons) (for example, cadmium has absorption characteristics around 680 nm), the continuum removal method is used to accurately extract the spectral absorption depth of each pixel in the image in that band. This value directly reflects the absorption intensity of a specific wavelength of light by the pixel material composition, and is a key spectral indicator for identifying pollutants.

[0053] Finally, enhanced spectral feature values ​​are generated through nonlinear fusion. This involves inputting the spectral absorption depth and the comprehensive historical pollution impact index of each pixel into a preset nonlinear response function for fusion, resulting in the enhanced spectral feature value for that pixel. The technical essence lies in the following: when the comprehensive historical pollution impact index of a pixel is low (i.e., weakly affected by historical pollution), the function output value is close to the original spectral absorption depth, and the enhancement effect is not significant; as the comprehensive historical pollution impact index increases, the nonlinear response function enters a rapidly rising region, significantly enhancing the spectral depth nonlinearly; when the comprehensive historical pollution impact index is extremely high, the enhancement amplitude tends to the saturation upper limit μ. This mechanism cleverly simulates the judgment logic that even if the current remote sensing signal is weak in areas with high historical pollution potential, it should still be given high attention and amplified for examination.

[0054] Finally, the above fusion calculation is performed on all pixels of the entire remote sensing image to obtain the enhanced spectral feature value of each pixel. All enhanced spectral feature values ​​are normalized to a uniform numerical range (e.g., [0,1]) and arranged strictly according to the spatial location and coordinate reference of the original image pixels, thus generating the final first pollution feature layer. This layer is a raster data layer, and the value of each pixel comprehensively expresses the result of nonlinear coupling between the "intensity of potential impact from historical pollution" and the "possibility of pollution indicated by the current spectral signal" at that location.

[0055] Furthermore, interpolating the soil pollutant concentration data into a second pollution concentration layer with the same spatial resolution as the first pollution feature layer, based on a preset anisotropic pollution diffusion model, includes:

[0056] Obtain the measured values ​​of soil pollutant concentrations, the three-dimensional geographic coordinates of all preset monitoring points, and the sampling depth within the polluted area to be divided.

[0057] Obtain the digital elevation model of the polluted area to be delineated, the spatial distribution data of groundwater flow direction, and the prevailing wind direction of the area;

[0058] Based on the spatial distribution data of groundwater flow direction and the prevailing wind direction in the region, a dominant pollution diffusion direction is calculated and determined for each planar location within the pollution area to be divided using a vector synthesis method.

[0059] Based on measured values ​​of soil pollutant concentrations and three-dimensional geographic coordinates, and combined with the dominant pollution diffusion direction at each planar location within the pollution area to be delineated, an anisotropic pollution diffusion model is constructed to characterize the spatial distribution correlation of pollutant concentrations.

[0060] The Kriging interpolation algorithm based on the anisotropic pollution diffusion model is used to spatially interpolate the measured values ​​of soil pollutant concentrations to predict the estimated pollutant concentrations at each horizontal position within the pollution area to be delineated at the preset standard assessment depth.

[0061] The estimated pollutant concentrations at all horizontal locations located at the preset standard assessment depth are arranged according to the three-dimensional geographic coordinates of all preset monitoring points to form a continuous two-dimensional concentration distribution surface.

[0062] The two-dimensional concentration distribution surface is resampled to a spatial resolution and geographic coordinate range that is completely consistent with the first pollution feature layer to generate the second pollution concentration layer.

[0063] In another specific embodiment, the specific implementation process for generating the second pollution concentration layer is as follows:

[0064] Obtain the measured values ​​of soil pollutant concentrations at all preset monitoring points. and its three-dimensional geographic coordinates After sampling depth, the system simultaneously acquires the digital elevation model of the polluted area to be delineated, groundwater flow direction grid data, and the prevailing wind direction of the area (usually expressed as an angle) obtained through statistical analysis.

[0065] First, the dominant direction of pollution diffusion field Determination: For each planar grid location (x, y) within the region, the system reads the groundwater flow direction angle at that location. (Derived from hydrogeological models or measurements) and the prevailing wind direction in the region. The dominant diffusion direction is calculated using the vector synthesis method:

[0066] ,

[0067] Here, arctan2(x,y) is a function that takes the x and y coordinates of a point in a Cartesian coordinate system as input and returns the angle between the vector from the origin (0,0) to the point (x,y) and the positive x-axis. and These are the weighting coefficients for the groundwater flow direction angle and the regional prevailing wind direction angle, respectively. They can be preset according to the main driving forces of pollutant migration (such as soluble pollutants being greatly affected by groundwater and volatile organic compounds being greatly affected by wind). This step generates a dominant diffusion direction field that covers the entire region and changes continuously with space.

[0068] Secondly, an anisotropic pollution diffusion model is constructed: the core of the anisotropic pollution diffusion model is an anisotropic spatial variability function, which is used to quantitatively describe the spatial autocorrelation structure of pollutant concentrations. The model construction steps are as follows:

[0069] The experimental variability function was calculated based on data from all monitoring points.

[0070] Define a dominant diffusion direction This is an elliptical variogram model along its major axis. Its mathematical expression (taking the exponential model as an example) is:

[0071] ,

[0072] ,

[0073] ,

[0074] In the formula, The variogram value is a measure of the degree of difference in soil pollutant concentration between two points in space. The nugget value represents the variogram value as the spatial separation distance approaches zero (i.e., repeated measurements at the same point). The arch height represents the variation function from the nugget value. The incremental portion of the increase to a plateau level reflects the concentration variance caused by the macroscopic spatial autocorrelation structure (i.e., the spatial distribution and diffusion patterns of pollutants) that can be described by an elliptic variogram model. For the variable range along the dominant diffusion direction, For the range perpendicular to the dominant diffusion direction, , This is a spatial separation vector, representing the coordinate difference between any two points in the original East-North (or Cartesian) coordinate system. For example, in interpolation, it represents a known monitoring point. With a point to be estimated The positional offset between them, where, , , To be By rotating the coordinate system, the direction of pollution diffusion is shifted to the dominant direction. Local coordinate system based on ( Components after coordinate system The spatial separation vector is perpendicular to the dominant diffusion direction ( Projected length on axis The spatial separation vector is in the dominant diffusion direction ( Projected length on axis To perform the square root operation;

[0075] Then, anisotropic kriging interpolation is performed: for the target location where concentration needs to be estimated. ,in If the preset standard evaluation depth is fixed, such as 0.2 meters, the system will perform the following operations:

[0076] Determine the dominant diffusion direction of the plane containing the target point T. ;

[0077] by Based on this, calculate the anisotropic distances between the target point T and all monitoring points.

[0078] Using the anisotropic variogram model constructed above, the covariance between all point pairs is calculated, and the Kriging equations are constructed.

[0079] Solving the Kriging equations yields the optimal linear unbiased estimate of the pollutant concentration at the target point T. and its estimated variance .

[0080] Next, a two-dimensional concentration distribution surface is generated and resampled: The above interpolation is performed on all horizontal target locations at the standard assessment depth throughout the entire area to obtain a continuous two-dimensional concentration estimation matrix. This matrix is ​​then registered according to geographic coordinates to form the initial two-dimensional concentration distribution surface. Finally, a resampling algorithm such as bilinear or cubic convolution is used to adjust the spatial resolution, pixel size, and geographic extent of this concentration distribution surface to be completely consistent with the first pollution feature layer, generating the final second pollution concentration layer.

[0081] Step S102: Using a preset pollutant spectral response discrimination model, determine whether the spectral absorption characteristics of at least one pixel in the first pollution feature layer indicate pollution, and whether the pollutant concentration at the position corresponding to the at least one pixel in the second pollution concentration layer exceeds a preset pollution threshold.

[0082] In this step, the pre-defined pollutant spectral response discrimination model employs a gradient boosting decision tree. For each monitoring point, a three-dimensional feature vector is constructed as the model input:

[0083] Spectral absorption depth: Extract the spectral absorption depth of the corresponding pixel at the center of the characteristic absorption band of the target pollutant;

[0084] First-order differential spectral value: Calculate the first-order differential spectral value of the pixel in the shoulder or adjacent band of the characteristic absorption band, which is used to enhance the identification of weak absorption and spectral morphology changes.

[0085] Historical impact weighted spectral index: Calculate the spatial attenuation weight of the pixel to all historically polluted areas, and use this weight to weight the reflectance of the pixel in multiple relevant bands to generate a customized spectral index (such as the weighted normalized difference index) that incorporates the spatial prior of historical pollution.

[0086] Model Training and Threshold Setting: The training data consists of the three-dimensional feature vectors of all monitoring points and their corresponding measured soil pollutant concentrations. The model's learning objective is to establish a mapping relationship from these three spectral-derived features to the "probability of exceeding the concentration limit." After training, the model can output a spectral pollution probability value (range 0-1) for any pixel. Simultaneously, based on operational needs (such as controlling the false alarm rate) or through validation set optimization, a first discrimination threshold (e.g., 0.65) is set.

[0087] Secondly, perform full-image meta-level parallel dual-channel discrimination.

[0088] For each pixel in the first pollution feature layer, the system simultaneously executes the following two independent judgment processes:

[0089] Channel 1: Spectral Characteristic Pollution Indication Judgment

[0090] Extract the three-dimensional feature vector corresponding to the pixel and input it into the trained pollutant spectral response discrimination model;

[0091] The model outputs the spectral contamination probability value for that pixel;

[0092] The probability value of spectral contamination is compared with a first discrimination threshold: if the probability value is greater than the threshold, the spectral absorption feature of the pixel is determined to be "indicating contamination"; otherwise, it is determined to be "not indicating contamination".

[0093] Channel Two: Judgment of Concentration Exceeding Standard at Corresponding Location

[0094] In the second pollution concentration layer that has been spatially precisely registered, the estimated pollutant concentration corresponding to the geographic location of the pixel is directly read.

[0095] The estimated concentration of this pollutant is compared with the preset pollution threshold to determine whether it exceeds the standard.

[0096] Finally, a two-state discrimination result is generated:

[0097] Status A (Spectral Evidence): Does the spectral feature indicate contamination (yes / no)?

[0098] Status B (Concentration Evidence): Whether the concentration at the corresponding location exceeds the threshold (yes / no);

[0099] These two states will be directly used for pixel classification and labeling in subsequent steps, accurately defining core pollution pixels, suspected pollution pixels, etc.

[0100] Step S103: Based on the judgment result, mark the at least one pixel as a core pollution pixel, a suspected pollution pixel, or a clean pixel, and spatially cluster at least two adjacent core pollution pixels to obtain at least one core pollution region, and spatially cluster at least two adjacent suspected pollution pixels to obtain at least one suspected pollution region.

[0101] In this step, if the spectral characteristics of a pixel indicate pollution, and the pollutant concentration at the corresponding location of the pixel exceeds a preset pollution threshold, then the pixel is marked as a core pollution pixel; if only the spectral absorption characteristics of a pixel indicate pollution, or only the pollutant concentration at the corresponding location of the pixel exceeds a preset pollution threshold, then the pixel is marked as a suspected pollution pixel; if the spectral characteristics of a pixel do not indicate pollution, and the pollutant concentration at the corresponding location of the pixel does not exceed a preset pollution threshold, then the pixel is marked as a clean pixel.

[0102] In one specific embodiment, spatial clustering analysis is performed on core contamination pixels and suspected contamination pixels respectively:

[0103] Clustering algorithm selection: A clustering algorithm based on eight-neighbor connectivity analysis is adopted, which identifies spatially adjacent (shared edges or corners) cells of the same type as the same connected component;

[0104] Generate contamination core regions: Using all pixels marked as contamination core pixels as input, perform connectivity analysis. Each independent connected component (i.e., a group of contamination core pixels connected to each other) is extracted and defined as an independent contamination core region. The system also records the geometric attributes of each region, such as area, bounding rectangle, centroid coordinates, etc.

[0105] Generate contamination-prone regions: Similarly, using all pixels marked as contamination-prone cells as input, perform the same connectivity analysis, and each independent connected component is defined as a contamination-prone region;

[0106] This clustering method outputs two key sets of spatial objects: at least one core pollution area and at least one suspected pollution area. Each area has the same cell category attribute internally and is spatially continuous.

[0107] Step S104: Determine whether the similarity of the first boundary migration features between the first suspected pollution area and the adjacent first pollution core area is greater than a preset suspected diffusion threshold, wherein the first suspected pollution area is any pollution core area among at least one pollution core area.

[0108] In this step, the center of the first core area of ​​the first pollution core area and the center of the first suspected pollution area are obtained, and the direction of the geometric line connecting the center of the first core area to the center of the first suspected pollution area is defined as the pollution assessment direction.

[0109] Obtain the first dominant pollution diffusion direction at the center of the first core area, and calculate the directional consistency coefficient based on the angle between the pollution assessment direction and the first dominant pollution diffusion direction. ,in, The angle between the direction of pollution assessment and the direction of primary pollution diffusion;

[0110] Taking the center of the first core area as the diffusion source point, based on the anisotropic pollution diffusion model, the diffusion probability value P of the pollutants diffused to the center of the first suspected area along the assessment direction is simulated and calculated.

[0111] Extract the common boundary line between the first suspected pollution area and the first core pollution area;

[0112] Along the common boundary line, a preset vertical buffer zone is extended to the side of the first suspected pollution area and the side of the first pollution core area, respectively, to obtain the first set of pixels in the buffer zone on the side of the first suspected pollution area and the second set of pixels in the buffer zone on the side of the first pollution core area.

[0113] Calculate the mean of the first feature values ​​of all pixels in the first pixel set on the first pollution feature layer, and the mean of the second feature values ​​of all pixels in the second pixel set on the first pollution feature layer, respectively.

[0114] Calculate the boundary feature cosine similarity β between the mean of the first feature value and the mean of the second feature value, and then apply the directional consistency coefficient. The diffusion probability value P and the boundary feature cosine similarity β are used to calculate the first boundary migration feature similarity, expressed as:

[0115] ,

[0116] In the formula, The first boundary migration feature similarity, This is the weighting coefficient for the directional consistency coefficient. These are the weighting coefficients for the diffusion probability values. is the weight coefficient of the cosine similarity of the boundary features, and .

[0117] The calculated similarity of the first boundary migration feature is compared with the preset suspicious diffusion threshold.

[0118] In one specific embodiment, the first step is: geometric and physical orientation consistency assessment.

[0119] Determine the regional representative points and assessment direction: Calculate the geometric centroids of the first pollution core area and the first pollution suspected area respectively, and use them as their respective spatial representative points (i.e., the center of the first core area and the center of the first suspected area). Connect the two points and define the vector direction from the center of the core area to the center of the suspected area as the pollution assessment direction.

[0120] Obtain the dominant diffusion direction: From the dominant pollution diffusion direction field generated in step S101, query the first dominant pollution diffusion direction at the location of the center of the first core area (e.g., the synthesized groundwater flow and wind direction).

[0121] The directional consistency coefficient α is calculated by determining the plane angle θ (ranging from 0° to 180°) between the pollution assessment direction and the primary dominant pollution diffusion direction. The directional consistency coefficient α is defined as α = cos(θ). When the assessment direction is perfectly aligned with the dominant diffusion direction (θ = 0°), α = 1; when they are completely opposite (θ = 180°), α = -1; and when they are perpendicular (θ = 90°), α = 0. The closer the α value is to 1, the greater the likelihood that the suspected area is located downstream / downwind of the pollution core.

[0122] Step 2: Probabilistic assessment based on the physical diffusion model

[0123] Taking the center of the first core area as the pollution source, based on the anisotropic pollution diffusion model constructed in step S101, the migration of pollutants in space is simulated. Specifically, the diffusion path of pollutants along the pollution assessment direction starting from the source is calculated, and the probability value P of reaching the center of the first suspected area is assessed. This probability value P comprehensively reflects the source intensity, the diffusion capacity of the medium, and the distance attenuation effect along the assessment direction, and quantifies the possibility of diffusion from a physical mechanism perspective.

[0124] Step 3: Continuity assessment of pollution characteristics in the boundary area

[0125] Extracting and Buffering Common Boundaries: Through spatial calculations, the common boundary line between the first suspected pollution area and the first core pollution area is extracted. Along this boundary line, vertical buffer zones of a preset fixed width (e.g., 5 meters) are created on both the suspected pollution area side and the core pollution area side.

[0126] Calculate the boundary feature similarity β: Extract all pixels located within the buffer zone of the suspicious region (first pixel set) and all pixels located within the buffer zone of the core region (second pixel set). Calculate the mean vector of the feature values ​​(i.e., F(x,y)) of all pixels in both sets on the first contamination feature layer. and Calculate the cosine similarity between these two mean vectors, and use it as the boundary feature similarity β. The higher the β value, the more similar the pollution spectral features on both sides of the boundary are, implying spatial continuity.

[0127] Step 4: Multi-indicator integration and comprehensive decision-making

[0128] The evaluation results of the above three dimensions are linearly weighted and fused to calculate the first boundary migration feature similarity.

[0129] Threshold Comparison and Decision: The calculated similarity of the first boundary migration features is compared with a preset suspected diffusion threshold (e.g., 0.7). If the similarity of the first boundary migration features is greater than the threshold, the correlation between the first suspected pollution area and the first pollution core area (i.e., the possibility of diffusion from the core area) is considered strong; otherwise, it is considered weak. This determination will directly determine whether to perform a fusion operation on these two areas in subsequent steps.

[0130] Specifically, after determining whether the similarity of the first boundary migration characteristics between the first suspected pollution area and the adjacent first pollution core area is greater than a preset suspected diffusion threshold,

[0131] If the pollution level exceeds the suspected diffusion threshold, the first suspected pollution area will be marked as an independent pollution patch, and soil profile pollution data of preset sampling points inside and at the edge of the independent pollution patch will be obtained.

[0132] Based on the pollution data of each soil profile, a three-dimensional interpolation method based on the vertical transport model of pollutants is used to reconstruct and correct the pollution range and boundary of the independent pollution patches in three dimensions, so as to obtain at least one target pollution area.

[0133] Based on preset pollution risk patch merging rules, at least one pollution core area, at least one pollution diffusion area, and at least one target pollution area are aggregated to generate a soil pollution area delineation map of the pollution area to be delineated.

[0134] In one specific embodiment, for any first suspected pollution area determined to have weak correlation with all adjacent core pollution areas (i.e., the similarity of the first boundary migration characteristics is not greater than a preset suspected diffusion threshold), the system re-marks it as an independent pollution patch. Subsequently, within the patch and its boundary area, a series of supplementary sampling points are planned and deployed based on preset rules (such as equally spaced grids or according to the high and low gradients of the first pollution feature layer values). These sampling points aim to obtain more diagnostically valuable soil profile pollution data, including but not limited to pollutant concentrations, soil texture, bulk density, moisture content, and permeability coefficients at different soil depths.

[0135] After obtaining soil profile data from supplementary sampling points, perform the following operations to accurately correct the contamination extent of this isolated contaminated patch:

[0136] A three-dimensional vertical transport model of pollutants is constructed: This is a simplified numerical model based on physical laws. Its governing equations typically consider the convection, dispersion, adsorption, and degradation of pollutants in the unsaturated-saturated zone. The model uses collected soil stratification attribute data (such as texture and organic matter content of each layer) and regional hydrogeological parameters (such as groundwater level fluctuations and rainfall infiltration) as key inputs.

[0137] Perform three-dimensional conditional interpolation and simulation: using pollutant concentration data at different depths of all sampling points (including the original discrete monitoring points and new supplementary profile points) as "hard data", and the possible transport range of pollutants obtained by the above vertical transport model as "soft constraints", the three-dimensional co-kriging or probabilistic inversion method based on the physical model is used to perform high-resolution interpolation and simulation of the three-dimensional spatial distribution of pollutants in the area where the independent pollution patch is located.

[0138] Extraction and correction of pollution range: From the three-dimensional simulation results, extract all three-dimensional voxels (voxels) whose pollutant concentration exceeds the preset threshold, project these voxels onto the horizontal plane, and comprehensively consider their vertical distribution characteristics (such as whether they reach groundwater), and redraw the two-dimensional pollution range boundary of the independent pollution patch at the surface or target depth. This boundary is the more accurate pollution range after verification by field data and correction by the physical model. The system defines the corrected area as one or more (if the simulation results show that there are multiple independent pollution sources) target pollution areas.

[0139] Based on the preset pollution risk patch merging rules, at least one pollution core area, at least one pollution diffusion area, and at least one target pollution area are aggregated to generate a soil pollution area delineation map of the pollution areas to be delineated.

[0140] Step S105: If the pollution level is greater than the suspected diffusion threshold, the first suspected pollution area and the first pollution core area are merged based on the preset pollution diffusion fusion rule to obtain the pollution diffusion area.

[0141] In this step, the center of the first core area of ​​the first pollution core area is taken as the diffusion source point;

[0142] Based on the anisotropic pollution diffusion model, the steady-state concentration distribution of pollutants at various locations in space is simulated and calculated when the pollutants diffuse from the diffusion source to the first suspected pollution area. The spatial range with concentration values ​​higher than the preset fusion concentration threshold is extracted and defined as the potential diffusion influence area.

[0143] Calculate the spatial intersection area between the potential diffusion impact area and the first suspected pollution area, and determine whether the proportion of the spatial intersection area to the area of ​​the first suspected pollution area is greater than a preset proportion threshold.

[0144] If the ratio is not greater than the preset threshold, the first suspected pollution area will not be merged with the first core pollution area.

[0145] If the potential diffusion impact area is greater than the preset proportional threshold, the potential diffusion impact area and the first pollution suspected area are spatially superimposed using the line connecting the diffusion source point and the center of the first suspected area as the main axis direction to obtain the superimposed area.

[0146] A morphological closing operation is performed on the superimposed region, and the continuous spatial range after the morphological closing operation is defined as the fused pollution diffusion region.

[0147] In one specific embodiment, the geometric center of the first pollution core area (the center of the first core area) is used as the representative point of the theoretical point source or area source. The anisotropic pollution diffusion model constructed in step S101 is invoked, and the source point is set to have a unit release intensity. The model simulates the migration and diffusion process of pollutants under steady-state or long-term average conditions, extending into the entire surrounding space, including the direction of the first suspected pollution area. The model outputs a simulated concentration field covering the relevant area. From this simulated concentration field, all raster pixels with simulated concentration values ​​higher than a preset fusion concentration threshold are extracted. The spatial range formed by these pixels is defined as the potential diffusion impact zone. This threshold is typically set to a low value to identify the range of "impact traces" that pollutants may reach.

[0148] Spatial intersection analysis and fusion necessity judgment:

[0149] Using the spatial overlay analysis function of Geographic Information System (GIS), the spatial intersection between the two polygon vectors of the potential diffusion impact area and the first suspected pollution area is calculated.

[0150] Calculate the area of ​​the intersection of the above parts respectively. and the original area of ​​the first suspected pollution site .

[0151] Calculate the overlap ratio = / The ratio is compared with a preset ratio threshold (e.g., 50% or 70%). If it is not greater than the ratio threshold, it indicates that most of the area of ​​the suspected pollution point does not fall on the main influence path of the simulated diffusion from the core area. Its cause may be relatively independent or complex. At this time, the system terminates the fusion operation and retains the suspected area as an independent object.

[0152] If the ratio is greater than the threshold, it indicates that the main part of the suspicious area is highly overlapping with the diffusion path simulated from the core area in space. The system determines that it meets the fusion conditions and continues to execute subsequent operations.

[0153] Spatial overlay and morphological optimization generate the final region:

[0154] The line connecting the center of the first core area to the center of the first suspected area is determined as the main axis. The potential diffusion impact area and the first suspected pollution area are spatially united to obtain a preliminary superimposed area that may contain internal cavities or jagged boundaries.

[0155] Perform anisotropic morphological closing operation:

[0156] Structural element design: Create an elliptical morphological structural element whose major axis is aligned with the principal axis direction in the anisotropic diffusion model. The length of the major axis can be set to the range along the principal axis direction in the anisotropic diffusion model. It is positively correlated with the length of the minor axis, while the length of the minor axis is positively correlated with the range in the vertical direction. They are positively correlated, thus making the shape of the structural element reflect the anisotropy of diffusion.

[0157] Operation execution: Using a pre-designed elliptical structuring element, a morphological closing operation (i.e., dilation followed by erosion) is performed on the superimposed region. The purpose of this operation is to: a) bridge any small gaps that may exist between the two original regions; b) smooth unnatural bumpy and concave boundaries at the connection points; c) fill small holes inside the region caused by simulation or classification noise, ultimately forming a spatially connected and smooth continuous region.

[0158] Define the pollution diffusion area: The continuous spatial range after the above morphological closing operation is defined as the fused pollution diffusion area.

[0159] Step S106: Based on the preset pollution risk patch merging rules, at least one pollution core area and at least one pollution diffusion area are aggregated to generate a soil pollution area delineation map of the pollution area to be delineated.

[0160] In this step, it is determined whether any two spatially adjacent regions in the at least one pollution core region and the at least one pollution diffusion region meet the preset merging conditions. The merging conditions include: calculating the feature similarity of the pixels on both sides of the common boundary between the two regions on the first pollution feature layer, and the directional compatibility between the direction of the line connecting the centers of the two regions and the dominant pollution diffusion direction at the corresponding location.

[0161] Each pair of adjacent regions that meets the merging conditions is marked as a candidate merging pair;

[0162] From all candidate merge pairs, select the pair with the highest comprehensive score for spatial merging to form a new merged region, and update the region set;

[0163] The above selection and merging operations are performed iteratively until no candidate merge pairs satisfy the merging conditions are found;

[0164] Each region in the final region set is defined as an independent pollution patch in the soil pollution zoning map;

[0165] Based on the average concentration value of each independent pollution patch on the second pollution concentration layer, a pollution risk level is assigned to each independent pollution patch, and a soil pollution area delineation map with the pollution risk level is generated.

[0166] In one specific embodiment, at least one pollution core region generated in step S103 and at least one pollution diffusion region generated in step S105 together constitute an initial set of regions to be merged. For any two spatially adjacent regions in this set (denoted as region A and region B), the system calculates merging condition indicators in two dimensions in parallel:

[0167] Boundary feature similarity β': Extract the common boundary line between region A and region B. Along this boundary line, create vertical buffers of a preset width on both sides. Extract the feature values ​​of all pixels in the buffer on side A on the first pollution feature layer and calculate its mean vector μA; similarly, calculate the mean vector μB of the buffer on side B. Calculate the cosine similarity between these two mean vectors as the boundary feature similarity β'. The higher the value of β', the more continuous the spectral pollution features of the two regions are at the boundary.

[0168] Diffusion direction compatibility α': Calculate the vector direction from the centroid of region A to the centroid of region B. Locate the dominant directions at the centroids of regions A and B respectively from the dominant pollution diffusion direction field. Calculate the angle θ' between the above vector direction and the dominant direction at the centroid of region B (or use the average direction of the two dominant directions as a reference). The direction compatibility coefficient α' is calculated as cosθ'. The closer the value of α' is to 1, the more consistent the direction from A to B is with the local dominant diffusion trend.

[0169] The system presets a boundary similarity threshold (e.g., 0.75) and a directional compatibility threshold (e.g., 0.5). Regions A and B are considered to meet the merging conditions and are marked as a candidate merging pair only when both β' > the boundary similarity threshold and α' > the directional compatibility threshold are satisfied. A comprehensive score is then calculated for this merging pair. , where w1 and w2 are preset weights.

[0170] Selection and Merging: From all candidate merge pairs, select the pair with the highest overall score. Merge these two regions into a new, contiguous region (called the merged region C) using the spatial union operation.

[0171] Update set: Remove the two original regions from the current set of regions to be merged, and add the new merged region C to the set;

[0172] Recalculation: Since a new region C is generated, the system recalculates the merging condition indices (β' and α') between C and all its spatially adjacent regions in the set, and updates the list of candidate merging pairs and their comprehensive scores.

[0173] Termination condition: Repeat the above steps until the candidate merge pair list is empty, that is, there are no adjacent region pairs that meet the merge condition, and the iteration terminates.

[0174] Define an independent contaminated patch: Each region in the final set of regions after iterative merging is formally defined as an independent contaminated patch in the soil contamination zoning map;

[0175] Calculate and assign risk levels for each individual contaminated patch:

[0176] Calculate the average concentration: On the second pollution concentration layer with precise spatial registration, extract the estimated pollutant concentration values ​​of all pixels within the patch area and calculate their arithmetic mean;

[0177] Graded assignment: Based on the preset grade classification rules (for example, the average concentration between the screening value and the control value is defined as "medium risk", and the average concentration exceeding the control value is defined as "high risk"), a pollution risk level (such as "high risk", "medium risk", "low risk" or the corresponding numerical level) is assigned to the patch.

[0178] Output: The system outputs the final soil pollution area delineation map, which is a spatial vector layer. Each patch element includes its geographical boundary, unique number, calculated average concentration, and assigned pollution risk level attribute.

[0179] In summary, the method presented in this application achieves deep coupling and feature enhancement of spectral, historical, and measured data at the multi-source information fusion level, laying a data foundation for high-precision identification. By constructing a dynamically decaying historical pollution influence field and nonlinearly fusing it with the feature absorption depth of hyperspectral images, a first pollution feature layer that simultaneously reflects "current spectral anomalies" and "historical pollution potential" is generated, effectively amplifying weak pollution signals. Simultaneously, an anisotropic dominant diffusion direction field is constructed based on hydrological and meteorological data, and this drives Kriging interpolation to generate a second pollution concentration layer that conforms to the physical laws of pollutant migration. The creation of these two fundamental layers provides a high-quality and spatially aligned information base for subsequent analysis, far exceeding that of a single data source.

[0180] At the pollution analysis and process simulation level, physical mechanisms were introduced, and intelligent discrimination and correlation analysis from pixels to regions were realized, significantly improving scientific rigor and accuracy. A dual-channel discrimination mechanism based on gradient boosting decision trees was adopted to independently and in parallel judge the spectral pollution probability and concentration exceedance status of each pixel, achieving mutual verification of evidence and accurate identification of uncertainties (suspicious pixels). Furthermore, a regional correlation quantitative assessment model integrating three elements—"directional physical consistency, diffusion process probability, and boundary feature continuity"—was proposed. This model combines the physical laws of anisotropic diffusion models with spatial topology and spectral characteristics to scientifically determine whether suspected pollution areas spread from known core areas, enabling the inference of pollution range to move from empirical speculation to quantitative calculation.

[0181] At the spatial integration and decision optimization level, hierarchical, adaptive, and interpretable regional fusion and aggregation rules were constructed to ensure the rationality and practicality of the partitioning results. For areas with strong correlation, spatial intersection verification and anisotropic morphological fusion were performed on the "potential impact zone" simulated by the diffusion model to generate "pollution diffusion areas" that conform to the diffusion morphology. Finally, an iterative merging algorithm considering the similarity of boundary features and the compatibility of diffusion directions was used to globally optimize and aggregate all polluted areas, forming a final patch with internal homogeneity and clear boundaries. The entire process was completed automatically based on quantification rules, eliminating subjective arbitrariness, and a flexible strategy adjustment interface was provided through parameter thresholds.

[0182] Please see Figure 2 The diagram shows a structural block diagram of a soil pollution zone delineation system according to this application.

[0183] like Figure 2 As shown, the soil pollution area classification system 200 includes a processing module 210, a first judgment module 220, a clustering module 230, a second judgment module 240, a fusion module 250, and a generation module 260.

[0184] The processing module 210 is configured to process the spectral remote sensing data of pollution sources, historical polluted site data, and soil pollutant concentration data of preset monitoring points in the polluted area to be divided according to preset data processing rules, to obtain a first pollution feature layer and a second pollution concentration layer; the first judgment module 220 is configured to determine, through a preset pollutant spectral response discrimination model, whether the spectral absorption characteristics of at least one pixel in the first pollution feature layer indicate pollution, and whether the pollutant concentration at the position corresponding to the at least one pixel in the second pollution concentration layer exceeds a preset pollution threshold; the clustering module 230 is configured to mark the at least one pixel as a pollution core pixel, a suspected pollution pixel, or a clean pixel according to the judgment result, and to spatially cluster at least two adjacent pollution core pixels to obtain at least one pollution core. The system comprises: a first pollution suspected area and a second judgment module 240 configured to determine whether the similarity of the first boundary migration feature between the first pollution suspected area and the adjacent first pollution core area is greater than a preset suspected diffusion threshold, wherein the first pollution suspected area is any pollution core area among at least one pollution core area; a fusion module 250 configured to, if greater than the suspected diffusion threshold, fuse the first pollution suspected area and the first pollution core area according to a preset pollution diffusion fusion rule to obtain a pollution diffusion area; and a generation module 260 configured to, based on a preset pollution risk patch merging rule, aggregate at least one pollution core area and at least one pollution diffusion area to generate a soil pollution area delineation map of the pollution area to be delineated.

[0185] It should be understood that Figure 2 The modules and references described in the document Figure 1 The steps described in the text correspond to those in the method described above. Therefore, the operations, features, and corresponding technical effects described above also apply to the method described in the text. Figure 2 The various modules in the document will not be described in detail here.

[0186] In other embodiments, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, causes the processor to perform the soil pollution area delineation method in any of the above method embodiments.

[0187] In one embodiment, the computer-readable storage medium of the present invention stores computer-executable instructions, which are configured as follows:

[0188] According to the preset data processing rules, the spectral remote sensing data of pollution sources, historical pollution site data, and soil pollutant concentration data of preset monitoring points in the pollution area to be divided are processed to obtain the first pollution feature layer and the second pollution concentration layer.

[0189] By using a preset pollutant spectral response discrimination model, it is determined whether the spectral absorption characteristics of at least one pixel in the first pollution feature layer indicate pollution, and whether the pollutant concentration at the position corresponding to the at least one pixel in the second pollution concentration layer exceeds a preset pollution threshold.

[0190] Based on the judgment result, at least one pixel is marked as a pollution core pixel, a pollution suspected pixel, or a clean pixel, and at least two adjacent pollution core pixels are spatially clustered to obtain at least one pollution core region, and at least two adjacent pollution suspected pixels are spatially clustered to obtain at least one pollution suspected region.

[0191] Determine whether the similarity of the first boundary migration features between the first suspected pollution area and the adjacent first pollution core area is greater than a preset suspected diffusion threshold, wherein the first suspected pollution area is any pollution core area among at least one pollution core area.

[0192] If the pollution level is greater than the suspected diffusion threshold, the first suspected pollution area and the first pollution core area are merged based on the preset pollution diffusion fusion rules to obtain the pollution diffusion area.

[0193] Based on preset pollution risk patch merging rules, at least one pollution core area and at least one pollution diffusion area are aggregated to generate a soil pollution area delineation map of the pollution area to be delineated.

[0194] Computer-readable storage media may include a stored program area and a stored data area, wherein the stored program area may store an operating system and an application program required for at least one function; the stored data area may store data created based on the use of the soil contaminated zoning system, etc. Furthermore, the computer-readable storage medium may include high-speed random access memory, and may also include memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, the computer-readable storage medium may optionally include memory remotely configured relative to a processor, which can be connected to the soil contaminated zoning system via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0195] Figure 3 This is a schematic diagram of the structure of the electronic device provided in the embodiment of the present invention, such as... Figure 3 As shown, the device includes a processor 310 and a memory 320. The electronic device may also include an input device 330 and an output device 340. The processor 310, memory 320, input device 330, and output device 340 can be connected via a bus or other means. Figure 3 Taking a bus connection as an example, the memory 320 is the computer-readable storage medium described above. The processor 310 executes various server functions and data processing by running non-volatile software programs, instructions, and modules stored in the memory 320, thereby implementing the soil contaminated area delineation method described in the above embodiment. The input device 330 can receive input digital or character information and generate key signal inputs related to user settings and function control of the soil contaminated area delineation system. The output device 340 may include a display screen or other display device.

[0196] The aforementioned electronic device can execute the method provided in the embodiments of the present invention, and has the corresponding functional modules and beneficial effects for executing the method. Technical details not described in detail in this embodiment can be found in the method provided in the embodiments of the present invention.

[0197] In one implementation, the above-described electronic device is applied to a soil contaminated area delineation system for a client application, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to:

[0198] According to the preset data processing rules, the spectral remote sensing data of pollution sources, historical pollution site data, and soil pollutant concentration data of preset monitoring points in the pollution area to be divided are processed to obtain the first pollution feature layer and the second pollution concentration layer.

[0199] By using a preset pollutant spectral response discrimination model, it is determined whether the spectral absorption characteristics of at least one pixel in the first pollution feature layer indicate pollution, and whether the pollutant concentration at the position corresponding to the at least one pixel in the second pollution concentration layer exceeds a preset pollution threshold.

[0200] Based on the judgment result, at least one pixel is marked as a pollution core pixel, a pollution suspected pixel, or a clean pixel, and at least two adjacent pollution core pixels are spatially clustered to obtain at least one pollution core region, and at least two adjacent pollution suspected pixels are spatially clustered to obtain at least one pollution suspected region.

[0201] Determine whether the similarity of the first boundary migration features between the first suspected pollution area and the adjacent first pollution core area is greater than a preset suspected diffusion threshold, wherein the first suspected pollution area is any pollution core area among at least one pollution core area.

[0202] If the pollution level is greater than the suspected diffusion threshold, the first suspected pollution area and the first pollution core area are merged based on the preset pollution diffusion fusion rules to obtain the pollution diffusion area.

[0203] Based on preset pollution risk patch merging rules, at least one pollution core area and at least one pollution diffusion area are aggregated to generate a soil pollution area delineation map of the pollution area to be delineated.

[0204] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0205] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for delineating soil pollution areas, characterized in that, include: Acquire spectral remote sensing data of pollution sources, historical polluted site data, and soil pollutant concentration data at preset monitoring points for the area to be delineated as a pollution zone; Spatially overlay and weightedly fuse the historical polluted site data with the spectral absorption features of at least one pixel in the spectral remote sensing image of the pollution source to generate a first pollution feature layer. The soil pollutant concentration data is interpolated into a second pollution concentration layer with the same spatial resolution as the first pollution feature layer according to a preset anisotropic pollution diffusion model. By using a preset pollutant spectral response discrimination model, it is determined whether the spectral absorption characteristics of at least one pixel in the first pollution feature layer indicate pollution, and whether the pollutant concentration at the position corresponding to the at least one pixel in the second pollution concentration layer exceeds a preset pollution threshold. If the spectral characteristics of a certain pixel indicate pollution, and the concentration of pollutants at the corresponding position of the certain pixel exceeds a preset pollution threshold, then the certain pixel is marked as a pollution core pixel. If the spectral absorption characteristics of only a certain pixel indicate pollution, or if the concentration of pollutants at the location corresponding to only that certain pixel exceeds a preset pollution threshold, then that certain pixel is marked as a suspected pollution pixel. If the spectral characteristics of a certain pixel do not indicate pollution, and the concentration of pollutants at the corresponding position of the certain pixel does not exceed a preset pollution threshold, then the certain pixel is marked as a clean pixel; Spatial clustering of at least two adjacent core pollution pixels yields at least one core pollution region, and spatial clustering of at least two adjacent suspected pollution pixels yields at least one suspected pollution region. Determine whether the similarity of the first boundary migration features between the first suspected pollution area and the adjacent first pollution core area is greater than a preset suspected diffusion threshold, wherein the first suspected pollution area is any pollution core area among at least one pollution core area. If the pollution level is greater than the suspected diffusion threshold, the first suspected pollution area and the first pollution core area are merged based on the preset pollution diffusion fusion rules to obtain the pollution diffusion area. Based on preset pollution risk patch merging rules, at least one pollution core area and at least one pollution diffusion area are aggregated to generate a soil pollution area delineation map of the pollution area to be delineated.

2. The method for delineating soil pollution areas according to claim 1, characterized in that, The step of spatially overlaying and weighted fusion of the historical contaminated site data with the spectral absorption features of at least one pixel in the pollution source spectral remote sensing image to generate a first pollution feature layer includes: Obtain the geographic center coordinates, pollution intensity level, and historical activity period of a specific site from the historical polluted site data; Based on a certain pollution intensity level and a certain number of historical activity years, a certain persistent impact weight factor for a certain land parcel is calculated, and a certain dynamic impact field centered on a certain geographic center coordinate is constructed according to the certain persistent impact weight factor, thus obtaining all dynamic impact fields for all land parcels. The spectral remote sensing image of the pollution source is preprocessed to extract the spectral absorption depth of each pixel in the preset pollutant characteristic band. The superposition value of all dynamic influence fields at a certain pixel location is calculated as a comprehensive historical pollution impact index for that pixel. The comprehensive historical pollution impact index and the spectral absorption depth of a pixel are input into a preset nonlinear response function to calculate the enhanced spectral characteristic value of the pixel. The expression for the nonlinear response function is as follows: ， In the formula, For pixels Enhanced spectral characteristic values, For pixels At the spectral absorption depth of the preset pollutant characteristic band, For the maximum enhancement coefficient, To control the curvature parameters for enhancing speed, For pixels The comprehensive historical pollution impact index, It is a natural exponential function; Traverse all pixels, normalize the calculated enhanced spectral feature values, and arrange them according to the spatial coordinates of each pixel in the pollution source spectral remote sensing image to generate the first pollution feature layer.

3. The method for delineating soil pollution areas according to claim 1, characterized in that, The step of interpolating the soil pollutant concentration data into a second pollution concentration layer with the same spatial resolution as the first pollution feature layer according to a preset anisotropic pollution diffusion model includes: Obtain the measured values ​​of soil pollutant concentrations, the three-dimensional geographic coordinates of all preset monitoring points, and the sampling depth within the polluted area to be divided. Obtain the digital elevation model of the polluted area to be delineated, the spatial distribution data of groundwater flow direction, and the prevailing wind direction of the area; Based on the spatial distribution data of groundwater flow direction and the prevailing wind direction in the region, a dominant pollution diffusion direction is calculated and determined for each planar location within the pollution area to be divided using a vector synthesis method. Based on the measured values ​​of soil pollutant concentration and three-dimensional geographic coordinates, and combined with the dominant pollution diffusion direction of each plane position in the pollution area to be divided, an anisotropic pollution diffusion model is constructed to characterize the spatial distribution correlation of pollutant concentration. The spatial correlation range of the anisotropic pollution diffusion model in the dominant pollution diffusion direction is greater than the spatial correlation range in the horizontal direction perpendicular to the dominant pollution diffusion direction. The Kriging interpolation algorithm based on the anisotropic pollution diffusion model is used to spatially interpolate the measured values ​​of soil pollutant concentrations to predict the estimated pollutant concentrations at each horizontal position within the pollution area to be delineated at the preset standard assessment depth. The estimated pollutant concentrations at all horizontal locations located at the preset standard assessment depth are arranged according to the three-dimensional geographic coordinates of all preset monitoring points to form a continuous two-dimensional concentration distribution surface. The two-dimensional concentration distribution surface is resampled to a spatial resolution and geographic coordinate range that is completely consistent with the first pollution feature layer to generate the second pollution concentration layer.

4. The method for delineating soil pollution areas according to claim 1, characterized in that, The determination of whether the similarity of the first boundary migration features between the first suspected pollution area and the adjacent first pollution core area is greater than a preset suspected diffusion threshold includes: Obtain the center of the first core area of ​​the first pollution core area and the center of the first suspected pollution area, and define the direction of the geometric line connecting the center of the first core area to the center of the first suspected area as the pollution assessment direction. Obtain the first dominant pollution diffusion direction at the center of the first core area, and calculate the directional consistency coefficient based on the angle between the pollution assessment direction and the first dominant pollution diffusion direction. ,in, The angle between the direction of pollution assessment and the direction of primary pollution diffusion; Taking the center of the first core area as the diffusion source point, based on the anisotropic pollution diffusion model, the diffusion probability value P of the pollutants diffused to the center of the first suspected area along the assessment direction is simulated and calculated. Extract the common boundary line between the first suspected pollution area and the first core pollution area; Along the common boundary line, a preset vertical buffer zone is extended to one side of the first suspected pollution area and one side of the first pollution core area, respectively, to obtain the first set of pixels in the buffer zone on the side of the first suspected pollution area and the second set of pixels in the buffer zone on the side of the first pollution core area. Calculate the mean of the first feature values ​​of all pixels in the first pixel set on the first pollution feature layer, and the mean of the second feature values ​​of all pixels in the second pixel set on the first pollution feature layer, respectively. Calculate the boundary feature cosine similarity β between the mean of the first feature value and the mean of the second feature value, and then calculate the similarity based on the directional consistency coefficient. The diffusion probability value P and the boundary feature cosine similarity β are used to calculate the first boundary migration feature similarity, expressed as: ， In the formula, The first boundary migration feature similarity, This is the weighting coefficient for the directional consistency coefficient. These are the weighting coefficients for the diffusion probability values. is the weight coefficient of the cosine similarity of the boundary features, and ; The calculated similarity of the first boundary migration feature is compared with the preset suspicious diffusion threshold.

5. The method for delineating soil pollution areas according to claim 1, characterized in that, After determining whether the similarity of the first boundary migration features between the first suspected pollution area and the adjacent first pollution core area is greater than a preset suspected diffusion threshold, the method further includes: If the pollution level exceeds the suspected diffusion threshold, the first suspected pollution area will be marked as an independent pollution patch, and soil profile pollution data of preset sampling points inside and at the edge of the independent pollution patch will be obtained. Based on the pollution data of each soil profile, a three-dimensional interpolation method based on the vertical transport model of pollutants is used to reconstruct and correct the pollution range and boundary of the independent pollution patches in three dimensions, so as to obtain at least one target pollution area. Based on preset pollution risk patch merging rules, at least one pollution core area, at least one pollution diffusion area, and at least one target pollution area are aggregated to generate a soil pollution area delineation map of the pollution area to be delineated.

6. The method for delineating soil pollution areas according to claim 4, characterized in that, The pollution diffusion fusion rule, based on a preset rule, merges the first suspected pollution area with the first core pollution area to obtain a pollution diffusion area including: The center of the first core area of ​​the first pollution core area is taken as the diffusion source point; Based on the anisotropic pollution diffusion model, the steady-state concentration distribution of pollutants at various locations in space is simulated and calculated when the pollutants diffuse from the diffusion source to the first suspected pollution area. The spatial range with concentration values ​​higher than the preset fusion concentration threshold is extracted and defined as the potential diffusion influence area. Calculate the spatial intersection area between the potential diffusion impact area and the first suspected pollution area, and determine whether the proportion of the spatial intersection area to the area of ​​the first suspected pollution area is greater than a preset proportion threshold. If the ratio is not greater than the preset threshold, the first suspected pollution area will not be merged with the first core pollution area. If the potential diffusion impact area is greater than the preset proportional threshold, the potential diffusion impact area and the first pollution suspected area are spatially superimposed using the line connecting the diffusion source point and the center of the first suspected area as the main axis direction to obtain the superimposed area. A morphological closing operation is performed on the superimposed region, and the continuous spatial range after the morphological closing operation is defined as the fused pollution diffusion region.

7. The method for delineating soil pollution areas according to claim 1, characterized in that, The method, based on preset pollution risk patch merging rules, aggregates at least one pollution core area and at least one pollution diffusion area to generate a soil pollution zoning map of the pollution area to be divided, including: Determine whether any two spatially adjacent regions in the at least one pollution core region and the at least one pollution diffusion region meet the preset merging conditions. The merging conditions include: calculating the feature similarity of the pixels on both sides of the common boundary between the two regions on the first pollution feature layer, and the directional compatibility between the direction of the line connecting the centers of the two regions and the dominant pollution diffusion direction at the corresponding location. Each pair of adjacent regions that meets the merging conditions is marked as a candidate merging pair; From all candidate merge pairs, select the pair with the highest comprehensive score for spatial merging to form a new merged region, and update the region set; The above selection and merging operations are performed iteratively until no candidate merge pairs satisfy the merging conditions are found; Each region in the final region set is defined as an independent pollution patch in the soil pollution zoning map; Based on the average concentration value of each independent pollution patch on the second pollution concentration layer, a pollution risk level is assigned to each independent pollution patch, and a soil pollution area delineation map with the pollution risk level is generated.

8. A soil pollution zone delineation system, used to implement the method according to any one of claims 1-7, characterized in that, The system includes: The processing module is configured to process the pollution source spectral remote sensing data, historical polluted site data, and soil pollutant concentration data at preset monitoring points in the polluted area to be divided according to preset data processing rules, so as to obtain a first pollution feature layer and a second pollution concentration layer. The first judgment module is configured to determine, through a preset pollutant spectral response discrimination model, whether the spectral absorption characteristics of at least one pixel in the first pollution feature layer indicate pollution, and whether the pollutant concentration at the position corresponding to the at least one pixel in the second pollution concentration layer exceeds a preset pollution threshold. The clustering module is configured to mark at least one pixel as a polluted core pixel, a suspected polluted pixel, or a clean pixel based on the judgment result, and to perform spatial clustering on at least two adjacent polluted core pixels to obtain at least one polluted core region, and to perform spatial clustering on at least two adjacent suspected polluted pixels to obtain at least one suspected polluted region. The second judgment module is configured to judge whether the similarity of the first boundary migration feature between the first suspected pollution area and the adjacent first pollution core area is greater than a preset suspected diffusion threshold, wherein the first suspected pollution area is any pollution core area among at least one pollution core area. The fusion module is configured to, if the pollution level is greater than the suspected diffusion threshold, merge the first suspected pollution area with the first core pollution area based on a preset pollution diffusion fusion rule to obtain the pollution diffusion area. The generation module is configured to aggregate at least one pollution core area and at least one pollution diffusion area based on a preset pollution risk patch merging rule, and generate a soil pollution area delineation map of the pollution area to be delineated.

Images