A Method for Identifying Soil Pollution Around Enterprises Based on Vegetation Heavy Metal Stress

By establishing elevation descent connections and stress phase continuity chains, the temporal sequence of vegetation heavy metal stress is transmitted in reverse, forming a reverse source aggregation of enterprise boundary stress. This solves the problem of inconsistency between the direction of pollution range expansion and the direction of terrain propagation in existing technologies, and achieves accurate soil pollution identification and source direction determination.

CN123112367APending Publication Date: 2026-11-17TECH CENT FOR SOIL AGRI & RURAL ECOLOGY & ENVIRONMENT MINIST OF ECOLOGY & ENVIRONMENT
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611332990.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-31
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing methods for identifying soil pollution around enterprises based on remote sensing images cannot effectively distinguish the source-direction relationship between the location of heavy metal stress in vegetation and the enterprise. This results in the pollution range expansion direction being inconsistent with the topographic propagation direction, unclear source-direction location at the enterprise boundary, and easy confusion between vegetation stress from different sources.

Method used

By acquiring remote sensing data of enterprise boundaries and multiple time phases, the vegetation heavy metal stress index is calculated, an elevation descent connection and stress phase continuity chain are established, and the vegetation heavy metal stress time phase sequence is transmitted in reverse along the topographic stress phase continuity chain to form a reverse source aggregation of enterprise boundary stress. Cross-time phase assessment and closure calculation are then performed to identify the range of soil pollution around the enterprise.

Benefits of technology

It enables accurate identification of soil pollution around enterprises, reduces the impact of overall brightness and vegetation growth changes on remote sensing data, distinguishes the spatial and temporal propagation characteristics of vegetation anomalies, and ensures the accuracy of pollution source determination.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123112367A_ABST
    Figure CN123112367A_ABST
Patent Text Reader

Abstract

This invention discloses a method for identifying soil pollution around enterprises based on vegetation heavy metal stress, relating to the fields of ecological environment monitoring and soil pollution identification technology. It acquires remote sensing data on the boundaries, elevations, and multi-temporal multispectral and hyperspectral data of operating enterprises; calculates the vegetation heavy metal stress index and forms a temporal sequence; constructs a topographic stress phase continuity chain based on elevation descent, stress continuity time difference, and stress phase continuity, compressing vegetation stress back to the enterprise boundary to form a reverse source convergence body at the enterprise boundary; determines the overlap range through cross-temporal misalignment testing and performs source-to-source return, calculating the enterprise source-to-source pollution closure amount; then constructs the shifted source-to-source pollution closure amount and misaligned source closure reference amount along the enterprise boundary, thereby determining the remote sensing identification range of soil pollution around the enterprise and the pollution source-to-boundary range.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of ecological environment monitoring and soil pollution identification technology, specifically a method for identifying soil pollution around enterprises based on heavy metal stress caused by vegetation. Background Technology

[0002] In soil pollution monitoring, heavy metal pollution can cause changes in vegetation chlorophyll content, leaf internal structure, canopy reflectance characteristics, and red edge position. Therefore, red light, red edge, and near-infrared spectral information can be used to analyze the stress state of vegetation. Based on the continuous changes in vegetation spectral status at multiple collection phases, potential heavy metal pollution areas in the soil surrounding enterprises can be identified. For soil pollution identification around operating enterprises, in addition to determining whether vegetation is under heavy metal stress, it is also necessary to further analyze whether the stress on vegetation at different spatial locations around the enterprise has source-direction characteristics that propagate along the terrain and point towards the current enterprise boundary.

[0003] Existing methods for identifying soil pollution around enterprises based on remote sensing imagery often use single- or multi-temporal vegetation indices, spectral characteristics, and heavy metal stress features as model inputs, employing machine learning models to obtain the spatial distribution of suspected soil pollution. While these methods can locate areas of vegetation spectral anomalies, the relationship between vegetation heavy metal stress locations and the current enterprise is typically analyzed only through distance, buffer zone, or spatial proximity. Further analysis is not conducted to determine whether different vegetation grids simultaneously satisfy the same downward elevation direction, the order of stress occurrence, and the continuity of stress change patterns. When roads, historical storage yards, other production areas, or existing soil anomalies coexist around the enterprise, different sources may generate similar vegetation spectral responses. Relying solely on the degree of vegetation heavy metal stress and spatial distance makes it difficult to determine whether abnormal vegetation can converge towards the current enterprise along topographical and temporal propagation relationships. This easily leads to the conflation of two distinct issues—"the existence of vegetation stress" and "the stress originating from the current enterprise"—in the same identification result.

[0004] The aforementioned issues arise primarily because the spatial propagation of pollution impacts does not simply spread uniformly outward from the enterprise center. Instead, it is influenced by factors such as elevation variations, the direction of surface subsidence, and differences in the duration of stress at different locations. Furthermore, the same propagation process may manifest in remote sensing imagery as spectral changes occurring first in high-altitude vegetation, followed by similar changes in low-altitude vegetation after several acquisition phases. When these spatial and temporal information are not collectively used in source-direction analysis, independent vegetation anomalies may be grouped into the same enterprise due to proximity, or they may be classified as different pollution areas due to shifts in anomaly location at different acquisition phases. This ultimately leads to problems such as inconsistencies between the direction of pollution expansion and the direction of topographic propagation, unclear source-direction locations at enterprise boundaries, mixed identification of vegetation stress from different sources, and biased attribution of soil pollution around the enterprise. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for identifying soil pollution around enterprises based on heavy metal stress caused by vegetation, thus solving the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for identifying soil pollution around enterprises based on heavy metal stress caused by vegetation, comprising the following steps: S1. Acquire the enterprise boundary, elevation data and UAV, satellite multispectral and hyperspectral remote sensing data of the current production enterprises at multiple acquisition time phases, extract vegetation spectral reflectance characteristics and calculate vegetation heavy metal stress index I, forming a vegetation heavy metal stress time phase sequence for each vegetation grid location. S2. Based on the elevation data, establish elevation descent connections between vegetation grid locations, perform cross-correlation calculations on the heavy metal stress time-phase sequences of adjacent vegetation, and combine the elevation descent rate Gx and the stress continuity time difference Tx to calculate and obtain the stress phase continuity amount KW, and form a topographic stress phase continuity chain. S3. The heavy metal stress time-phase sequence of each vegetation is transmitted in reverse along the topographic stress phase continuity chain, and the reverse transmission amount flowing into the enterprise boundary is compressed and aggregated according to the stress phase continuity amount to form the enterprise boundary stress reverse source back aggregate; the median value of the misaligned back aggregate formed by the cross-temporal misalignment of the enterprise boundary stress reverse source back aggregate is evaluated, and the source back processing is triggered according to the evaluation results. S4. Extract the cross-temporal overlap range of the enterprise boundary stress reverse source back-aggregate, allocate the back-aggregate in the opposite direction of the original reverse transmission connection according to the original contribution ratio of each branch, and perform closed-loop calculation with the obtained source-to-back-aggregate stress quantity FM and the original vegetation heavy metal stress temporal sequence to form the enterprise source-to-pollution closed-loop quantity Cre. S5. Based on the enterprise source-to-pollution closure quantity Cre and the enterprise boundary stress reverse source back-aggregate, a displacement source-to-pollution closure quantity is formed. The above quartile position value is used as the misdirection closure reference quantity QCre. The enterprise source-to-pollution closure quantity Cre and the misdirection closure reference quantity QCre are comprehensively evaluated. Based on the evaluation results, the current remote sensing identification range of soil pollution around the enterprise and the pollution source-to-boundary range are output.

[0007] Preferably, S1 includes S11 and S12; S11. Obtain the enterprise boundary and elevation data of the surrounding area of ​​the enterprise currently in production, and obtain UAV multispectral remote sensing data, UAV hyperspectral remote sensing data and satellite remote sensing data from multiple collection phases within the same identification period; Using the grid size and grid center position of elevation data as a spatial reference, geometric registration and resampling standardization are performed on UAV multispectral remote sensing data, UAV hyperspectral remote sensing data and satellite remote sensing data to ensure that the same surface location falls into the same grid in various types of remote sensing data. Vegetation spectral reflectance features were extracted from each grid cell, including red reflectance Rg, red edge reflectance Rb, and near-infrared reflectance Rw. For hyperspectral remote sensing data, the average values ​​of the reflectance of continuous narrow bands falling within the red band, red edge band, and near-infrared band are used to obtain the red reflectance, red edge reflectance, and near-infrared reflectance, respectively. For multispectral remote sensing data, the reflectance of the corresponding band can be read directly; Retain the grid positions with near-infrared reflectivity Rw greater than red reflectivity Rg as vegetation grid positions, and establish a unified grid space foundation based on enterprise boundaries, elevation data, and each vegetation grid position; S12. For each vegetation grid location at each acquisition time phase, the red edge chlorophyll carrying capacity RE and the near-infrared structural carrying capacity RN are constructed using red light reflectance Rg, red edge reflectance Rb, and near-infrared reflectance Rw, respectively, as shown in the following formulas: ; ; In the formula, This represents a calculation constant to prevent the denominator from being zero, and its value is 0.001. The red-edge chlorophyll carrying capacity (RE) of all vegetation grid locations within the same acquisition phase was arranged from smallest to largest. The red-edge chlorophyll carrying capacity (RE) of the first quartile and the third quartile were read and marked as 25RE and 75RE, respectively. The near-infrared structural load RN formed by all vegetation grid locations within the same acquisition phase was arranged from smallest to largest. The near-infrared structural load RN at the first quartile and the third quartile were read and labeled as 25RN and 75RN, respectively. The normalized red-edge chlorophyll stress nRE and the normalized near-infrared structural stress nRN were calculated and obtained using the following formulas: ; In the formula, This represents the normalized red-edge chlorophyll stress at the i-th vegetation grid location during the t-th data collection phase. This represents the red-edge chlorophyll carrying capacity at the third quartile position during the t-th acquisition phase. This represents the red-edge chlorophyll carrying capacity at the first quartile position during the t-th acquisition phase; ; In the formula, This represents the normalized near-infrared structural stress at the i-th vegetation grid location during the t-th acquisition phase. This represents the near-infrared structural load at the third quartile position during the t-th acquisition phase. This represents the near-infrared structural load capacity at the first quartile position during the t-th acquisition phase; The vegetation heavy metal stress index I was obtained by combining the normalized red-edge chlorophyll stress value nRE and the normalized near-infrared structural stress value nRN, as shown in the following formula: ; In the formula, This represents the vegetation heavy metal stress index at the i-th vegetation grid location during the t-th data collection phase. For the same vegetation grid location, the vegetation heavy metal stress index I is arranged sequentially according to the order of the collection time phases to form the vegetation heavy metal stress time phase sequence for the current vegetation grid location; the same processing is performed on all vegetation grid locations to obtain the vegetation heavy metal stress time phase sequence for the entire area surrounding the enterprise.

[0008] Preferably, S2 includes S21, which reads the elevation of the current vegetation grid position and the elevation of each of the adjacent vegetation grid positions in the surrounding eight neighborhoods; takes the adjacent vegetation grid positions with elevations lower than the current vegetation grid position as the descending receiving position of the current vegetation grid position, and establishes an elevation descending connection with the current vegetation grid position as the high end and the descending receiving position as the low end. For each elevation drop connection, the elevation drop rate Gx is calculated based on the elevation difference between the high and low ends and the grid center distance, using the following formula: ; In the formula, Gx specifically represents the elevation descent rate of the x-th elevation descent connection. This represents the elevation of the high-level vegetation grid position connected by the xth elevation descent link. Dx represents the elevation of the low-end vegetation grid position of the xth elevation descent connection, and Dx represents the distance between the centers of the high-end and low-end grids. Read the vegetation heavy metal stress time-phase sequence of the xth elevation descent connection between the high and low ends: The high-order sequences are as follows: ; The low-order sequences are as follows: Where m represents the number of acquisition phases; Keep the high-end vegetation heavy metal stress time phase sequence position unchanged, and starting from 0 shifts in the collection time phase, shift the low-end vegetation heavy metal stress time phase sequence successively along the direction of increasing collection time phase. When the data collection phases are shifted by q, the vegetation heavy metal stress indices from the 1st to the mqth indices at the higher end are paired with the vegetation heavy metal stress indices from the (1+q)th to the mth indices at the lower end, and the shift cross-correlation coefficient between the two sequences is calculated using Pearson cross-correlation. ; When q=0, compare the stress changes of the two vegetation grid locations at the same sampling time phase; When q=1, compare the stress change of the current acquisition phase at the higher end with the next acquisition phase at the lower end; execute sequentially until the remaining paired phases can no longer be cross-correlation calculated; Read the misalignment phase number q with the largest cross-correlation coefficient among all misalignment positions, and use the current misalignment phase number as the stress continuity time difference Tx of the x-th elevation descent connection, as shown in the formula: ; In the formula, Indicates the search for The maximum value of q is obtained, and the cross-correlation coefficient of the current shift position is denoted as q. .

[0009] Preferably, S2 further includes S22, where S22 reads the elevation decrease rate Gx of each elevation decrease connection formed outward from the same high-level vegetation grid position for all elevation decrease connections, and calculates the terrain distribution flow LGx of the current connection based on the ratio of the elevation decrease rate of the current elevation decrease connection to the sum of the elevation decrease rates of all connections at the same high level, as shown in the following formula: ; In the formula, The elevation decrease rate represents the r-th elevation decrease connection starting from the same high-level vegetation grid position, and n represents the number of elevation decrease connections formed starting from the current high-level vegetation grid position. Cross-correlation coefficients of shift Normalization is performed to convert it to the range of 0 to 1, and the cross-temporal propagation length is compressed using the stress continuity time difference Tx. The stress phase continuity amount KW is then calculated using the following formula: ; In the formula, This represents the stress phase continuity of the xth elevation descent connection; Based on the stress phase continuity amount KW, continuous connection is made at the high and low ends of each elevation descent connection: When the low end of the xth elevation descent connection and the high end of another elevation descent connection belong to the same vegetation grid position, the two connections will be continuously connected along the elevation descent direction. When the same vegetation grid location forms an elevation descent connection to multiple low-level ends, it extends to each low-level end to form a continuing branch. At the same time, the stress continuity time difference Tx of each connection is accumulated step by step along the current branch, so that each low-level vegetation grid location retains the cumulative shift time phase number relative to the starting position of the branch. The stress phase continuity amount KW of each connection is attached to the current connection until it is connected to a position where there is no next low-level vegetation grid location, forming a terrain stress phase continuity chain.

[0010] Preferably, S3 includes S31, S31: Read the terrain stress phase continuation chain, find the end vegetation grid position in the chain where there is no next low-end vegetation grid position, take the vegetation heavy metal stress time phase sequence of the current end vegetation grid position as the reverse transmission starting amount, and transmit it step by step to the enterprise boundary in the opposite direction of the terrain stress phase continuation chain. By advancing the low-end vegetation heavy metal stress time-phase sequence by Tx acquisition phases, the vegetation heavy metal stress index at the (t+Tx)th acquisition phase of the low end participates in the reverse transmission at the (t)th acquisition phase of the high end, thus obtaining the reverse phase reset amount of the current connection, as shown in the following formula: ; In the formula, This represents the reverse phase reset amount of the t-th acquisition phase of the x-th elevation descent connection. This represents the amount to be transmitted at the lower end of the current connection during the (t+Tx)th acquisition phase; In this process, the vegetation heavy metal stress index is used as the initial amount to be transferred at the end position. Each time the reverse direction of the enterprise passes through the first-level terrain stress phase continuity chain, the phase is first reset through Tx, and then the reverse transfer amount is formed through the stress phase continuity amount KW. After multiple reverse transfer amounts are merged into the same vegetation grid position, they are compressed and aggregated. The resulting compression and aggregation result continues to be used as the amount to be transferred when the current vegetation grid position is reversed to the next layer, until it is transferred to the enterprise boundary. When multiple low-level vegetation grids converge in reverse along different connections to the same high-level vegetation grid location, the reverse transmission amount is compressed according to the proportion of the current stress phase carrying capacity KW in the total stress phase carrying capacity KW of all converging connections, as shown in the following formula: ; In the formula, This represents the amount of pressure-cohesion transfer formed at the current high-level vegetation grid location during the t-th acquisition phase, and nc represents the number of connections flowing back into the current high-level location. This represents the stress phase continuity of the c-th connection that flows in reverse into the same high-end vegetation grid position; The current processing requires a branch to undergo three consecutive actions before it can continue to propagate toward the enterprise boundary: "Tx reverse phase reset - KW continuation weighting - branch compression". Tx first eliminates the temporal lag generated step by step during propagation, and KW then retains the terrain and stress continuation characteristics extracted by S2. Therefore, distant vegetation stress will not directly participate in the back-convergence just because the spatial path can reach the enterprise boundary. After the reverse propagation reaches the enterprise boundary, a fixed starting point on the enterprise boundary is used as the arc length starting point. The arc length is continuously calculated along the enterprise boundary, and the enterprise boundary arc length position is defined. The compression propagation amount reaching the same enterprise boundary arc length position and the same acquisition phase is accumulated to obtain the reverse source repolymerization value at the enterprise boundary arc length position s at the acquisition phase t. ; The inverse source back-aggregate is formed by the inverse source back-aggregate values ​​at each enterprise boundary arc length position at each acquisition time phase, and the inverse source propagation path from which each inverse source back-aggregate value originates.

[0011] The method for dividing the arc length of the enterprise boundary is as follows: Read the grid edge lengths of the unified grid space base in S1 and obtain the enterprise boundary. First, select the vertex with the smallest x-coordinate among all the vertices of the enterprise boundary as the starting point of the boundary arc length; when there are multiple vertices with the same x-coordinate, select the vertex with the smallest y-coordinate as the starting point of the boundary arc length. Starting from the starting point of the boundary arc length, move continuously along the boundary in a clockwise direction, and use the grid edge length in S1 as a single segment length to continuously segment the enterprise boundary; Specifically, starting from the starting point of the boundary arc length, move one grid cell length along the enterprise boundary, and record the range of enterprise boundary passed through as the first enterprise boundary arc length position; then, from the end point of the first enterprise boundary arc length position, continue moving one grid cell length clockwise, and record the range of enterprise boundary passed through as the second enterprise boundary arc length position; continue dividing along the enterprise boundary in the above manner until it approaches the starting point of the boundary arc length again. If the remaining enterprise boundary length is less than one grid cell length, the remaining boundary is recorded as the last enterprise boundary arc length position. This divides the entire enterprise boundary into multiple enterprise boundary arc length positions arranged consecutively in a clockwise direction. During the reverse transmission process of S31, when a terrain-stressed phase continuation chain reaches the enterprise boundary, the position where the current reverse transmission branch first intersects with the enterprise boundary is obtained, and the branch is traced clockwise from the starting point of the boundary arc length to the current intersection position. Based on the enterprise boundary segment into which the current intersection position falls, the current reverse transmission branch is assigned to the current enterprise boundary arc length position. For the same acquisition phase, if the first intersection of multiple reverse propagation branches falls at the same enterprise boundary arc length position, the compression propagation amount when these reverse propagation branches reach the enterprise boundary is accumulated, and the accumulated result is used as the reverse source back-polymerization value at the current enterprise boundary arc length position in the current acquisition phase; if different reverse propagation branches fall at different enterprise boundary arc length positions, reverse source back-polymerization values ​​are formed at their respective enterprise boundary arc length positions.

[0012] The reverse source propagation path is obtained as follows: During the reverse propagation along the terrain stress phase continuation chain, each end vegetation grid position is taken as the starting position, and the vegetation grid positions passed through when the current end vegetation grid position is propagated back to the enterprise boundary, as well as the elevation descent connections between adjacent vegetation grid positions, are recorded sequentially until the current reverse propagation reaches the enterprise boundary; the current end vegetation grid position, the vegetation grid positions passed through in sequence, and the elevation descent connections are continuously recorded in the order of reverse propagation to form the reverse source propagation path of the current end vegetation grid position.

[0013] Preferably, S3 further includes S32, where S32, based on the reverse source back-convergence of enterprise boundary stress, the back-convergence proportion of enterprise boundary arc length position s at acquisition time t is obtained by proportionalizing the reverse source back-convergence values ​​of all enterprise boundary arc length positions at each acquisition time phase with respect to the current acquisition time phase. : The process is as follows: For each acquisition phase, first sum the inverse source convergence values ​​at all arc length positions on the enterprise boundary to obtain the total boundary convergence for the current acquisition phase; then divide the inverse source convergence value at each arc length position on the enterprise boundary by the total boundary convergence for the current phase to obtain the convergence percentage at the current arc length position. After processing all acquisition phases in the same way, the convergence percentage at different arc length positions on the enterprise boundary for each acquisition phase can be obtained. Without moving the enterprise boundary convergence distribution at any acquisition time phase, the convergence percentage at the same enterprise boundary arc length position across all acquisition time phases is geometrically aggregated and accumulated along the entire enterprise boundary, as follows: ; In the formula, L represents the isotopic reagglomeration amount, L represents the number of enterprise boundary arc length positions, and m represents the number of sampling phases. Maintain the convergence percentage of the first set phase With the spatial position fixed, the enterprise boundary is used as the long axis of the annular arc connecting the beginning and end, and the second to the mth acquisition phases are cyclically shifted respectively; For the h-th misalignment, the t-th acquisition phase is moved (t-1)h arc lengths along the enterprise boundary, and the part that exceeds the end of the enterprise boundary continues to be arranged from the starting point of the enterprise boundary. Each misalignment maintains the original convergence values ​​and spatial morphology of each acquisition phase, and only destroys the co-positional aggregation state of different acquisition phases on the enterprise boundary. For each acquisition phase after the h-th misalignment, the same geometric aggregation as the same in-situ re-aggregation amount H0 is re-executed to obtain the misaligned re-aggregation amount Hh; h is changed and the process is repeated to form multiple misaligned re-aggregation amounts, and the median value of all misaligned re-aggregation amounts is read and denoted as HM; When the isotopic back-aggregate H0 is greater than the median value of the misaligned back-aggregate HM, it indicates that the cross-temporal aggregation in the current enterprise boundary stress reverse source back-aggregate cannot be explained by the result of maintaining the original back-aggregate form but destroying spatial isotopicity. The enterprise boundary stress reverse source back-aggregate is retained and the source-to-back processing of S4 is triggered. When the isotopic repolymerization H0 is not greater than the median value of the misaligned repolymerization HM, the heavy metal stress of the vegetation surrounding the enterprise will continue to be retained as a result of remote sensing monitoring.

[0014] Preferably, S4 includes S41, S41, reading the aggregation occupancy of each enterprise boundary arc length position in each acquisition phase in the enterprise boundary stress reverse source aggregation body, obtaining the enterprise boundary arc length position where aggregation occupancy exists in all acquisition phases, and connecting the continuously adjacent positions along the enterprise boundary to form a cross-phase overlapping range; Read the inverse source back-convergence value at the boundary arc length position of each enterprise within the cross-temporal overlap range and the inverse source propagation path that forms the current inverse source back-convergence value, and use the current inverse source back-convergence value as the back-convergence starting amount; For multiple reverse source transmission paths connected at the same enterprise boundary arc length position, read the compression transmission amount when each reverse source transmission path reaches the enterprise boundary, calculate the proportion of the compression transmission amount of each reverse source transmission path to the sum of the total compression transmission amounts, and allocate the return starting amount using the current proportion. The reverse source transmission path is transmitted step by step from the enterprise boundary to the vegetation grid position. When the return amount reaches the vegetation grid position where multiple reverse source transmission paths converge, the reverse phase reset amount and the stress phase continuity amount KW formed by each connection are read respectively. The ratio of the product of the two to the sum of all input products at the current vegetation grid position is used as the return allocation ratio of the current connection. The return amount reaching the current vegetation grid position is then transmitted to each low-end vegetation grid position.

[0015] Preferably, S4 also includes S42, where S42 is to divide the return amount allocated to each elevation descent connection in S41 by the current connection's stress phase continuity amount KW, and restore the stress value before the low end of the current connection participates in the reverse transmission. Based on the current stress continuity time difference Tx, the recovered stress value is shifted by Tx acquisition time phases in the direction of increasing acquisition time phase, so that it returns to the original acquisition time phase; Process continuously along each reverse source propagation path until returning to each vegetation grid location that forms the current enterprise boundary stress reverse source backflow body. Record the return result obtained at the i-th vegetation grid location at the t-th acquisition phase as the source-to-return stress quantity. ; Among them, the source-to-back stress quantity How to obtain: Read the inverse source back-convergence value within the overlapping temporal range of the enterprise boundary. Using the current inverse source back-convergence value as the starting amount for the back-convergence, proceed along the elevation descending connections traversed when the current inverse source back-convergence value was formed, and back-converge step by step from the enterprise boundary to the vegetation grid location. When the back-convergence reaches the location of a vegetation grid with multiple connections, allocate the current back-convergence to each connection according to the original contribution ratio of each connection in forming the compression transfer amount. For the back-convergence allocated to each connection, remove the effect of the logarithmic value of the stress phase continuity amount KW, and based on the stress continuity time difference Tx of the current connection, move the back-convergence amount Tx times in the direction of increasing acquisition time phase to restore the back-convergence amount to the original acquisition time phase. Process step by step in the above manner until the back-convergence reaches the location of each vegetation grid that participated in the formation of the enterprise boundary stress inverse source back-convergence. Record the back-convergence amount finally obtained at the i-th vegetation grid location in the t-th acquisition time phase as the source-to-backward stress quantity. ; Read the vegetation heavy metal stress index at the same vegetation grid location and the same collection time phase, calculate the degree of closure between the total source-to-back stress amount and the vegetation heavy metal stress index, and obtain the enterprise's source-to-pollution closure amount Cre, as shown in the following formula: In the formula, M represents the number of vegetation grid locations involved in the formation of the overlapping temporal range. Specifically, this represents the source-to-back stress at the i-th vegetation grid location during the t-th acquisition phase.

[0016] Preferably, S5 includes S51, S51: read the cross-temporal overlap range used by the enterprise source to pollution closure quantity Cre, and obtain the number of continuous enterprise boundary arc length positions included in the current cross-temporal overlap range; Keeping the current number of arc length positions and the continuous arrangement order between each arc length position unchanged, starting from the first enterprise boundary arc length position after the current cross-phase overlap range, move one enterprise boundary arc length position clockwise along the enterprise boundary each time to form the 1st, 2nd and up to the gth shift range. When the displacement range exceeds the end of the enterprise boundary, continue arranging from the starting point of the arc length of the enterprise boundary, and exclude any displacement that is exactly the same as the original overlapping range across time phases. For the g-th shift range, the inverse source back-convergence value at the boundary arc length of each enterprise within the current range is read, along with the elevation descent connections that were sequentially passed through when the current inverse source back-convergence value was formed. Using each inverse source back-convergence value as the starting amount for the back-convergence, at locations with multiple elevation descent connections, the back-convergence amount is allocated according to the original contribution ratio when each connection forms the compression transfer amount. The numerical effect of the stress phase continuity amount KW is removed from the back-convergence amount allocated to each elevation descent connection. Then, based on the stress continuity time difference Tx, it is restored to the original acquisition time phase along the direction of increasing acquisition time phase, and processed step-by-step to each vegetation grid location to obtain the source-to-back back-convergence stress amount at each vegetation grid location under the g-th shift. ; Read the vegetation heavy metal stress index of each vegetation grid position at the same acquisition time phase, and calculate the g-th migration source to pollution closure amount gCre.

[0017] Preferably, S5 further includes S52, in which all the shift source to contamination closure quantities formed in S51 are arranged from small to large, and the shift source to contamination closure quantity at the upper quartile position is read as the fault source closure reference quantity QCre; The enterprise's pollution source closure quantity Cre is compared with the misdirection closure reference quantity QCre to obtain the assessment results: When the enterprise source-to-pollution closure amount Cre is not less than the mis-source closure reference amount QCre, it means that the recovery degree of the source-to-pollution return result formed by the original cross-temporal overlap range to the vegetation heavy metal stress temporal sequence exceeds most of the enterprise boundary shift positions. The location of each vegetation grid participating in the calculation of the enterprise source-to-pollution closure amount Cre is determined as the current remote sensing identification range of soil pollution around the enterprise, and the cross-temporal overlap range is determined as the pollution source-to-pollution boundary range. The current enterprise source-to-pollution identification result is output. When the pollution closure amount Cre of the enterprise source direction is less than the mis-source closure reference amount QCre, it means that after migrating the cross-temporal overlapping range to other locations at the enterprise boundary, a similar return recovery result can still be formed. The current spatial distribution of heavy metal stress in vegetation is insufficient to be classified into the current enterprise source direction. The current enterprise does not form a source direction soil pollution identification result, and the temporal sequence of heavy metal stress in vegetation is retained as subsequent remote sensing monitoring data.

[0018] This invention provides a method for identifying soil pollution around enterprises based on heavy metal stress caused by vegetation, which has the following beneficial effects: (1) The vegetation heavy metal stress time-phase sequence is transmitted in reverse along the topographic stress phase continuity chain, and the temporal misalignment generated during the propagation process is restored by using the stress continuity time difference Tx. Then, the stress phase continuity amount KW is used to control the reverse compression of different connections, so that the scattered vegetation stress around the enterprise is finally transformed into the reverse source back-aggregation distribution on the enterprise boundary. Subsequently, the back-aggregation results of different collection time phases are cyclically misaligned on the annular arc length axis of the enterprise boundary to maintain the spatial morphology of each time phase and destroy the cross-temporal iso-positional aggregation to form the misaligned back-aggregation amount, and it is compared with the iso-positional back-aggregation amount. Thus, the current remote sensing results of the enterprise itself can be used to test whether the back-aggregation position on the enterprise boundary has cross-temporal continuity without introducing historical pollution events or other enterprise samples as references, so that the occasional single-temporal boundary aggregation and the continuous enterprise source-to-aggregation can be processed separately.

[0019] (2) The red-edge chlorophyll carrying capacity (RE) and near-infrared structural carrying capacity (RN) within the same acquisition phase are sorted, and the normalized red-edge chlorophyll stress capacity (nRE) and normalized near-infrared structural stress capacity (nRN) are constructed using the data from the first and third quartile positions. This ensures that each vegetation grid position is scaled according to the overall distribution of vegetation around the enterprise at the current acquisition phase, rather than directly comparing absolute spectral values ​​from different remote sensing platforms. This processing can reduce the influence of changes in overall brightness, sensor dimensions, and general vegetation growth status on stress values ​​at different acquisition phases, enabling subsequent comparisons of vegetation stress changes at a unified scale between different acquisition phases. This establishes a continuously usable temporal data foundation for the cross-correlation calculation of adjacent vegetation grid positions.

[0020] (3) Based on the established elevation descent connection, instead of directly assuming the existence of stress propagation between the high-level and low-level ends, the vegetation heavy metal stress time-phase sequences at both ends are read separately. The high-level sequence position is kept unchanged, and the low-level sequence is shifted successively along the direction of increasing collection time phase. The cross-correlation coefficient of the shift is calculated for each shift position. Then, the number of shift time phases with the largest cross-correlation coefficient is read as the stress continuity time difference Tx. The current processing transforms the process of "a certain vegetation stress change first appears at the high-level end, and a similar change appears at the low-level end after several collection time phases" into a calculable time phase difference, so that simple synchronous changes, random fluctuations, and connections that only have topographic descent conditions but no time continuity characteristics can be distinguished, thus establishing a temporal basis for judging whether the vegetation anomaly around the enterprise is propagated step by step along the topographic direction.

[0021] (4) After the formation of the reverse source convergence under enterprise boundary stress, the judgment is not made directly based on the magnitude of the reverse source convergence value at a certain boundary location. Instead, the iso-convergence amount H0 at the same enterprise boundary arc length position in different acquisition phases is calculated first. Then, while keeping the convergence distribution pattern of each acquisition phase unchanged, the different acquisition phases are cyclically misaligned along the enterprise boundary to form multiple misaligned convergence amounts Hh. The median value HM of the misaligned convergence amounts is used to test whether the cross-phase aggregation at the original enterprise boundary position is persistent. The current processing enables the separate handling of high convergence positions that occasionally appear in a single acquisition phase, convergence phenomena with continuously changing positions in different acquisition phases, and situations where multiple acquisition phases continuously point to the same enterprise boundary range. This solves the problem that a single remote sensing anomaly is easily interpreted as the direction of enterprise pollution sources. Attached Figure Description

[0022] Figure 1 This is a schematic diagram illustrating the steps of the method for identifying soil pollution around enterprises based on heavy metal stress caused by vegetation, as described in this invention. Figure 2 This is a schematic diagram of the shift verification and recognition result output process of the present invention. Detailed Implementation

[0023] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0024] Example 1 This invention provides a method for identifying soil pollution around enterprises based on heavy metal stress caused by vegetation. Please refer to [link / reference]. Figure 1 and Figure 2 This includes the following steps: S1. Acquire the enterprise boundary, elevation data and UAV, satellite multispectral and hyperspectral remote sensing data of the current production enterprises at multiple acquisition time phases, extract vegetation spectral reflectance characteristics and calculate vegetation heavy metal stress index I, forming a vegetation heavy metal stress time phase sequence for each vegetation grid location. S2. Based on the elevation data, establish elevation descent connections between vegetation grid locations, perform cross-correlation calculations on the heavy metal stress time-phase sequences of adjacent vegetation, and combine the elevation descent rate Gx and the stress continuity time difference Tx to calculate and obtain the stress phase continuity amount KW, and form a topographic stress phase continuity chain. S3. The heavy metal stress time-phase sequence of each vegetation is transmitted in reverse along the topographic stress phase continuity chain, and the reverse transmission amount flowing into the enterprise boundary is compressed and aggregated according to the stress phase continuity amount to form the enterprise boundary stress reverse source back aggregate; the median value of the misaligned back aggregate formed by the cross-temporal misalignment of the enterprise boundary stress reverse source back aggregate is evaluated, and the source back processing is triggered according to the evaluation results. S4. Extract the cross-temporal overlap range of the enterprise boundary stress reverse source back-aggregate, allocate the back-aggregate in the opposite direction of the original reverse transmission connection according to the original contribution ratio of each branch, and perform closed-loop calculation with the obtained source-to-back-aggregate stress quantity FM and the original vegetation heavy metal stress temporal sequence to form the enterprise source-to-pollution closed-loop quantity Cre. S5. Based on the enterprise source-to-pollution closure quantity Cre and the enterprise boundary stress reverse source back-aggregate, a displacement source-to-pollution closure quantity is formed. The above quartile position value is used as the misdirection closure reference quantity QCre. The enterprise source-to-pollution closure quantity Cre and the misdirection closure reference quantity QCre are comprehensively evaluated. Based on the evaluation results, the current remote sensing identification range of soil pollution around the enterprise and the pollution source-to-boundary range are output.

[0025] In this embodiment, enterprise boundary and elevation data, as well as UAV, satellite multispectral, and hyperspectral remote sensing data from multiple acquisition phases, are first unified to the same vegetation raster location. Red light, red edge, and near-infrared spectral information are then used together to calculate the vegetation heavy metal stress index I. Specifically, red edge chlorophyll stress and near-infrared structural stress are coupled using a product-coupled approach, ensuring that only raster locations where both types of vegetation physiological responses simultaneously exhibit abnormalities generate a large vegetation heavy metal stress index. This index is then compiled into a vegetation heavy metal stress temporal sequence according to the acquisition phases. This approach reduces the interference of individual leaf color changes, individual canopy structure changes, and spatial scale differences between different remote sensing data on subsequent pollution identification, ensuring that the vegetation stress information used in subsequent analysis has both a spatial location basis and a temporal variation basis.

[0026] Instead of directly determining the pollution source based on the spatial distance between the location of vegetation stress and the enterprise, this method uses the cross-correlation results of elevation descent connectivity, elevation descent rate Gx, and the staggered phase sequences of heavy metal stress in adjacent vegetation to obtain the stress continuity time difference Tx and stress phase continuity amount KW, and forms a topographic stress phase continuity chain accordingly. This allows for the simultaneous inclusion of "topographical allowance for migration in a certain direction" and "continuous occurrence of vegetation stress at certain time differences between adjacent locations" in the propagation analysis. This prevents vegetation anomalies that are only spatially adjacent but lack topographic and temporal continuity characteristics from directly entering the enterprise source direction determination, and can be used to distinguish spatial proximity anomalies generated by roads around the enterprise, historical storage yards, and other production areas.

[0027] The back-collection values ​​of the reverse source within the overlapping temporal phase range are returned to each vegetation grid location according to the original contribution ratio. Based on KW and Tx, the original stress values ​​and collection time phases of each vegetation grid location are restored to form the source-to-back stress quantity FM. Then, the degree of recovery between FM and the original vegetation heavy metal stress time phase sequence is used to form the enterprise source-to-pollution closure quantity Cre. At the same time, the overlapping temporal phase range is shifted sequentially along the enterprise boundary to form multiple shifted source-to-pollution closure quantities. The values ​​at the upper quartile positions are used to form the mis-source closure reference quantity QCre.

[0028] Comparing the enterprise source-to-pollution closure quantity Cre with the misdirection closure reference quantity QCre can further determine whether the current cross-temporal overlap range is better able to restore the original vegetation stress spatial-temporal structure than other locations at the enterprise boundary. Based on this, the remote sensing identification range of soil pollution around the enterprise and the pollution source-to-boundary range are output, so that the identification results not only answer "where is suspected soil pollution", but also further answer "whether the current suspected pollution has source-to-boundary characteristics pointing to the currently operating enterprise". Example 2 Please refer to Figure 1 Specifically: S1 includes S11 and S12; S11. Obtain the enterprise boundary and elevation data of the surrounding area of ​​the enterprise currently in production, and obtain UAV multispectral remote sensing data, UAV hyperspectral remote sensing data and satellite remote sensing data from multiple collection phases within the same identification period; Using the grid size and grid center position of elevation data as a spatial reference, geometric registration and resampling standardization are performed on UAV multispectral remote sensing data, UAV hyperspectral remote sensing data and satellite remote sensing data to ensure that the same surface location falls into the same grid in various types of remote sensing data. Vegetation spectral reflectance features were extracted from each grid cell, including red reflectance Rg, red edge reflectance Rb, and near-infrared reflectance Rw. Retain the grid positions with near-infrared reflectivity Rw greater than red reflectivity Rg as vegetation grid positions, and establish a unified grid space foundation based on enterprise boundaries, elevation data, and each vegetation grid position; S12. For each vegetation grid location at each acquisition time phase, the red edge chlorophyll carrying capacity RE and the near-infrared structural carrying capacity RN are constructed using red light reflectance Rg, red edge reflectance Rb, and near-infrared reflectance Rw, respectively, as shown in the following formulas: ; ; In the formula, This represents a calculation constant to prevent the denominator from being zero; The red-edge chlorophyll carrying capacity (RE) of all vegetation grid locations within the same acquisition phase was arranged from smallest to largest. The red-edge chlorophyll carrying capacity (RE) of the first quartile and the third quartile were read and marked as 25RE and 75RE, respectively. The near-infrared structural load RN formed by all vegetation grid locations within the same acquisition phase was arranged from smallest to largest. The near-infrared structural load RN at the first quartile and the third quartile were read and labeled as 25RN and 75RN, respectively. The normalized red-edge chlorophyll stress nRE and the normalized near-infrared structural stress nRN were calculated and obtained using the following formulas: ; In the formula, This represents the normalized red-edge chlorophyll stress at the i-th vegetation grid location during the t-th data collection phase. This represents the red-edge chlorophyll carrying capacity at the third quartile position during the t-th acquisition phase. This represents the red-edge chlorophyll carrying capacity at the first quartile position during the t-th acquisition phase; ; In the formula, This represents the normalized near-infrared structural stress at the i-th vegetation grid location during the t-th acquisition phase. This represents the near-infrared structural load at the third quartile position during the t-th acquisition phase. This represents the near-infrared structural load capacity at the first quartile position during the t-th acquisition phase; The vegetation heavy metal stress index I was obtained by combining the normalized red-edge chlorophyll stress value nRE and the normalized near-infrared structural stress value nRN, as shown in the following formula: ; In the formula, This represents the vegetation heavy metal stress index at the i-th vegetation grid location during the t-th data collection phase. For the same vegetation grid location, the vegetation heavy metal stress index I is arranged sequentially according to the order of the collection time phases to form the vegetation heavy metal stress time phase sequence for the current vegetation grid location; the same processing is performed on all vegetation grid locations to obtain the vegetation heavy metal stress time phase sequence for the entire area surrounding the enterprise.

[0029] In this embodiment, by using the raster size and center position of the elevation data as a unified spatial reference, geometric registration and resampling are performed on UAV multispectral remote sensing data, UAV hyperspectral remote sensing data, and satellite remote sensing data. This ensures that spectral information acquired from different remote sensing platforms and at different acquisition times falls into the same surface raster position, and simultaneously preserves enterprise boundaries, elevation information, and vegetation raster positions on the basis of a unified raster space. This provides a clear and consistent spatial basis for subsequent red reflectance Rg, red-edge reflectance Rb, near-infrared reflectance Rw, and vegetation heavy metal stress index I, solving the problem of difficulty in continuous analysis of the same surface location due to different spatial resolutions and sampling positions of multi-source remote sensing data. It establishes a unified raster unit for subsequent elevation propagation analysis and enterprise boundary inverse source analysis.

[0030] Vegetation grid locations were screened by using near-infrared reflectance (Rw) greater than red reflectance (Rg), allowing subsequent treatments to focus on surface areas where soil heavy metal stress could be reflected through vegetation spectral states. Subsequently, red reflectance (Rg), red-edge reflectance (Rb), and near-infrared reflectance (Rw) were used to calculate red-edge chlorophyll carrying capacity (RE) and near-infrared structural carrying capacity (RN), respectively, enabling vegetation chlorophyll state and canopy and leaf structural states to participate in stress analysis. Compared to methods that rely solely on a single vegetation index to determine anomalies, this treatment simultaneously observes two types of spectral changes in vegetation color, physiological state, and structural state. Spectral anomalies caused solely by leaf color changes or canopy structural fluctuations do not directly determine the final vegetation heavy metal stress index (I), which is beneficial for distinguishing single spectral anomalies caused by some non-heavy metal factors.

[0031] The vegetation heavy metal stress index I was formed by multiplying and taking the square root of the product of the normalized red-edge chlorophyll stress value (nRE) and the normalized near-infrared structural stress value (nRN). This ensures that the final index is jointly controlled by both chlorophyll stress and vegetation structural stress. When one stress value is low, even if the other stress value is high alone, it will not directly result in a high vegetation heavy metal stress index I. Subsequently, the vegetation heavy metal stress indices I at the same vegetation grid location were arranged according to the collection time to form a vegetation heavy metal stress temporal sequence. This transforms vegetation anomalies that were originally scattered across multiple collection time phases into continuous analysis objects with a temporal order. This allows for the identification of stress lag, propagation sequence, and consistency of change between different vegetation grid locations, providing a basis for determining whether vegetation anomalies around an enterprise continuously point to the currently operating enterprise along the terrain direction.

[0032] Example 3 Please refer to Figure 1Specifically: S2 includes S21. S21 reads the elevation of the current vegetation raster position and the elevation of each of the eight neighboring vegetation raster positions; takes the adjacent vegetation raster positions with lower elevations than the current vegetation raster position as the descending receiving positions of the current vegetation raster position, and establishes an elevation descending connection with the current vegetation raster position as the high end and the descending receiving positions as the low end. For each elevation drop connection, the elevation drop rate Gx is calculated based on the elevation difference between the high and low ends and the grid center distance, using the following formula: ; In the formula, This represents the elevation of the high-level vegetation grid position connected by the xth elevation descent link. Dx represents the elevation of the low-end vegetation grid position of the xth elevation descent connection, and Dx represents the distance between the centers of the high-end and low-end grids. Read the vegetation heavy metal stress time-phase sequence of the xth elevation descent connection between the high and low ends: The high-order sequences are as follows: ; The low-order sequences are as follows: Where m represents the number of acquisition phases; Keep the high-end vegetation heavy metal stress time phase sequence position unchanged, and starting from 0 shifts in the collection time phase, shift the low-end vegetation heavy metal stress time phase sequence successively along the direction of increasing collection time phase. When the data collection phases are shifted by q, the vegetation heavy metal stress indices from the 1st to the mqth indices at the higher end are paired with the vegetation heavy metal stress indices from the (1+q)th to the mth indices at the lower end, and the shift cross-correlation coefficient between the two sequences is calculated using Pearson cross-correlation. ; Read the misalignment phase number q with the largest cross-correlation coefficient among all misalignment positions, and use the current misalignment phase number as the stress continuity time difference Tx of the x-th elevation descent connection, as shown in the formula: ; In the formula, Indicates the search for The maximum value of q is obtained, and the cross-correlation coefficient of the current shift position is denoted as q. .

[0033] S2 also includes S22. S22 reads the elevation descent rate Gx of each elevation descent connection formed outward from the same high-level vegetation grid position, and calculates the terrain distribution flow LGx of the current connection as the ratio of the elevation descent rate of the current elevation descent connection to the sum of the elevation descent rates of all connections at the same high level. The formula is as follows: ; In the formula, The elevation decrease rate represents the r-th elevation decrease connection starting from the same high-level vegetation grid position, and n represents the number of elevation decrease connections formed starting from the current high-level vegetation grid position. Cross-correlation coefficients of shift Normalization is performed to convert it to the range of 0 to 1, and the cross-temporal propagation length is compressed using the stress continuity time difference Tx. The stress phase continuity amount KW is then calculated using the following formula: ; In the formula, This represents the stress phase continuity of the xth elevation descent connection; Based on the stress phase continuity amount KW, continuous connection is made at the high and low ends of each elevation descent connection: When the low end of the xth elevation descent connection and the high end of another elevation descent connection belong to the same vegetation grid position, the two connections will be continuously connected along the elevation descent direction. When the same vegetation grid location forms an elevation descent connection to multiple low-level ends, it extends to each low-level end to form a continuing branch; at the same time, the stress continuity time difference Tx of each connection is accumulated step by step along the current branch, and the stress phase continuity amount KW of each connection is attached to the current connection until it is connected to a location where there is no next low-level vegetation grid location, forming a terrain stress phase continuity chain.

[0034] In this embodiment, an elevation descent connection is first established based on the elevation difference between the current vegetation grid location and the locations of the eight surrounding neighboring vegetation grid locations. Then, the elevation descent rate Gx is calculated using the elevation difference between the high and low ends and the grid center distance, transforming the spatial proximity relationship between vegetation grids around the enterprise into a topographic connection with a clear descent direction and degree. Compared with methods that determine the direction of pollution propagation solely based on enterprise center distance, buffer radius, or adjacent pixels, the current processing can distinguish vegetation grid locations that are "spatially close but lack topographic descent support conditions." This limits subsequent vegetation heavy metal stress propagation analysis to spatial connections with topographic support, solving the problem in the prior art where proximity anomalies are easily directly attributed to the same enterprise's pollution range.

[0035] Further, the topographic flow LGx is calculated by utilizing all elevation descent connections formed outward from the same high-level end. This eliminates the need for individual connections to be judged independently; instead, their elevation descent rate is used in subsequent calculations based on their proportion within all descent directions at the current high-level end. Simultaneously, the misalignment cross-correlation coefficient, topographic flow LGx, and stress continuity time difference Tx are compressed into a stress phase continuity quantity KW. Therefore, a connection can only generate a relatively high stress phase continuity quantity KW if it simultaneously possesses a relatively clear topographic descent direction, relatively consistent stress phase changes, and a relatively reasonable propagation time difference. If only steep slopes, temporal similarities, or excessively long stress continuity time differences exist, the role of the current connection in subsequent transmission will be limited, making it difficult to incorporate spatial proximity anomalies such as roads, historical storage yards, and other production areas into enterprise source-direction analysis based on a single condition.

[0036] Based on the high and low ends of each elevation descent connection, the stress phase continuity quantity KW is continuously connected, and the stress continuity time difference Tx is accumulated step by step along each continuity branch to form a terrain stress phase continuity chain that simultaneously retains information on spatial descent direction, stress propagation strength, and temporal lag. The current chain structure organizes the vegetation heavy metal stress information, which was originally scattered in different vegetation grid locations and different collection time phases, into a continuously traceable propagation sequence. This allows subsequent investigations to trace the source direction of each vegetation stress along the current chain in reverse, and to analyze whether different propagation sequences converge at the enterprise boundary. This provides a direct processing basis for distinguishing between "vegetation stress around the enterprise" and "current vegetation stress has source-direction characteristics pointing to the currently operating enterprise."

[0037] Example 4 Please refer to Figure 1 Specifically: S3 includes S31. S31: Read the terrain stress phase continuation chain, find the end vegetation grid position in the chain where there is no next low-end vegetation grid position, take the vegetation heavy metal stress time phase sequence of the current end vegetation grid position as the reverse transmission starting amount, and transmit it to the enterprise boundary level by level in the opposite direction of the terrain stress phase continuation chain. By advancing the low-end vegetation heavy metal stress time-phase sequence by Tx acquisition phases, the vegetation heavy metal stress index at the (t+Tx)th acquisition phase of the low end participates in the reverse transmission at the (t)th acquisition phase of the high end, thus obtaining the reverse phase reset amount of the current connection, as shown in the following formula: ; In the formula, This represents the reverse phase reset amount of the t-th acquisition phase of the x-th elevation descent connection. This represents the amount to be transmitted at the lower end of the current connection during the (t+Tx)th acquisition phase; When multiple low-level vegetation grids converge in reverse along different connections to the same high-level vegetation grid location, the reverse transmission amount is compressed according to the proportion of the current stress phase carrying capacity KW in the total stress phase carrying capacity KW of all converging connections, as shown in the following formula: ; In the formula, This represents the amount of pressure-cohesion transfer formed at the current high-level vegetation grid location during the t-th acquisition phase, and nc represents the number of connections flowing back into the current high-level location. This represents the stress phase continuity of the c-th connection that flows in reverse into the same high-end vegetation grid position; After the reverse propagation reaches the enterprise boundary, a fixed starting point on the enterprise boundary is used as the arc length starting point. The arc length is continuously calculated along the enterprise boundary, and the enterprise boundary arc length position is defined. The compression propagation amount reaching the same enterprise boundary arc length position and the same acquisition phase is accumulated to obtain the reverse source repolymerization value at the enterprise boundary arc length position s at the acquisition phase t. ; The inverse source back-aggregate is formed by the inverse source back-aggregate values ​​at each enterprise boundary arc length position at each acquisition time phase, and the inverse source propagation path from which each inverse source back-aggregate value originates.

[0038] S3 also includes S32. S32, based on the reverse source re-aggregate of enterprise boundary stress, obtains the re-aggregate proportion of enterprise boundary arc length position s at acquisition time t by proportionalizing the reverse source re-aggregate values ​​of all enterprise boundary arc length positions at each acquisition time phase and the current acquisition time phase. : Without moving the enterprise boundary convergence distribution at any acquisition time phase, the convergence percentage at the same enterprise boundary arc length position across all acquisition time phases is geometrically aggregated and accumulated along the entire enterprise boundary, as follows: ; In the formula, L represents the isotopic reagglomeration amount, L represents the number of enterprise boundary arc length positions, and m represents the number of sampling phases. Maintain the convergence percentage of the first set phase With the spatial position fixed, the enterprise boundary is used as the long axis of the annular arc connecting the beginning and end, and the second to the mth acquisition phases are cyclically shifted respectively; For the h-th misalignment, the t-th acquisition phase is moved (t-1)h arc lengths along the enterprise boundary, and the portion that exceeds the end of the enterprise boundary continues to be arranged from the starting point of the enterprise boundary. For each acquisition phase after the h-th misalignment, the same geometric aggregation as the same in-situ re-aggregation amount H0 is re-executed to obtain the misaligned re-aggregation amount Hh; h is changed and the process is repeated to form multiple misaligned re-aggregation amounts, and the median value of all misaligned re-aggregation amounts is read and denoted as HM; When the isotopic back-aggregate H0 is greater than the median value of the misaligned back-aggregate HM, it indicates that the cross-temporal aggregation in the current enterprise boundary stress reverse source back-aggregate cannot be explained by the result of maintaining the original back-aggregate form but destroying spatial isotopicity. The enterprise boundary stress reverse source back-aggregate is retained and the source-to-back processing of S4 is triggered. When the isotopic repolymerization H0 is not greater than the median value of the misaligned repolymerization HM, the heavy metal stress of the vegetation surrounding the enterprise will continue to be retained as a result of remote sensing monitoring.

[0039] S4 includes S41. S41 reads the back-aggregation occupancy of each enterprise boundary arc length position in each acquisition phase of the enterprise boundary stress reverse source back-aggregate, obtains the enterprise boundary arc length position where back-aggregation occupancy exists in all acquisition phases, and connects the continuously adjacent positions along the enterprise boundary to form a cross-phase overlapping range. Read the inverse source back-convergence value at the boundary arc length position of each enterprise within the cross-temporal overlap range and the inverse source propagation path that forms the current inverse source back-convergence value, and use the current inverse source back-convergence value as the back-convergence starting amount; For multiple reverse source transmission paths connected at the same enterprise boundary arc length position, read the compression transmission amount when each reverse source transmission path reaches the enterprise boundary, calculate the proportion of the compression transmission amount of each reverse source transmission path to the sum of the total compression transmission amounts, and allocate the return starting amount using the current proportion. The reverse source transmission path is transmitted step by step from the enterprise boundary to the vegetation grid position. When the return amount reaches the vegetation grid position where multiple reverse source transmission paths converge, the reverse phase reset amount and the stress phase continuity amount KW formed by each connection are read respectively. The ratio of the product of the two to the sum of all input products at the current vegetation grid position is used as the return allocation ratio of the current connection. The return amount reaching the current vegetation grid position is then transmitted to each low-end vegetation grid position.

[0040] S4 also includes S42, which is the return amount allocated to each elevation descent connection by S41, divided by the current connection's stress phase continuity amount KW, to restore the stress value before the current connection's low end participates in the reverse transmission. Based on the current stress continuity time difference Tx, the recovered stress value is shifted by Tx acquisition time phases in the direction of increasing acquisition time phase, so that it returns to the original acquisition time phase; Process continuously along each reverse source propagation path until returning to each vegetation grid location that forms the current enterprise boundary stress reverse source backflow body. Record the return result obtained at the i-th vegetation grid location at the t-th acquisition phase as the source-to-return stress quantity. ; Read the vegetation heavy metal stress index at the same vegetation grid location and the same collection time phase, calculate the degree of closure between the total source-to-back stress amount and the vegetation heavy metal stress index, and obtain the enterprise's source-to-pollution closure amount Cre, as shown in the following formula: In the formula, M represents the number of vegetation grid locations involved in the formation of the overlapping temporal range.

[0041] In this embodiment, the vegetation heavy metal stress time-phase sequence is transmitted step by step in the opposite direction of the topographic stress phase continuity chain. The stress continuity time difference Tx is used to advance the vegetation heavy metal stress state at the lower end to a collection time phase that matches the higher end. Then, the reverse transmission intensity of each connection is controlled according to the stress phase continuity amount KW. This ensures that vegetation stress at different spatial locations is no longer simply traced back to the enterprise boundary based on geographical distance, but is simultaneously constrained by topographic direction, the order of stress propagation, and the consistency of temporal changes. This further distinguishes cases in the background technology where vegetation is included in the enterprise's pollution range only because of its abnormal location near the enterprise, giving vegetation stress entering the enterprise boundary analysis a continuous spatial propagation basis and a temporal continuity basis.

[0042] When multiple low-level vegetation grid locations converge in the opposite direction to the same high-level vegetation grid location, the reverse transmission amount is compressed and aggregated using the KW ratio of the stress phase continuity of each connection. After the reverse transmission reaches the enterprise boundary, the reverse source back-aggregation value is accumulated according to the arc length position of the enterprise boundary and the acquisition time phase to form a reverse source back-aggregation value. The reverse source back-aggregation values ​​of multiple acquisition time phases together constitute the enterprise boundary stress reverse source back-aggregate. The current processing can gradually converge the scattered vegetation heavy metal stress around the enterprise to the specific location on the enterprise boundary. This transforms the remote sensing results, which originally only indicated "abnormal vegetation exists near the enterprise," into source-direction information of "whether different abnormal vegetation points to a continuous range at the enterprise boundary." This is beneficial for identifying the boundary location of enterprise pollution sources that is difficult to determine in the background technology.

[0043] For the enterprise boundary stress reverse source back-aggregates that have passed the cross-temporal aggregation test, the cross-temporal overlap range is further extracted. Using the reverse source back-aggregate value as the starting amount for the back-aggregation, the back-aggregate is distributed to the vegetation grid locations level by level according to the original compression contribution ratio of each reverse source propagation path. Subsequently, the effect of the stress phase continuity KW logarithmic value is removed, and the original acquisition time phase is restored based on the stress continuity time difference Tx. The source-direction back-aggregate stress amount FM at each vegetation grid location is obtained. Then, the closure degree of FM is calculated with the original vegetation heavy metal stress time phase sequence to form the enterprise source-direction pollution closure amount Cre. The current processing does not only determine whether vegetation stress can reach the enterprise in reverse, but further examines whether the back-aggregate results at the enterprise boundary can reappear to the original vegetation stress spatial-temporal state after returning along the original propagation structure. Therefore, it can further distinguish between "accidentally being traceable to the enterprise boundary" and "vegetation heavy metal stress with a stable enterprise source-direction structure," establishing a more binding discriminative basis for subsequently determining the remote sensing identification range of soil pollution around the enterprise and the pollution source-direction boundary range.

[0044] Example 5 For a method to identify soil pollution around enterprises based on heavy metal stress caused by vegetation, please refer to [reference needed]. Figure 2 Specifically: S5 includes S51. S51 reads the cross-temporal overlap range used by the enterprise source to pollution closure quantity Cre, and obtains the number of continuous enterprise boundary arc length positions included in the current cross-temporal overlap range. Keeping the current number of arc length positions and the continuous arrangement order between each arc length position unchanged, starting from the first enterprise boundary arc length position after the current cross-phase overlap range, move one enterprise boundary arc length position clockwise along the enterprise boundary each time to form the 1st, 2nd and up to the gth shift range. When the displacement range exceeds the end of the enterprise boundary, continue arranging from the starting point of the arc length of the enterprise boundary, and exclude any displacement that is exactly the same as the original overlapping range across time phases. For the g-th shift range, the inverse source back-convergence value at the boundary arc length of each enterprise within the current range is read, along with the elevation descent connections that were sequentially passed through when the current inverse source back-convergence value was formed. Using each inverse source back-convergence value as the starting amount for the back-convergence, at locations with multiple elevation descent connections, the back-convergence amount is allocated according to the original contribution ratio when each connection forms the compression transfer amount. The numerical effect of the stress phase continuity amount KW is removed from the back-convergence amount allocated to each elevation descent connection. Then, based on the stress continuity time difference Tx, it is restored to the original acquisition time phase along the direction of increasing acquisition time phase, and processed step-by-step to each vegetation grid location to obtain the source-to-back back-convergence stress amount at each vegetation grid location under the g-th shift. ; Read the vegetation heavy metal stress index of each vegetation grid position at the same acquisition time phase, and calculate the g-th migration source to pollution closure amount gCre.

[0045] S5 also includes S52, where all the shift source to contamination closure quantities formed by S51 are arranged from small to large, and the shift source to contamination closure quantity at the upper quartile position is read as the fault source closure reference quantity QCre; The enterprise's pollution source closure quantity Cre is compared with the misdirection closure reference quantity QCre to obtain the assessment results: When the enterprise source-to-pollution closure amount Cre is not less than the mis-source closure reference amount QCre, it means that the recovery degree of the source-to-pollution return result formed by the original cross-temporal overlap range to the vegetation heavy metal stress temporal sequence exceeds most of the enterprise boundary shift positions. The location of each vegetation grid participating in the calculation of the enterprise source-to-pollution closure amount Cre is determined as the current remote sensing identification range of soil pollution around the enterprise, and the cross-temporal overlap range is determined as the pollution source-to-pollution boundary range. The current enterprise source-to-pollution identification result is output. When the enterprise's source-to-pollution closure amount Cre is less than the mis-source closure reference amount QCre, it means that after migrating the cross-temporal overlap range to other locations at the enterprise boundary, a return recovery result can still be formed. The current enterprise has not formed a source-to-soil pollution identification result, and the vegetation heavy metal stress temporal sequence is retained as subsequent remote sensing monitoring data.

[0046] In this embodiment, by keeping the number of enterprise boundary arc length positions and their consecutive arrangement order within the cross-temporal overlap range unchanged, the current cross-temporal overlap range is shifted along the enterprise boundary arc length positions one by one. Each shift only changes the location of the source-direction boundary, without changing the length and internal structure of the source-direction boundary itself. Subsequently, the inverse source back-aggregation value is reread within each shifted range, and source-direction back-processing is performed. This current method can construct multiple hypothetical source-direction positions under the conditions of the same enterprise, the same enterprise boundary, and the same batch of vegetation heavy metal stress data, so that the original cross-temporal overlap range has a directly comparable internal spatial reference within the enterprise, without relying on other enterprises, historical pollution events, or manually selected reference areas.

[0047] For each shifted area, instead of simply comparing the magnitude of its reverse source back-aggregation value, the reverse source back-aggregation value within the current shifted area is used as the starting amount for the back-aggregation. The back-aggregation amount is then distributed step-by-step according to the contribution ratio of each elevation descent connection when the original compression transfer amount was formed. At the same time, the effect of the stress phase continuity amount KW on the back-aggregation value is removed, and the original collection phase is restored using the stress continuity time difference Tx. Finally, the source-to-back stress amount at each vegetation grid location is reconstructed. Therefore, each shifted area needs to undergo a spatial back-aggregation and temporal phase restoration process at the same depth as the original cross-temporal overlap area. This allows for verification of whether other locations at the enterprise boundary can also reconstruct the original vegetation heavy metal stress spatial-temporal structure, rather than simply assuming that the current location belongs to the pollution source-to-boundary based on the large back-aggregation amount at a certain enterprise boundary location.

[0048] By calculating the source-direction pollution closure value gCre for each displacement range and arranging all source-direction pollution closure values ​​gCre in ascending order, a mis-source closure reference value QCre is constructed using the values ​​of its upper quartile positions. This mis-source closure reference value QCre directly reflects the closure level that most non-original source-direction positions at the current enterprise boundary can achieve after complete back-return processing. Compared with the method of judging enterprise source direction using fixed empirical values, the current reference value is jointly generated by the current enterprise's own boundary structure, reverse source back-convergence state, terrain transfer structure, and vegetation heavy metal stress temporal sequence. It can change with different enterprise boundary morphologies and different stress spatial distributions, which can reduce the problem of mismatch of uniform discrimination scale caused by differences in enterprise size, boundary shape, and terrain.

[0049] The pollution source closure quantity Cre, formed by the original cross-temporal overlap range, is compared with the mis-source closure reference quantity QCre. When the pollution source closure quantity Cre is not less than the mis-source closure reference quantity QCre, it indicates that the original cross-temporal overlap range is better able to recover the vegetation heavy metal stress temporal sequence involved in the calculation than most of the shifted positions of the enterprise boundary. In this case, the vegetation raster position involved in the calculation of the pollution source closure quantity Cre is taken as the current remote sensing identification range of soil pollution around the enterprise, and the original cross-temporal overlap range is taken as the pollution source boundary range. When the pollution source closure quantity Cre is less than the mis-source closure reference quantity QCre, the current vegetation anomaly is not classified into the current enterprise source direction. The current processing can further distinguish between "the existence of vegetation heavy metal stress around the enterprise" and "the current vegetation heavy metal stress has source direction characteristics pointing to the current operating enterprise," which correspondingly solves the problems in the background technology where anomalies generated by roads, historical stockpiles, and other production areas are easily mixed with current enterprise pollution and the pollution source boundary position is difficult to determine.

[0050] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended technical solutions and their equivalents.

Claims

1. A method for identifying soil pollution around enterprises based on heavy metal stress caused by vegetation, characterized in that: Includes the following steps: S1. Acquire the enterprise boundary, elevation data and UAV, satellite multispectral and hyperspectral remote sensing data of the current production enterprises at multiple acquisition time phases, extract vegetation spectral reflectance characteristics and calculate vegetation heavy metal stress index I, forming a vegetation heavy metal stress time phase sequence for each vegetation grid location. S2. Based on the elevation data, establish elevation descent connections between vegetation grid locations, perform cross-correlation calculations on the heavy metal stress time-phase sequences of adjacent vegetation, and combine the elevation descent rate Gx and the stress continuity time difference Tx to calculate and obtain the stress phase continuity amount KW, and form a topographic stress phase continuity chain. S3. The heavy metal stress time sequence of each vegetation is transmitted in reverse along the topographic stress phase continuity chain, and the reverse transmission amount flowing into the enterprise boundary is compressed and aggregated according to the stress phase continuity amount to form the enterprise boundary stress reverse source aggregate. The median value of the misaligned back-aggregate formed by the cross-temporal phase misalignment of the enterprise boundary-stressed reverse source back-aggregate is evaluated, and the source-to-back processing is triggered based on the evaluation results. S4. Extract the cross-temporal overlap range of the enterprise boundary stress reverse source back-aggregate, allocate the back-aggregate in the opposite direction of the original reverse transmission connection according to the original contribution ratio of each branch, and perform closed-loop calculation with the obtained source-to-back-aggregate stress quantity FM and the original vegetation heavy metal stress temporal sequence to form the enterprise source-to-pollution closed-loop quantity Cre. S5. Based on the enterprise source-to-pollution closure quantity Cre and the enterprise boundary stress reverse source back-aggregate, a displacement source-to-pollution closure quantity is formed. The above quartile position value is used as the misdirection closure reference quantity QCre. The enterprise source-to-pollution closure quantity Cre and the misdirection closure reference quantity QCre are comprehensively evaluated. Based on the evaluation results, the current remote sensing identification range of soil pollution around the enterprise and the pollution source-to-boundary range are output.

2. The method for identifying soil pollution around enterprises based on vegetation heavy metal stress according to claim 1, characterized in that: S1 includes S11 and S12; S11. Obtain the enterprise boundary and elevation data of the surrounding area of ​​the enterprise currently in production, and obtain UAV multispectral remote sensing data, UAV hyperspectral remote sensing data and satellite remote sensing data from multiple collection phases within the same identification period; Using the grid size and grid center position of elevation data as a spatial reference, geometric registration and resampling standardization are performed on UAV multispectral remote sensing data, UAV hyperspectral remote sensing data and satellite remote sensing data to ensure that the same surface location falls into the same grid in various types of remote sensing data. Extract vegetation spectral reflectance features from each grid cell, including red reflectance Rg, red edge reflectance Rb, and near-infrared reflectance Rw; Retain the grid positions with near-infrared reflectivity Rw greater than red reflectivity Rg as vegetation grid positions, and establish a unified grid space foundation based on enterprise boundaries, elevation data, and the positions of each vegetation grid. S12. For each vegetation grid location at each acquisition time phase, the red edge chlorophyll carrying capacity RE and the near-infrared structural carrying capacity RN are constructed using red light reflectance Rg, red edge reflectance Rb, and near-infrared reflectance Rw, respectively, as shown in the following formulas: ; ; In the formula, This represents a calculation constant to prevent the denominator from being zero; The red-edge chlorophyll carrying capacity (RE) of all vegetation grid locations within the same acquisition phase was arranged from smallest to largest. The red-edge chlorophyll carrying capacity (RE) of the first quartile and the third quartile were read and marked as 25RE and 75RE, respectively. The near-infrared structural load RN formed by all vegetation grid locations within the same acquisition phase was arranged from smallest to largest. The near-infrared structural load RN at the first quartile and the third quartile were read and labeled as 25RN and 75RN, respectively. The normalized red-edge chlorophyll stress nRE and the normalized near-infrared structural stress nRN were calculated and obtained using the following formulas: ; In the formula, This represents the normalized red-edge chlorophyll stress at the i-th vegetation grid location during the t-th data collection phase. This represents the red-edge chlorophyll carrying capacity at the third quartile position during the t-th acquisition phase. This represents the red-edge chlorophyll carrying capacity at the first quartile position during the t-th acquisition phase; ; In the formula, This represents the normalized near-infrared structural stress at the i-th vegetation grid location during the t-th acquisition phase. This represents the near-infrared structural load at the third quartile position during the t-th acquisition phase. This represents the near-infrared structural load capacity at the first quartile position during the t-th acquisition phase; The vegetation heavy metal stress index I was obtained by combining the normalized red-edge chlorophyll stress value nRE and the normalized near-infrared structural stress value nRN, as shown in the following formula: ; In the formula, This represents the vegetation heavy metal stress index at the i-th vegetation grid location during the t-th data collection phase. For the same vegetation grid location, the vegetation heavy metal stress index I is arranged sequentially according to the order of the collection time phases to form the vegetation heavy metal stress time phase sequence for the current vegetation grid location; the same processing is performed on all vegetation grid locations to obtain the vegetation heavy metal stress time phase sequence for the entire area surrounding the enterprise.

3. The method for identifying soil pollution around enterprises based on vegetation heavy metal stress according to claim 2, characterized in that: S2 includes S21. S21 reads the elevation of the current vegetation raster position and the elevation of each of the eight neighboring vegetation raster positions; takes the adjacent vegetation raster positions with lower elevations than the current vegetation raster position as the descending receiving position of the current vegetation raster position, and establishes an elevation descending connection with the current vegetation raster position as the high end and the descending receiving position as the low end. For each elevation drop connection, the elevation drop rate Gx is calculated based on the elevation difference between the high and low ends and the grid center distance, using the following formula: ; In the formula, This represents the elevation of the high-level vegetation grid position connected by the xth elevation descent link. Dx represents the elevation of the low-end vegetation grid position of the xth elevation descent connection, and Dx represents the distance between the centers of the high-end and low-end grids. Read the vegetation heavy metal stress time-phase sequence of the xth elevation descent connection between the high and low ends: The high-order sequences are as follows: ; The low-order sequences are as follows: Where m represents the number of acquisition phases; Keep the high-end vegetation heavy metal stress time phase sequence position unchanged, and starting from 0 shifts in the collection time phase, shift the low-end vegetation heavy metal stress time phase sequence successively along the direction of increasing collection time phase. When the data collection phases are shifted by q, the vegetation heavy metal stress indices from the 1st to the mqth indices at the higher end are paired with the vegetation heavy metal stress indices from the (1+q)th to the mth indices at the lower end, and the shift cross-correlation coefficient between the two sequences is calculated using Pearson cross-correlation. ; Read the displacement phase number q with the largest cross-correlation coefficient among all displacement positions, and use the current displacement phase number as the stress continuity time difference Tx of the x-th elevation descent connection, as shown in the formula: ; In the formula, Indicates the search for The maximum value of q is obtained, and the cross-correlation coefficient of the current shift position is denoted as q. .

4. The method for identifying soil pollution around enterprises based on vegetation heavy metal stress according to claim 3, characterized in that: S2 also includes S22. S22 reads the elevation descent rate Gx of each elevation descent connection formed outward from the same high-level vegetation grid position, and calculates the terrain distribution flow LGx of the current connection as the ratio of the elevation descent rate of the current elevation descent connection to the sum of the elevation descent rates of all connections at the same high level. The formula is as follows: ; In the formula, The elevation decrease rate represents the r-th elevation decrease connection starting from the same high-level vegetation grid position, and n represents the number of elevation decrease connections formed starting from the current high-level vegetation grid position. Cross-correlation coefficients of shift Normalization is performed to convert it to the range of 0 to 1, and the cross-temporal propagation length is compressed using the stress continuity time difference Tx. The stress phase continuity amount KW is then calculated using the following formula: ; In the formula, This represents the stress phase continuity of the xth elevation descent connection; Based on the stress phase continuity amount KW, continuous connection is made at the high and low ends of each elevation descent connection: When the low end of the xth elevation descent connection and the high end of another elevation descent connection belong to the same vegetation grid position, the two connections will be continuously connected along the elevation descent direction. When the same vegetation grid location forms an elevation descent connection to multiple low-level ends, it extends to each low-level end to form a continuing branch; at the same time, the stress continuity time difference Tx of each connection is accumulated step by step along the current branch, and the stress phase continuity amount KW of each connection is attached to the current connection until it is connected to a location where there is no next low-level vegetation grid location, forming a terrain stress phase continuity chain.

5. The method for identifying soil pollution around enterprises based on vegetation heavy metal stress according to claim 4, characterized in that: S3 includes S31. S31 reads the terrain stress phase continuation chain, finds the end vegetation grid position in the chain where there is no next low-end vegetation grid position, takes the vegetation heavy metal stress time phase sequence of the current end vegetation grid position as the reverse transmission starting amount, and transmits it to the enterprise boundary level by level in the opposite direction of the terrain stress phase continuation chain. By advancing the low-end vegetation heavy metal stress time-phase sequence by Tx acquisition phases, the vegetation heavy metal stress index at the (t+Tx)th acquisition phase of the low end participates in the reverse transmission at the (t)th acquisition phase of the high end, thus obtaining the reverse phase reset amount of the current connection, as shown in the following formula: ; In the formula, This represents the reverse phase reset amount of the t-th acquisition phase of the x-th elevation descent connection. This represents the amount to be transmitted at the lower end of the current connection during the (t+Tx)th acquisition phase; When multiple low-level vegetation grids converge in reverse along different connections to the same high-level vegetation grid location, the reverse transmission amount is compressed according to the proportion of the current stress phase carrying capacity KW in the total stress phase carrying capacity KW of all converging connections, as shown in the following formula: ; In the formula, This represents the amount of pressure-cohesion transfer formed at the current high-level vegetation grid location during the t-th acquisition phase, and nc represents the number of connections flowing back into the current high-level location. This represents the stress phase continuity of the c-th connection that flows in reverse into the same high-end vegetation grid position; After the reverse propagation reaches the enterprise boundary, a fixed starting point on the enterprise boundary is used as the starting point of the arc length. The arc length is continuously calculated along the enterprise boundary, and the position of the arc length on the enterprise boundary is defined. By accumulating the compression transfer amounts reaching the same enterprise boundary arc length position and the same acquisition time phase, the reverse source repolymerization value at the enterprise boundary arc length position s at the acquisition time phase t is obtained. ; The inverse source back-aggregate is formed by the inverse source back-aggregate values ​​at each enterprise boundary arc length position at each acquisition time phase, and the inverse source propagation path from which each inverse source back-aggregate value originates.

6. The method for identifying soil pollution around enterprises based on vegetation heavy metal stress according to claim 5, characterized in that: S3 also includes S32. S32, based on the reverse source back-convergence of enterprise boundary stress, obtains the back-convergence proportion of enterprise boundary arc length position s at acquisition time t by proportionalizing the reverse source back-convergence values ​​of all enterprise boundary arc length positions at each acquisition time phase and the current acquisition time phase. : Without moving the enterprise boundary convergence distribution at any acquisition time phase, the convergence percentage at the same enterprise boundary arc length position across all acquisition time phases is geometrically aggregated and accumulated along the entire enterprise boundary, as follows: ; In the formula, L represents the isotopic reagglomeration amount, L represents the number of enterprise boundary arc length positions, and m represents the number of sampling phases. Maintain the re-convergence percentage of the first set phase With the spatial position fixed, the enterprise boundary is used as the long axis of the annular arc connecting the beginning and end, and the second to the mth acquisition phases are cyclically shifted respectively; For the h-th misalignment, the t-th acquisition phase is moved (t-1)h arc lengths along the enterprise boundary, and the portion that exceeds the end of the enterprise boundary continues to be arranged from the starting point of the enterprise boundary. For each acquisition phase after the h-th misalignment, the same geometric aggregation as the same in-situ re-aggregation amount H0 is re-executed to obtain the misaligned re-aggregation amount Hh; h is changed and the process is repeated to form multiple misaligned re-aggregation amounts, and the median value of all misaligned re-aggregation amounts is read and denoted as HM; When the isotopic back-aggregate H0 is greater than the median value of the misaligned back-aggregate HM, it indicates that the cross-temporal aggregation in the current enterprise boundary stress reverse source back-aggregate cannot be explained by the result of maintaining the original back-aggregate form but destroying spatial isotopicity. The enterprise boundary stress reverse source back-aggregate is retained and the source-to-back processing of S4 is triggered. When the isotopic repolymerization H0 is not greater than the median value of the misaligned repolymerization HM, the heavy metal stress of the vegetation surrounding the enterprise will continue to be retained as a result of remote sensing monitoring.

7. The method for identifying soil pollution around enterprises based on vegetation heavy metal stress according to claim 6, characterized in that: S4 includes S41. S41 reads the back-aggregation occupancy of each enterprise boundary arc length position in each acquisition phase of the enterprise boundary stress reverse source back-aggregate, obtains the enterprise boundary arc length position where back-aggregation occupancy exists in all acquisition phases, and connects the continuously adjacent positions along the enterprise boundary to form a cross-phase overlapping range. Read the inverse source back-convergence value at the boundary arc length position of each enterprise within the cross-temporal overlap range and the inverse source propagation path that forms the current inverse source back-convergence value, and use the current inverse source back-convergence value as the back-convergence starting amount; For multiple reverse source transmission paths connected at the same enterprise boundary arc length position, read the compression transmission amount when each reverse source transmission path reaches the enterprise boundary, calculate the proportion of the compression transmission amount of each reverse source transmission path to the sum of the total compression transmission amounts, and allocate the return starting amount using the current proportion. The reverse source transmission path is transmitted step by step from the enterprise boundary to the vegetation grid position. When the return amount reaches the vegetation grid position where multiple reverse source transmission paths converge, the reverse phase reset amount and the stress phase continuity amount KW formed by each connection are read respectively. The ratio of the product of the two to the sum of all input products at the current vegetation grid position is used as the return allocation ratio of the current connection. The return amount reaching the current vegetation grid position is then transmitted to each low-end vegetation grid position.

8. The method for identifying soil pollution around enterprises based on vegetation heavy metal stress according to claim 7, characterized in that: S4 also includes S42, which is the return amount allocated to each elevation descent connection by S41, divided by the current connection's stress phase continuity amount KW, to restore the stress value before the current connection's low end participates in the reverse transmission. Based on the current stress continuity time difference Tx, the recovered stress value is shifted by Tx acquisition time phases in the direction of increasing acquisition time phase, so that it returns to the original acquisition time phase; Process continuously along each reverse source propagation path until returning to each vegetation grid location that forms the current enterprise boundary stress reverse source backflow body. Record the return result obtained at the i-th vegetation grid location at the t-th acquisition phase as the source-to-return stress quantity. ; Read the vegetation heavy metal stress index at the same vegetation grid location and the same collection time phase, calculate the degree of closure between the total source-to-back stress amount and the vegetation heavy metal stress index, and obtain the enterprise's source-to-pollution closure amount Cre, as shown in the following formula: In the formula, M represents the number of vegetation grid locations involved in the formation of the overlapping temporal range.

9. The method for identifying soil pollution around enterprises based on vegetation heavy metal stress according to claim 8, characterized in that: S5 includes S51. S51 reads the cross-temporal overlap range used by the enterprise source to pollution closure quantity Cre, and obtains the number of continuous enterprise boundary arc length positions included in the current cross-temporal overlap range. Keeping the current number of arc length positions and the continuous arrangement order between each arc length position unchanged, starting from the first enterprise boundary arc length position after the current cross-phase overlap range, move one enterprise boundary arc length position clockwise along the enterprise boundary each time to form the 1st, 2nd and up to the gth shift range. When the displacement range exceeds the end of the enterprise boundary, continue arranging from the starting point of the arc length of the enterprise boundary, and exclude any displacement that is exactly the same as the original overlapping range across time phases. For the g-th shift range, the inverse source back-convergence value at the boundary arc length of each enterprise within the current range is read, along with the elevation descent connections that were sequentially passed through when the current inverse source back-convergence value was formed. Using each inverse source back-convergence value as the starting amount for the back-convergence, at locations with multiple elevation descent connections, the back-convergence amount is allocated according to the original contribution ratio when each connection forms the compression transfer amount. The numerical effect of the stress phase continuity amount KW is removed from the back-convergence amount allocated to each elevation descent connection. Then, based on the stress continuity time difference Tx, it is restored to the original acquisition time phase along the direction of increasing acquisition time phase, and processed step-by-step to each vegetation grid location to obtain the source-to-back back-convergence stress amount at each vegetation grid location under the g-th shift. ; Read the vegetation heavy metal stress index of each vegetation grid position at the same acquisition time phase, and calculate the g-th migration source to pollution closure amount gCre.

10. The method for identifying soil pollution around enterprises based on vegetation heavy metal stress according to claim 9, characterized in that: S5 also includes S52, where all the shift source to contamination closure quantities formed by S51 are arranged from small to large, and the shift source to contamination closure quantity at the upper quartile position is read as the fault source closure reference quantity QCre; The enterprise's pollution source closure quantity Cre is compared with the misdirection closure reference quantity QCre to obtain the assessment results: When the enterprise source-to-pollution closure amount Cre is not less than the mis-source closure reference amount QCre, it means that the recovery degree of the source-to-pollution return result formed by the original cross-temporal overlap range to the vegetation heavy metal stress temporal sequence exceeds most of the enterprise boundary shift positions. The location of each vegetation grid participating in the calculation of the enterprise source-to-pollution closure amount Cre is determined as the current remote sensing identification range of soil pollution around the enterprise, and the cross-temporal overlap range is determined as the pollution source-to-pollution boundary range. The current enterprise source-to-pollution identification result is output. When the enterprise's source-to-pollution closure amount Cre is less than the mis-source closure reference amount QCre, it means that after migrating the cross-temporal overlap range to other locations at the enterprise boundary, a return recovery result can still be formed. The current enterprise has not formed a source-to-soil pollution identification result, and the vegetation heavy metal stress temporal sequence is retained as subsequent remote sensing monitoring data.