A soil environment monitoring method and system based on multi-stage sampling

Abstract

This invention relates to the field of environmental monitoring, and more particularly to a soil environmental monitoring method and system based on multi-level sampling. The soil environmental monitoring method provided by this invention includes the following steps: acquiring a GIS model of a target area; based on the GIS model of the target area, deploying a first-level data acquisition module within the target area at a macro scale, and using the first-level data acquired by the first-level data acquisition module to identify at least one abnormal sub-area within the target area; based on the GIS model of the abnormal sub-area, deploying a second-level data acquisition module within the abnormal sub-area at a meso scale, and using the second-level data acquired by the second-level data acquisition module to identify at least one abnormal point within the abnormal sub-area; based on the GIS model of the abnormal point, deploying a third-level data acquisition module at each abnormal point at a micro scale, and using the third-level data acquired by each third-level data acquisition module to identify the soil condition at each abnormal point.

Description

Technical Field

[0001] This invention relates to the field of monitoring, and in particular to a method and system for monitoring soil environment based on multi-level sampling. Background Technology

[0002] With the rapid advancement of urbanization and industrialization, heavy metal pollution, persistent organic toxins, and the accumulation of excessive nutrients such as nitrogen and phosphorus are constantly eroding the soil environment of farmland, industrial parks, and construction sites, posing potential ecological and health risks. Therefore, there is an urgent need for a multi-level sampling soil environmental monitoring method and system that can combine large-scale screening with precise local diagnosis to achieve efficient, intelligent, and dynamic pollution identification and assessment. Summary of the Invention

[0003] This invention provides a soil environmental monitoring method based on multi-level sampling, the method comprising the following steps:

[0004] Obtain the GIS model of the target area;

[0005] Based on the GIS model of the target area, a first-level data acquisition module is deployed in the target area at a macro scale, and the first-level data collected by the first-level data acquisition module is used to identify at least one abnormal sub-region in the target area.

[0006] Based on the GIS model of the anomalous sub-region, second-level data acquisition modules are deployed in the anomalous sub-regions at the meso-scale, and the second-level data collected by the second-level data acquisition modules is used to identify at least one anomalous point in the anomalous sub-region.

[0007] Based on the GIS model of anomaly points, a third-level data acquisition module is deployed at each anomaly point at a micro scale, and the third-level data collected by each third-level data acquisition module is used to identify the soil conditions at each anomaly point.

[0008] In this embodiment or other embodiments, the soil environmental monitoring method based on multi-level sampling provided by the present invention,

[0009] The carrier of the first-level data acquisition module is a space platform, and the first-level data is used to assess the soil environmental status of the target area.

[0010] The second-level data acquisition module is carried by an aerial platform, and the second-level data is used to assess the soil environmental status of abnormal sub-regions.

[0011] The third-level data acquisition module is carried by a ground platform, and the third-level data is used to assess the soil conditions at abnormal locations.

[0012] In this embodiment or some other embodiments, based on the GIS model, the first-level data acquisition module is deployed in the target area at a macro scale, including the following steps:

[0013] Based on the GIS model of the target area, its horizontal spatial region is divided at the first resolution to obtain several first-level sub-regions.

[0014] In this embodiment or some other embodiments, the step of identifying at least one abnormal sub-region within the target area using the first-level data collected by the first-level data acquisition module includes the following steps:

[0015] Based on first-level data from a single source, identify at least one anomalous sub-region within the target area, where "single source" refers to the first-level data being unique to the first-level target data; or

[0016] Based on multi-source first-level data, at least one abnormal sub-region within the target area is identified, wherein the multi-source refers to the first-level data including at least two types of first-level target data.

[0017] In this embodiment or some other embodiments, identifying at least one abnormal sub-region within the target area based on multi-source first-level data includes the following steps:

[0018] By utilizing each type of first-level target data from multiple sources, abnormal sub-regions within the target area are identified, and then these abnormal sub-regions are merged in the GIS model of the target area based on each type of first-level target data.

[0019] In this embodiment or some other embodiments, identifying at least one abnormal sub-region within the target area based on single-source first-level data includes the following steps:

[0020] Preprocess the single-source data of each sub-region at each sampling time to obtain the first-level dataset. In the first-level dataset, each data point is the standardized instantaneous value corresponding to the single-source data of the sub-region at a sampling time.

[0021] Using the first-level dataset, obtain at least one temporal feature vector and at least one spatial feature vector for each data point;

[0022] The anomaly score of each data point is generated by using the temporal and spatial feature vectors of each data point, and the average anomaly score of each first-level sub-region is obtained based on the anomaly score of the single-source data at each sampling time for each first-level sub-region.

[0023] Set the sliding window and the movement step size, and obtain several first-level sub-region clustering groups based on the sliding window and the movement step size;

[0024] Based on the average anomaly scores of multiple first-level sub-regions in each cluster group, a score threshold for the corresponding cluster group is generated. The score threshold minimizes the binary classification entropy of the corresponding cluster group. A global score threshold is obtained based on the score threshold of each cluster group.

[0025] Based on the global score threshold, the average anomaly score of each first-level sub-region is compared one by one, and the first-level sub-regions whose average anomaly score exceeds the global score threshold are marked as anomaly sub-regions.

[0026] In this embodiment or some other embodiments, based on the GIS model, a second-level data acquisition module is deployed in each anomaly sub-region at a meso-scale, including the following steps:

[0027] Based on the GIS model corresponding to the abnormal sub-region, its horizontal spatial region is divided at a second resolution to obtain several secondary sub-regions, where the second resolution is higher than the first resolution.

[0028] A second-level data acquisition module is deployed in each second-level sub-region to collect at least one second-level target data in the corresponding sub-region.

[0029] In this embodiment or some other embodiments, based on the second-level data collected by the second-level data acquisition module, identifying at least one abnormal point within each abnormal sub-region includes the following steps:

[0030] Based on the secondary target data and its corresponding anomaly threshold, abnormal secondary sub-regions are marked from the several secondary sub-regions, and the center position of each abnormal secondary sub-region is set as the anomaly point.

[0031] In this embodiment or some other embodiments, based on the GIS model, a third-level data acquisition module is deployed at each anomaly location at a micro scale, including the following steps:

[0032] Based on the GIS model corresponding to the anomaly points, the vertical spatial region is divided at the third resolution to obtain several soil sampling points. At each soil sampling point, a soil sample is collected using the third-level data acquisition module. The soil sample is used to assess the soil condition.

[0033] On the other hand, the soil environment monitoring system based on multi-level sampling provided by the present invention includes a processor and a memory in one embodiment of the present invention. The memory stores a computer program, and when the computer program is executed by the processor, it implements the soil environment monitoring method based on multi-level sampling provided in any of the above embodiments.

[0034] The soil environment monitoring method and system based on multi-level sampling provided by this invention have the following advantages: This invention achieves rapid screening, accurate location and high-precision profile soil data monitoring of soil environmental anomalies in target areas through multi-level sampling coordination and multi-level data fusion from macro to meso to micro levels. Attached Figure Description

[0035] Figure 1 A flowchart of a soil environmental monitoring method based on multi-level sampling provided in this embodiment of the invention;

[0036] Figure 2 This is a flowchart of anomaly sub-region identification based on single-source data provided in an embodiment of the present invention;

[0037] Figure 3 This is a schematic diagram of the system structure of the soil environmental monitoring method based on multi-level sampling provided in an embodiment of the present invention. Detailed Implementation

[0038] In the following description, specific details such as systems, structures, and techniques are set forth for illustrative purposes and not limiting, in order to provide a thorough understanding of the embodiments of this application. Those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details.

[0039] It should be noted that detailed descriptions of well-known systems, devices, circuits, and methods have been omitted in the description of this application to avoid unnecessary details from hindering the description of this application; in addition, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0040] In one embodiment, see Figure 1 , Figure 1 This is a flowchart of a soil environmental monitoring method based on multi-level sampling provided in an embodiment of the present invention; as shown below. Figure 1 As shown in the embodiments of the present invention, the soil environmental monitoring method based on multi-level sampling includes the following steps:

[0041] S01. Obtain the GIS model of the target area.

[0042] It is understood that the target area described in this invention refers to the geographical area where soil environmental monitoring is required, which can be delineated by factors such as administrative boundaries, topographic features, or suspected pollution areas.

[0043] Furthermore, the GIS model of the target area described in this invention is a comprehensive data structure that digitally represents the target area; it includes a digital elevation model (DEM), soil attribute layers, land use, and multi-source spatial information such as hydrological systems and land features; it forms a unified spatial reference framework through geocoding and vector / raster data fusion to support path planning, positioning and navigation, and data analysis of subsequent multi-level data acquisition modules.

[0044] Specifically, step S01 can be obtained for any target area's GIS model through the following steps: acquiring high-precision terrain data through UAV aerial surveying or satellite remote sensing, and importing information such as soil texture, soil organic matter, land use type, and hydrogeographic elements from the existing cadastral database to complete the construction of the regional digital elevation model and attribute layer.

[0045] In some other embodiments, the GIS model obtained in step S01 also marks obstacles that may affect monitoring (such as buildings, dense vegetation or water bodies) so that they can be avoided or focused on during path planning and sampling point selection at the meso and micro levels, thereby laying the spatial foundation for subsequent macro screening and dynamic scheduling.

[0046] S02. Based on the GIS model of the target area, deploy the first-level data acquisition module in the target area at a macro scale, and use the first-level data collected by the first-level data acquisition module to identify at least one abnormal sub-region in the target area.

[0047] It should be noted that the first-level data acquisition module of this invention is deployed on a space platform, which includes, but is not limited to, space-borne platforms such as remote sensing satellites, high-altitude balloons, or suborbital unmanned aerial vehicles.

[0048] It is understood that the space platform upon which this invention is based can conduct all-weather, large-scale continuous monitoring of target areas of tens to hundreds of square kilometers in high-altitude or near-space environments, so as to ensure real-time understanding and dynamic updating of the overall soil environment status of the region.

[0049] Furthermore, relying on the aforementioned space platform, this embodiment utilizes at least one first-level data acquisition module to acquire at least one first-level target data within the target area.

[0050] Specifically, the first-level data acquisition module of the present invention includes, but is not limited to, one or more sensing devices selected from multispectral / hyperspectral imagers, infrared thermal radiation sensors, synthetic aperture radar (SAR), and lidar (LiDAR).

[0051] Among them, the multispectral / hyperspectral imager is used to acquire reflectivity information of the Earth's surface in the visible, near-infrared and short-wave infrared bands; the infrared thermal radiation sensor is used to measure the distribution of thermal radiation temperature on the Earth's surface; the synthetic aperture radar is used to penetrate clouds and vegetation to acquire radar echo profile signals; and the lidar is used to acquire high-precision digital elevation models and vegetation canopy structure data.

[0052] Furthermore, based on the aforementioned sensing devices, multi-band reflectivity data, surface temperature field data, radar scattering coefficient and echo profile, digital elevation model and canopy height / density, as well as quantitative inversion of soil surface moisture content achieved through multi-source fusion, can be obtained as primary target data, providing reliable spatial and temporal basis for subsequent meso-level gridded fine-grained point layout and anomaly sub-region location.

[0053] In this embodiment, to better identify abnormal sub-regions within the target area, step S02 further includes the following steps:

[0054] Based on the GIS model of the target area, its horizontal spatial region is divided at the first resolution to obtain several first-level sub-regions.

[0055] Specifically, in step S02, the target area is gridded on the horizontal plane according to a fixed size, that is, the target area is divided into several non-overlapping first grid units, and each first grid unit corresponds to a first-level sub-region; furthermore, the first resolution can be equivalent to the geometric parameters of the first grid unit, such as side length, area, etc.

[0056] In one specific embodiment, for a total area of ​​approximately 12 km² 2 The agricultural experimental field, based on its GIS model, is divided into first grid units on the horizontal plane with a fixed resolution of 50m×50m in the corresponding area. The spatial characteristics inside each first grid unit are relatively uniform, which can provide a clear geometric reference for the deployment of data acquisition modules and avoid blind spots and redundant coverage.

[0057] It is understandable that, in some other specific embodiments, the setting of the first resolution can also be flexibly adjusted according to other factors, such as cost, monitoring accuracy requirements, road network conditions, etc., to adapt to different application scenarios.

[0058] In this embodiment, the identification of any abnormal sub-region is achieved by analyzing the first-level data collected by the first-level data acquisition module during a sampling period; further, the identification of at least one abnormal sub-region within the target area in step S02 includes the following steps:

[0059] Based on first-level data from a single source, identify at least one anomalous sub-region within the target area, where "single source" refers to the first-level data being a unique type of first-level target data; or

[0060] Based on multi-source first-level data, at least one abnormal sub-region within the target area is identified, wherein the multi-source refers to the first-level data including at least two types of first-level target data;

[0061] The identification of at least one anomalous sub-region within the target area based on multi-source first-level data includes the following steps:

[0062] By utilizing each type of first-level target data from multiple sources, abnormal sub-regions within the target area are identified, and then these abnormal sub-regions are merged in the GIS model of the target area based on each type of first-level target data.

[0063] In this embodiment, based on any type of primary target data, the identification process for abnormal sub-regions is as follows: Figure 2 As shown:

[0064] S021. Preprocess the single-source data of each level sub-region at each sampling time to obtain the first-level dataset. In the first-level dataset, each data point is the standardized instantaneous value corresponding to the single-source data of the next level sub-region at a sampling time.

[0065] It is understood that the single-source data mentioned in step S021 is a type of first-level target data.

[0066] Furthermore, the preprocessing described in step S021 includes, but is not limited to, noise filtering, missing value step size, and standardization.

[0067] Specifically, the standardization of single-source data for any first-level sub-region satisfies the following formula: ,in, The standardized instantaneous value of the next-level sub-region i at sampling time t. This refers to single-source data from the next level sub-region i at sampling time t. This represents the average value of single-source data in sub-region i within the sampling period. This represents the standard deviation of the single-source data for the first-level sub-region i within the sampling period.

[0068] S022. Using the first-level dataset, obtain at least one time feature vector and at least one spatial feature vector for each data point.

[0069] Furthermore, the time feature vector based on each data point includes the standardized instantaneous value (i.e., the data itself), the moving average of the standardized instantaneous value within the sampling period, the difference between the standardized instantaneous value and the standardized instantaneous value at the previous sampling time, and the fluctuation range of the standardized instantaneous value within the sampling period. It should be noted that in some other embodiments, the time feature vector may also include other time-series feature statistics.

[0070] Furthermore, the spatial feature vector based on each data point includes the neighborhood average standardized instantaneous value and the maximum difference of the neighborhood standardized instantaneous values; it should be noted that in some other embodiments, the spatial feature vector may also include other spatial feature statistics.

[0071] S023. Using the time feature vector and spatial feature vector of each data point, generate the anomaly score of the corresponding data, and obtain the average anomaly score of the corresponding first-level sub-region based on the anomaly score of the single source data at each sampling time for each first-level sub-region.

[0072] Furthermore, the outlier score for any data point is obtained by a weighted sum of the multidimensional Mahalanobis distance score and the vector autoregression residual score, i.e. , This is the fusion weight, used to balance the weights of the multidimensional Mahalanobis distance score and the vector autoregression residual score. Its specific parameter values ​​can be adjusted according to actual circumstances. For multidimensional Mahalanobis distance scoring, This is the score for the vector autoregression residuals.

[0073] Specifically, the multidimensional Mahalanobis distance score for any data is obtained through the following calculation model: ,in, The multidimensional Mahalanobis distance score corresponds to the standardized instantaneous value of the next-level sub-region i at sampling time t. The total feature vector is composed of the temporal and spatial feature vectors of the next-level sub-region i at sampling time t. The value of m is the total number of time feature vectors and spatial feature vectors. , This represents the vector composed of the empirical means corresponding to the elements in the total eigenvector. For vectors transpose, , Representing vectors The covariance matrix.

[0074] It should be noted that the larger the multidimensional Mahalanobis distance score, the more the overall performance of the data (including trends, fluctuations, and neighborhood differences) in the corresponding first-level sub-region deviates from the normal performance in history.

[0075] Specifically, the vector autoregression residual score for any data is obtained through the following calculation model: , ,in, , , and The prediction parameter matrix and prediction parameter vector are obtained by fitting the least squares method based on the standardized instantaneous values ​​of the first-level sub-region i before sampling time t. This represents the total feature vector composed of the temporal and spatial feature vectors of the next-level sub-region i at sampling time t-1. This represents the total feature prediction vector composed of the temporal and spatial feature vectors of the next-level sub-region i at sampling time t. This represents the second norm between the total feature vector and the total feature prediction vector of the next level sub-region i at sampling time t-1. This is the corresponding vector autoregression residual score.

[0076] It should be noted that the larger the vector autoregression residual score, the more serious the deviation between the data in the corresponding first-level sub-region and the "normal" evolution model based on the past day, which may correspond to sudden anomalies.

[0077] Furthermore, the average anomaly score of any first-level sub-region is the average of the anomaly scores of that first-level sub-region at different sampling times within the sampling period.

[0078] S024. Set the sliding window and the movement step size, and obtain several first-level sub-region clustering groups based on the sliding window and the movement step size.

[0079] It is understandable that the sliding window is a fixed-size area that can include at least one first-level sub-region, and the movement step size is the number of grids the sliding window moves within the grid area.

[0080] Furthermore, the sliding window and the step size can be set according to the actual situation, thereby balancing computational efficiency and result precision while maintaining the local threshold adaptability, thus efficiently and accurately identifying abnormal first-level sub-regions.

[0081] S025. Based on the average anomaly scores of multiple first-level sub-regions in each cluster group, generate a score threshold for the corresponding cluster group. The score threshold minimizes the binary classification entropy of the corresponding cluster group. Based on the score threshold of each cluster group, obtain a global score threshold.

[0082] Furthermore, for the average anomaly score of multiple first-level sub-regions in each cluster group, different score thresholds are tried one by one until the optimal score threshold that minimizes the binary classification entropy between the abnormal first-level sub-regions and the normal first-level sub-regions is found.

[0083] Specifically, the score threshold for any cluster group is obtained through the following objective model: , , , ,in, The table represents the total number of first-level sub-regions in the k-th cluster group. This represents the average anomaly score of the j-th first-level sub-region within the k-th cluster group. and These represent the criteria based on the score threshold. The percentage of first-level sub-regions judged as "normal" or "abnormal" Indicates based on score threshold The obtained binary classification entropy.

[0084] Furthermore, the global score threshold is the average of the best score thresholds for different cluster groups. While taking into account the local optimal discrimination effect within each cluster group, it can provide a unified anomaly judgment standard for the entire target area.

[0085] S026. Based on the global score threshold, compare the average abnormal score of each first-level sub-region one by one, and mark the first-level sub-regions whose average abnormal score exceeds the global score threshold as abnormal sub-regions.

[0086] Furthermore, when multiple anomalous sub-regions are spatially continuous, these spatially adjacent anomalous sub-regions are aggregated into a single integrated anomalous sub-region. This ensures that complete and connected contamination patches can be uniformly deployed and sampled in a single operation, thereby improving sampling efficiency, avoiding redundant coverage, and ensuring the integrity of the anomalous region boundaries.

[0087] S03. Based on the GIS model of the abnormal sub-region, deploy second-level data acquisition modules in the abnormal sub-region at the meso-scale, and use the second-level data collected by the second-level data acquisition modules to identify at least one abnormal point in the abnormal sub-region.

[0088] In this embodiment, the second-level data acquisition module is deployed on an aviation platform, which includes, but is not limited to, multi-rotor drones, fixed-wing drones, or manned helicopters.

[0089] Understandably, aerial platforms are capable of conducting flexible and mobile grid-based flight inspections of identified anomalous sub-regions within a height range of tens to hundreds of meters, ensuring high-resolution, low-latency monitoring of the near-field environment of hotspot areas.

[0090] Furthermore, this embodiment incorporates at least one second-level data acquisition module on an aviation platform to achieve refined acquisition of key chemical and physical parameters within the abnormal sub-region.

[0091] Specifically, the second-level data acquisition module can be one or a combination of a portable multi-parameter electrochemical sensor array, a portable X-ray fluorescence spectrometer (XRF), a handheld near-infrared spectrometer, or a high-resolution visible light camera: the electrochemical sensor array is used to measure the pH, conductivity, and redox potential (ORP) of the soil surface layer online; the portable XRF is used for on-site qualitative or semi-quantitative analysis of heavy metal elements (such as Cd, Pb, As, and Hg); the near-infrared spectrometer is used to capture reflectance in the 400–2500 nm band to estimate soil organic matter content and moisture content; and the high-resolution camera is used to acquire images of landform and vegetation cover.

[0092] Based on the aforementioned sensing devices, this embodiment can obtain secondary target data such as soil surface pH / conductivity / ORP data, semi-quantitative heavy metal concentration data, near-infrared spectral reflectance data, and high-precision surface image data, providing accurate positioning and prediction basis for subsequent micro-level in-depth profile sampling and risk assessment.

[0093] Furthermore, to better achieve data collection in the abnormal sub-region, step S03 also includes the following steps:

[0094] Based on the GIS model corresponding to the abnormal sub-region, its horizontal spatial region is divided at a second resolution to obtain several secondary sub-regions, where the second resolution is higher than the first resolution.

[0095] A second-level data acquisition module is deployed in each second-level sub-region to collect at least one second-level target data in the corresponding sub-region.

[0096] In this embodiment, based on the second-level data collected by the second-level data acquisition module, at least one abnormal point is identified in each abnormal sub-region, including the following steps:

[0097] Based on the secondary target data and its corresponding anomaly threshold, abnormal secondary sub-regions are marked from the several secondary sub-regions, and the center position of each abnormal secondary sub-region is set as the anomaly point.

[0098] It should be noted that the abnormal threshold mentioned in this embodiment refers to the pre-set critical value that can distinguish between the normal range and the abnormal range for various monitoring indicators (such as soil surface pH, electrical conductivity, ORP, semi-quantitative concentration of heavy metals, near-infrared reflectance, etc.) in the second-level data, and is used to mark whether there are abnormal signals in the second-level sub-region.

[0099] To ensure the scientific rigor and reliability of the assessment, the abnormal threshold can be set with reference to the empirical statistical values ​​of the limits for specific pollutants or physicochemical indicators in national or local soil environmental quality regulations and industry technical specifications. For example, the reasonable range of pH (6.5 to 8.5), the content of heavy metal cadmium (Cd) not exceeding 0.3 mg / kg, and the conductivity not exceeding 400 μS / cm can be used as the upper and lower limits of the corresponding indicators for abnormality. When the measured value in a certain secondary sub-region exceeds the above standards, it can be determined that there is a pollution risk in that region.

[0100] S04. Based on the GIS model of anomaly points, a third-level data acquisition module is deployed at each anomaly point at a micro scale, and the third-level data collected by each third-level data acquisition module is used to identify the soil conditions at each anomaly point.

[0101] The third-level data acquisition module described in this embodiment is deployed on a ground platform, which includes, but is not limited to, automated drilling vehicles, wheeled or tracked unmanned ground vehicles, and is equipped with a core sampling device and a sample preservation and preliminary processing unit.

[0102] Understandably, ground platforms can perform automated, multi-depth stratified sampling of anomaly points within a scale of centimeters to meters to ensure accurate reproduction and high-precision detection of soil profile structure and pollutants at different depths.

[0103] Furthermore, this embodiment is equipped with a core sampling system, an automatic stratified sample dispenser, and field sensors on a ground platform: the core sampling system is used to automatically obtain soil core samples at preset depths such as 0–10 cm, 10–30 cm, and 30–50 cm; the stratified sample dispenser is used to number and seal core samples at different depths in real time; and the field sensors (such as portable soil moisture and temperature probes) are used to measure the real-time moisture content and temperature of the sample points.

[0104] Based on the above equipment, this embodiment can obtain stratified soil samples, in-situ moisture / temperature data, and on-site early warning indicators, providing high-quality, traceable samples and data support for subsequent laboratory tests such as heavy metal quantification (ICP-MS), organic pollutant analysis (GC-MS), total organic carbon (TOC) determination, and ecological enzyme activity detection.

[0105] In one embodiment, based on the soil environmental monitoring method based on multi-level sampling provided in the above embodiments, a soil environmental monitoring system based on multi-level sampling is also provided in the embodiment, which includes a processor and a memory. The memory stores a computer program, and when the computer program is executed by the processor, it implements the soil environmental monitoring method based on multi-level sampling provided in any of the above embodiments.

[0106] Specifically, the processor provided in this embodiment is connected to the first-level data acquisition module, the second-level data acquisition module, and the third-level data acquisition module deployed in the above method to obtain the first-level data, the second-level data, and the third-level data collected by the corresponding data acquisition modules; further, based on the first-level data, the second-level data, and the third-level data, the identification of abnormal sub-regions, the location of abnormal points, and the acquisition of soil samples at the abnormal points are realized.

[0107] The above description is merely an illustrative example of the present invention in a specific implementation process and is not intended to limit the scope of protection of the present invention. All equivalent substitutions or changes to the technical solutions described in the spirit and claims of the present invention should be included in the scope of protection of the present invention.

Claims

1. A soil environmental monitoring method based on multi-level sampling, characterized in that, Includes the following steps: Obtain the GIS model of the target area; Based on the GIS model of the target area, a first-level data acquisition module is deployed in the target area at a macro scale, and the first-level data collected by the first-level data acquisition module is used to identify at least one abnormal sub-region in the target area. Based on the GIS model of the anomalous sub-region, second-level data acquisition modules are deployed in the anomalous sub-regions at the meso-scale, and the second-level data collected by the second-level data acquisition modules is used to identify at least one anomalous point in the anomalous sub-region. Based on the GIS model of anomaly points, a third-level data acquisition module is deployed at each anomaly point at a micro-scale. The third-level data collected by each third-level data acquisition module is used to identify the soil conditions at each anomaly point. Based on the first-level data from a single source, at least one abnormal sub-region within the target area is identified. The single source refers to the first-level data being unique first-level target data. The single-source data of each first-level sub-region at each sampling time is preprocessed to obtain the first-level dataset. Using the first-level dataset, obtain at least one temporal feature vector and at least one spatial feature vector for each data point; The anomaly score of each data point is generated by using the temporal and spatial feature vectors of each data point, and the average anomaly score of each first-level sub-region is obtained based on the anomaly score of the single-source data at each sampling time for each first-level sub-region. Set the sliding window and the movement step size, and obtain several first-level sub-region clustering groups based on the sliding window and the movement step size; Based on the average anomaly scores of multiple first-level sub-regions in each cluster group, a score threshold for the corresponding cluster group is generated. The score threshold minimizes the binary classification entropy of the corresponding cluster group. A global score threshold is obtained based on the score threshold of each cluster group. Based on the global score threshold, the average anomaly score of each first-level sub-region is compared one by one, and the first-level sub-regions whose average anomaly score exceeds the global score threshold are marked as anomaly sub-regions.

2. The soil environmental monitoring method based on multi-level sampling according to claim 1, characterized in that: The carrier of the first-level data acquisition module is a space platform, and the first-level data is used to assess the soil environmental status of the target area. The second-level data acquisition module is carried by an aerial platform, and the second-level data is used to assess the soil environmental status of abnormal sub-regions. The third-level data acquisition module is carried by a ground platform, and the third-level data is used to assess the soil conditions at abnormal locations.

3. The soil environmental monitoring method based on multi-level sampling according to claim 2, characterized in that, Based on the GIS model, the first-level data acquisition module is deployed in the target area at a macro scale, including the following steps: Based on the GIS model of the target area, its horizontal spatial region is divided at the first resolution to obtain several first-level sub-regions.

4. The soil environmental monitoring method based on multi-level sampling according to claim 3, characterized in that, The step of identifying at least one abnormal sub-region within the target area using the first-level data acquired by the first-level data acquisition module includes the following steps: Based on first-level data from a single source, identify at least one anomalous sub-region within the target area, where "single source" refers to the first-level data being uniquely classified as first-level target data; or Based on multi-source first-level data, at least one abnormal sub-region within the target area is identified, wherein the multi-source refers to the first-level data including at least two types of first-level target data.

5. The soil environmental monitoring method based on multi-level sampling according to claim 4, characterized in that, The identification of at least one anomalous sub-region within the target area based on multi-source first-level data includes the following steps: By utilizing each type of first-level target data from multiple sources, abnormal sub-regions within the target area are identified, and then these abnormal sub-regions are merged in the GIS model of the target area based on each type of first-level target data.

6. The soil environmental monitoring method based on multi-level sampling according to claim 3, characterized in that, Based on the GIS model, a second-level data acquisition module is deployed in each anomaly sub-region at the meso-scale, including the following steps: Based on the GIS model corresponding to the abnormal sub-region, its horizontal spatial region is divided at a second resolution to obtain several secondary sub-regions, where the second resolution is higher than the first resolution. A second-level data acquisition module is deployed in each second-level sub-region to collect at least one second-level target data in the corresponding sub-region.

7. The soil environmental monitoring method based on multi-level sampling according to claim 6, characterized in that, Based on the second-level data collected by the second-level data acquisition module, at least one abnormal point is identified in each abnormal sub-region, including the following steps: Based on the secondary target data and its corresponding anomaly threshold, abnormal secondary sub-regions are marked from the several secondary sub-regions, and the center position of each abnormal secondary sub-region is set as the anomaly point.

8. The soil environmental monitoring method based on multi-level sampling according to claim 2, characterized in that, Based on the GIS model, a third-level data acquisition module is deployed at each anomaly location at a micro scale, including the following steps: Based on the GIS model corresponding to the anomaly points, the vertical spatial region is divided at the third resolution to obtain several soil sampling points. At each soil sampling point, a soil sample is collected using the third-level data acquisition module. The soil sample is used to assess the soil condition.

9. A soil environmental monitoring system based on multi-level sampling, characterized in that, It includes a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, it implements the soil environmental monitoring method based on multi-level sampling as described in any one of claims 1-8.

Images