Soil monitoring sampling methods, systems, devices, and media

Abstract

The application provides a soil monitoring sampling method, system, device and medium, the region to be monitored is divided into grid by spatial position and historical soil information, forming a candidate sampling unit with spatial correlation, in each candidate sampling unit, the physical parameters representing the state of the soil surface and shallow layer are collected, and a high-dimensional feature vector is generated; based on all high-dimensional feature vectors, a dynamic sampling priority scoring model reflecting the degree of spatial difference of soil is constructed, and a target sampling unit is determined; a depth stratified sampling strategy is performed on the target sampling unit, the sampling depth interval is adaptively determined, and then a representative soil sample is obtained; the soil sample is associated with the corresponding spatial position, sampling depth and soil state information, and structured soil sampling data is generated. The scheme of the application can realize the spatial-depth joint optimization of soil sampling, and accurately capture the real state and spatial variation law of the soil environment.

Description

Technical Field

[0001] This application relates to the field of soil sampling technology, and more specifically, to a soil monitoring and sampling method, system, equipment, and medium. Background Technology

[0002] Soil monitoring sampling is a fundamental technical means in environmental science, agricultural geology, and other fields. It aims to systematically collect and analyze soil samples to obtain information on their physical, chemical, and pollutant composition, in order to assess soil quality, trace pollution sources, or study ecological processes. Traditional soil sampling methods typically rely on pre-defined grids or expert experience for sampling points, collecting samples at fixed depths. However, existing technologies suffer from the following drawbacks: First, soil properties, such as pollutant distribution, moisture content, and density, exhibit high spatial heterogeneity and inhomogeneity. Fixed, uniform grid sampling strategies lack specificity, easily overlooking key pollution sites or areas of variation, resulting in insufficient spatial representativeness of the obtained samples and an inability to accurately depict the spatial pattern of pollution. Second, conventional methods often employ uniform standards for sampling depth, failing to consider the vertical differentiation characteristics of pollutants in the soil profile caused by water migration, adsorption, and desorption, thus limiting the representativeness of samples in the depth dimension. Furthermore, for large-scale monitoring areas, attempting to improve accuracy through high-density, indiscriminate sampling incurs high time and economic costs, making it impractical. Therefore, how to achieve spatial-depth joint optimization of soil sampling in order to accurately capture the true state and spatial variation patterns of the soil environment has become a challenge for the industry. Summary of the Invention

[0003] This application provides a soil monitoring and sampling method, system, equipment, and medium that can achieve spatial-depth joint optimization of soil sampling to accurately capture the true state and spatial variation patterns of the soil environment.

[0004] In a first aspect, this application provides a soil monitoring and sampling method, comprising the following steps: Obtain the spatial location and historical soil information of the area to be monitored; Based on the spatial location and historical soil information, the area to be monitored is divided into grids to form candidate sampling units with spatial correlation. Within each candidate sampling unit, physical parameters characterizing the state of the soil surface and shallow layers are collected, and high-dimensional feature vectors reflecting the differences in soil microenvironment are generated. A priority scoring model for dynamic sampling that reflects the degree of spatial difference in soil is constructed based on all high-dimensional feature vectors, and the target sampling unit is dynamically determined through the priority scoring model. A depth stratified sampling strategy is implemented for the target sampling unit, and the sampling depth range is adaptively determined according to the vertical variation trend of soil physical parameters, so as to obtain representative soil samples. The obtained soil samples are associated and labeled with their corresponding spatial location, sampling depth, and soil condition information to generate structured soil sampling data.

[0005] In conjunction with the first aspect, in one possible implementation, the area to be monitored is divided into grids based on the spatial location and historical soil information to form candidate sampling units with spatial correlation, specifically including: Based on the historical soil information, the spatial variation coefficients of key soil properties are determined, and then the grid division size is dynamically determined. Based on the area boundary defined by the spatial location and the grid division size, the area to be monitored is regularly divided into multiple continuous grid units; Each grid cell is assigned a unique identifier and its spatial geometric properties are calculated to form candidate sampling cells with spatial relationships.

[0006] In conjunction with the first aspect, in one possible implementation, physical parameters characterizing the state of the soil surface and shallow layers are collected within each candidate sampling unit, and a high-dimensional feature vector reflecting the differences in the soil microenvironment is generated. Specifically, this includes: Collect physical parameters characterizing the surface and shallow soil conditions for each candidate sampling unit; The collected physical parameters are standardized and quality verified to obtain a standardized set of physical parameters; The standardized physical parameter set corresponding to each candidate sampling unit and its associated historical soil attribute data are fused and dimensionality reduced to generate a high-dimensional feature vector that reflects the differences in soil microenvironment.

[0007] In conjunction with the first aspect, one possible implementation involves constructing a priority scoring model for dynamic sampling that reflects the degree of spatial variability in soil based on all high-dimensional feature vectors. This specifically includes: Unsupervised clustering analysis was performed based on all high-dimensional feature vectors to divide the soil microenvironment feature space into multiple categories and determine the centroid of each category. Based on the clustering results, the distance from the high-dimensional feature vector of each candidate sampling unit to the centroid of its class, as well as the minimum distance to the centroids of other classes, are calculated, thereby determining the priority scoring model for dynamic sampling that reflects the degree of spatial difference in soil.

[0008] In conjunction with the first aspect, in one possible implementation, a depth-stratified sampling strategy is performed on the target sampling unit, adaptively determining the sampling depth range based on the vertical variation trend of soil physical parameters, thereby obtaining representative soil samples. Specifically, this includes: Vertical continuous detection of soil physical parameters is carried out at the designated location of the target sampling unit to obtain the vertical variation curve reflecting the parameter changes with depth; Based on the vertical variation curve, identify the characteristic depth points where the rate of change of soil properties exceeds a preset threshold, and define the sampling depth range based on the identified characteristic depth points; Based on the defined sampling depth range, stratified borehole sampling is performed at the designated location of the target sampling unit to obtain soil samples of the corresponding depth range.

[0009] In conjunction with the first aspect, one possible implementation involves associating and labeling the obtained soil samples with their corresponding spatial location, sampling depth, and soil condition information to generate structured soil sampling data. Specifically, this includes: Generate and assign globally unique sample identifiers to the acquired soil samples; The sample identifier is digitally associated and bound with the spatial location, sampling depth range information of the corresponding target sampling unit, and soil state information collected at the target sampling unit. The associated information obtained after binding is encapsulated into a complete structured data record, thereby forming structured soil sampling data.

[0010] In conjunction with the first aspect, in one possible implementation, the physical parameters include at least the water content state, density characteristics, or electrical response characteristics.

[0011] Secondly, this application provides a soil monitoring and sampling system, comprising: The acquisition module is used to acquire the spatial location and historical soil information of the area to be monitored; The processing module is used to divide the area to be monitored into grids based on the spatial location and historical soil information, forming candidate sampling units with spatial correlation. Within each candidate sampling unit, physical parameters characterizing the state of the soil surface and shallow layers are collected, and a high-dimensional feature vector reflecting the differences in soil microenvironment is generated. The processing module is also used to construct a dynamic sampling priority scoring model that reflects the degree of spatial difference in soil based on all high-dimensional feature vectors, and to dynamically determine the target sampling unit through the priority scoring model. The processing module is also used to execute a depth stratification sampling strategy on the target sampling unit, adaptively determine the sampling depth range according to the vertical variation trend of soil physical parameters, and thus obtain representative soil samples. The execution module is used to associate and label the obtained soil samples with their corresponding spatial location, sampling depth, and soil condition information to generate structured soil sampling data.

[0012] Thirdly, this application provides a computer device, the computer device including a memory and a processor, the memory for storing a computer program, and the processor for calling and running the computer program from the memory, so that the computer device performs the above-described soil monitoring and sampling method.

[0013] Fourthly, this application provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the aforementioned soil monitoring and sampling method.

[0014] The technical solution provided in this application has the following beneficial effects: This application proposes two main methods. First, by dividing the monitoring area into grids based on spatial location and historical soil information, the continuous and complex soil region is discretized into candidate sampling units with clear spatial boundaries and adjacency relationships. This transforms the uncontrollable soil spatial structure into an analyzable object. Physical parameters characterizing the surface and shallow soil states are collected within each candidate sampling unit, and the multidimensional heterogeneous physical information is uniformly transformed into high-dimensional feature vectors. This effectively characterizes the degree of difference in the soil microenvironment between different units, providing a quantitative basis for subsequent judgment of "where significant spatial changes exist," thus supporting optimized decision-making in soil sampling at the spatial dimension from the source. Second, by constructing a dynamic sampling priority scoring model based on high-dimensional feature vectors to reflect the degree of soil spatial differences, sampling decisions are transformed from traditional fixed-point placement or experience-based judgment to a dynamic selection process based on soil state differences. This priority scoring model can highlight areas with significant soil changes and strong heterogeneity, while weakening the sampling weight of homogeneous areas and areas with redundant information. Therefore, under limited sampling conditions, priority coverage of key spatially variable areas is achieved, providing core support for accurate sampling at the "spatial level." Subsequently, within the defined target sampling unit, the sampling depth range is adaptively determined based on the vertical variation trend of soil physical parameters. This eliminates reliance on fixed strata or empirical settings for sampling depth. This process identifies inflection points in soil state changes at different depth levels, avoids oversampling of gently changing strata, and strengthens the collection efforts in areas with significant vertical differences. This ensures that the sampling depth matches the actual soil structure, providing effective assurance for the "depth dimension" in spatial-depth joint optimization. Finally, by associating and labeling soil samples with their corresponding spatial locations, sampling depths, and soil state information, structured soil sampling data is formed. Each sample possesses complete spatial and state background information. This structured data not only supports lateral comparisons between different sampling units but also longitudinal analysis at different depths within the same location. It can reconstruct the spatial-depth joint distribution characteristics of soil, thus providing a reliable data foundation for accurately revealing the true state of the soil environment and its spatial variation patterns. In summary, this scheme can achieve spatial-depth joint optimization of soil sampling to accurately capture the true state and spatial variation patterns of the soil environment. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1This is an exemplary flowchart of a soil monitoring and sampling method according to some embodiments of this application; Figure 2 This is an exemplary flowchart illustrating the generation of high-dimensional feature vectors according to some embodiments of this application; Figure 3 This is an exemplary flowchart illustrating the execution of a depth-layered sampling strategy according to some embodiments of this application; Figure 4 This is a schematic diagram of the structure of a soil monitoring and sampling system according to some embodiments of this application; Figure 5 This is a schematic diagram of the structure of a computer device for implementing a soil monitoring and sampling method according to some embodiments of this application. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0018] refer to Figure 1 The figure is an exemplary flowchart of a soil monitoring and sampling method according to some embodiments of this application. The soil monitoring and sampling method mainly includes the following steps: In step 101, the spatial location and historical soil information of the area to be monitored are obtained.

[0019] In practice, the spatial location and historical soil information of the area to be monitored can be obtained in the following way: First, a mobile terminal device equipped with a differential global positioning system or real-time dynamic positioning technology can be used to walk along the boundary of the area to be monitored and record or directly input its vertex coordinates to obtain a precise set of geographic boundary coordinates defining the area. At the same time, from an authoritative soil geographic database or past environmental survey reports, a raster format historical soil type distribution map, a vector format historical soil sampling point dataset, and its corresponding laboratory test index data that match the area are retrieved. Then, in the geographic information system software platform, the set of geographic boundary coordinates is defined as an analysis mask, and all historical soil data are spatially registered and unified to the same projected coordinate system. For discrete historical sampling point data, based on their location coordinates and test indicators, a Kriging spatial interpolation algorithm is used to generate a continuous spatial distribution raster map covering the entire monitored area. Each pixel value of the raster map represents a historical soil attribute value such as soil organic matter content, pH value, or specific heavy metal concentration. Finally, the obtained precise geographic boundary coordinate set and a series of historical soil attribute spatial distribution raster maps are used as the input basis for subsequent gridding and analysis, thereby completing the standardized acquisition and preprocessing of spatial location and historical soil information. The historical soil information includes attribute data such as historical soil type, organic matter content, pH value, heavy metal concentration, and soil texture. Other methods can also be used in other embodiments, which are not specifically limited here.

[0020] It should be noted that the spatial location in this application refers to the spatial data that quantifies the geographical extent of the area to be monitored, which is used to provide an accurate spatial framework and geometric benchmark for the soil monitoring and sampling scheme; the historical soil information in this application refers to the spatialized data that can reflect the inherent properties or past state of the soil within the area to be monitored, which is used to provide prior knowledge for understanding the background heterogeneity and spatial distribution patterns of regional soils, and serves as an important reference benchmark for dynamic grid division and judging the current changes or spatial differences in soil state.

[0021] In step 102, the area to be monitored is divided into grids based on the spatial location and historical soil information to form candidate sampling units with spatial correlation. Within each candidate sampling unit, physical parameters characterizing the state of the soil surface and shallow layers are collected, and high-dimensional feature vectors reflecting the differences in soil microenvironment are generated.

[0022] In some embodiments, the gridding of the area to be monitored based on the spatial location and historical soil information to form candidate sampling units with spatial relationships can be achieved through the following steps: Based on the historical soil information, the spatial variation coefficients of key soil properties are determined, and then the grid division size is dynamically determined. Based on the area boundary defined by the spatial location and the grid division size, the area to be monitored is regularly divided into multiple continuous grid units; Each grid cell is assigned a unique identifier and its spatial geometric properties are calculated to form candidate sampling cells with spatial relationships.

[0023] In specific implementation, the spatial variation coefficients of key soil attributes are determined based on the historical soil information, and then the grid division size is dynamically determined. This can be achieved as follows: First, the key attributes that best reflect the basic fertility and pollution migration characteristics of the soil, such as an organic matter content distribution raster map, are selected from the historical soil information as the calculation object. Then, the sliding window method in spatial statistics is used to traverse the raster map, and the ratio of the standard deviation to the mean of the pixel values ​​in each local window is calculated to obtain the variation coefficient of the central pixel point of the window. The variation coefficients of all pixel points are statistically analyzed globally, and the average value is taken as the entire monitored area. The representative spatial variation coefficient of this key attribute is determined; finally, the grid size is dynamically determined according to the preset variation coefficient-grid size mapping rule: for example, when the calculated global spatial variation coefficient is higher than 0.5, it is determined that the soil spatial heterogeneity is strong, and a smaller grid size, such as 5m×5m, is used to capture detailed variation; when the spatial variation coefficient is lower than 0.3, it is determined that the spatial homogeneity is strong, and a larger grid size, such as 20m×20m, can be used to improve the survey efficiency; the mapping rule can be based on historical experience data or pre-experiment calibration to ensure that the division scale is adapted to the spatial distribution complexity of the actual soil properties.

[0024] In specific implementation, the area to be monitored can be regularly divided into multiple continuous grid units according to the area boundary defined by the spatial location and the grid division size. This can be achieved in the following way: First, the precise geographic boundary coordinate set defined by the spatial location is constructed into a closed polygonal region in a geographic information system or computational geometry library. Then, using the minimum bounding rectangle of this polygonal region as a reference, and with the dynamically determined grid division size as a fixed step size, a regular square grid covering the entire bounding rectangle is generated starting from the lower left corner vertex of the rectangle along the latitude and longitude direction. Then, the "clipping" operation in spatial overlay analysis is used to perform intersection processing between the generated regular grid and the constructed closed polygonal region, retaining only the grids whose center point or any part falls within the polygonal region. These retained adjacent and non-overlapping grids constitute multiple continuous regular grid units covering the entire area to be monitored.

[0025] In specific implementation, assigning a unique identifier to each grid cell and calculating its spatial geometric attributes to form candidate sampling cells with spatial relationships can be achieved in the following way: For each grid cell retained in the above steps, sequentially encode it from west to east and from south to north, for example, G001, G002, ..., to generate its unique identifier; at the same time, calculate multiple spatial geometric attributes of each grid cell, for example: by calculating the centroid of its polygon geometry, obtain the coordinates of the center point of the cell, which will serve as the precise location for subsequent on-site sampling; calculate its polygon boundary to obtain the cell boundary coordinate string for spatial display and association; record the unique identifiers of its adjacent cells to form an adjacency list; finally, combine the unique identifier, center point coordinates, cell boundary coordinate string and adjacency list of each grid cell into a data structure with a complete spatial description, which represents a candidate sampling cell; the set of data structures of all cells constitutes a network with a clear spatial topological relationship, that is, it forms candidate sampling cells with spatial relationships; other methods can also be used in other embodiments, which are not limited here.

[0026] It should be noted that the spatial variation coefficient in this application refers to a statistical index that quantifies the heterogeneity or dispersion of a representative soil property in the spatial distribution of the area to be monitored. It enables the determination of grid size to adapt to the degree of spatial heterogeneity of the soil itself, thereby ensuring the scientificity and efficiency of the division scheme. The grid cell in this application refers to the basic geometric unit formed after the area to be monitored is regularly divided according to the determined grid division size. The candidate sampling cell in this application refers to a complete data object that is given a unique identity and precise spatial description information based on the grid cell. Its function is not only to identify a potential sampling location, but also to construct a network in the digital space that clearly expresses the location, shape and proximity relationship between units through its built-in spatial geometric attributes and topological relationships, thereby supporting spatial location-based analysis, screening and data association.

[0027] In some embodiments, reference Figure 2 As shown, this figure is an exemplary flowchart of generating high-dimensional feature vectors in some embodiments of this application. In this embodiment, the collection of physical parameters characterizing the state of the soil surface and shallow layers within each candidate sampling unit, and the generation of high-dimensional feature vectors reflecting the differences in the soil microenvironment, can be achieved through the following steps: In step 1021, physical parameters characterizing the surface and shallow soil conditions are collected for each candidate sampling unit; In step 1022, the collected physical parameters are standardized and quality verified to obtain a standardized set of physical parameters; In step 1023, the standardized physical parameter set corresponding to each candidate sampling unit and its associated historical soil attribute data are subjected to feature fusion and dimensionality reduction processing to generate a high-dimensional feature vector reflecting the differences in soil microenvironment.

[0028] In practice, the physical parameters of each candidate sampling unit can be collected in the following way: at a designated spatial location of each candidate sampling unit, usually its center point coordinates, a mobile soil multi-parameter measuring instrument integrating a dielectric sensor, a cone resistance sensor, and a four-electrode conductivity sensor is used to perform in-situ synchronous measurements; during operation, the sensor probe is vertically inserted into a preset shallow depth range below the soil surface, for example, 0 to 30 cm. During the insertion process, the dielectric sensor measures and records the volumetric water content of the soil in real time to characterize the water content state, the cone resistance sensor synchronously measures and records the penetration resistance of the soil to characterize the compaction characteristics, and the four-electrode conductivity sensor measures and records the apparent conductivity of the soil to characterize the electrical response characteristics; finally, for each candidate sampling unit, a set of physical parameters including the average volumetric water content, average soil compaction, and average apparent conductivity values ​​are output. The physical parameters include at least the water content state, compaction characteristics, or electrical response characteristics.

[0029] In practice, the collected physical parameters are standardized and quality-verified to obtain a standardized set of physical parameters. This can be achieved as follows: First, the raw data of all received physical parameters are quality-verified, including checking whether the data is within the sensor's measurement range and eliminating obvious outliers caused by transient interference from the equipment based on the Raida criterion or interquartile range method. Then, all valid raw data are standardized. For each physical parameter, such as volumetric water content, soil compaction, and apparent conductivity, the mean and standard deviation of the physical parameter for all candidate sampling units are calculated. The Z-score standardization method can be used, subtracting the overall mean of the parameter from the original value of the parameter for each candidate sampling unit and then dividing by the overall standard deviation to obtain its standardized value. After this step, the raw data of all physical parameters are transformed into a standardized set of physical parameters with a mean of 0 and a standard deviation of 1, thereby eliminating the differences in dimensions and numerical ranges between different parameters.

[0030] In practice, the standardized physical parameter set corresponding to each candidate sampling unit and its associated historical soil attribute data are fused and dimensionality reduced to generate a high-dimensional feature vector reflecting the differences in the soil microenvironment. This can be achieved in the following way: First, feature fusion is performed. For example, for a candidate sampling unit, three standardized physical parameter values ​​are extracted from its standardized physical parameter set. At the same time, based on the center point coordinates of the candidate sampling unit, the attribute values ​​at the corresponding coordinate positions are extracted from the previously generated historical soil attribute spatial distribution raster map, such as the organic matter content distribution map and the pH value distribution map. This forms an initial feature array containing the field physical parameters and historical soil attribute data. Historical soil attribute data includes organic matter content, pH value, heavy metal concentration, soil texture, and moisture content. Then, dimensionality reduction is performed: the initial feature arrays of all candidate sampling units in the entire monitoring area are combined into an initial feature matrix. Principal component analysis is applied to this matrix to calculate its eigenvalues ​​and eigenvectors, and the top K principal components with a cumulative variance contribution rate exceeding a preset threshold, such as 85%, are selected. Finally, the initial feature array of each candidate sampling unit is projected onto the directions of these K principal components to obtain a K-dimensional vector. This vector is the high-dimensional feature vector representing the soil microenvironmental differences of the candidate sampling unit. Other methods can also be used in other embodiments, which are not limited here.

[0031] It should be noted that the physical parameters in this application refer to the original measured values ​​characterizing the water content, compaction, and electrical properties of the soil in its current state; the standardized physical parameter set in this application refers to the data set obtained after quality control and standardization transformation of the original physical parameter data; the associated historical soil attribute data in this application refers to soil attribute values ​​such as organic matter, pH value, heavy metals, texture, and structure extracted from historical soil information corresponding to the spatial location of the specified candidate sampling unit. These values ​​are used to spatially align and associate historical background information reflecting the long-term inherent properties of the soil with physical parameters reflecting the current state; the high-dimensional feature vector in this application refers to a multi-dimensional mathematical vector generated by fusing the standardized physical parameter set with the associated historical soil attribute data and performing dimensionality reduction processing. This vector is used to compress and transform diverse and heterogeneous soil information into a comprehensive and measurable digital representation. This vector can retain and highlight the comprehensive differences in soil microenvironment between different candidate sampling units to the greatest extent with a lower dimension.

[0032] In step 103, a priority scoring model for dynamic sampling that reflects the degree of spatial difference in soil is constructed based on all high-dimensional feature vectors, and the target sampling unit is dynamically determined through the priority scoring model.

[0033] In some embodiments, constructing a priority scoring model for dynamic sampling that reflects the degree of spatial variability in soil based on all high-dimensional feature vectors can be achieved through the following steps: Unsupervised clustering analysis was performed based on all high-dimensional feature vectors to divide the soil microenvironment feature space into multiple categories and determine the centroid of each category. Based on the clustering results, the distance from the high-dimensional feature vector of each candidate sampling unit to the centroid of its class, as well as the minimum distance to the centroids of other classes, are calculated, thereby determining the priority scoring model for dynamic sampling that reflects the degree of spatial difference in soil.

[0034] In practice, unsupervised clustering analysis based on all high-dimensional feature vectors is used to divide the soil microenvironment feature space into multiple categories and determine the centroids of each category. This can be achieved as follows: First, high-dimensional feature vectors from all candidate sampling units are used as the input dataset. Then, the K-means clustering algorithm is used to perform unsupervised clustering analysis on this dataset. The number of clusters N is determined by the elbow rule or by empirical preset, for example, set to 3 to 5 categories. After executing the algorithm, all high-dimensional feature vectors are assigned to N different categories, each category representing a soil microenvironment pattern similar in the feature space. Finally, for each category obtained after the algorithm converges, the mean of all high-dimensional feature vectors belonging to that category is calculated. This mean vector is the centroid of that category, thus completing the division of the soil microenvironment feature space and determining the representative points of each category.

[0035] In practical implementation, based on the clustering results, the distance from the high-dimensional feature vector of each candidate sampling unit to the centroid of its class, and the minimum distance to the centroids of other classes, are calculated. This determines the priority scoring model for dynamic sampling, reflecting the degree of spatial variation in soil. This can be implemented as follows: First, for any candidate sampling unit, its class is determined based on its cluster label, and the Euclidean distance from its K-dimensional high-dimensional feature vector to the centroid of that class is calculated. This distance is the intra-class distance, used to measure the degree of deviation of the unit from the typical characteristics of its class. Then, the Euclidean distances from the high-dimensional feature vector of the candidate sampling unit to the centroids of the remaining (N-1) classes are calculated, and the unit is selected from these distances. The minimum value is found, and this distance is the inter-class minimum distance, which is used to measure the similarity between the unit and the nearest out-of-class feature. Finally, the core algorithm of the priority scoring model is defined as follows: for each candidate sampling unit, calculate the ratio of its intra-class distance to the inter-class minimum distance, or map these two distances into a scalar score according to specific rules. This ratio or score is used as the final priority score of the candidate sampling unit. Thus, a dynamic sampling priority scoring model based on clustering results, using distance calculation as the rule, and capable of outputting the sampling priority of each candidate sampling unit is constructed and determined. Other methods can also be used to implement this in other embodiments, which are not limited here.

[0036] It should be noted that the clustering results in this application refer to a complete dataset containing multiple category divisions, centroids of each category, and the category label of each candidate sampling unit; the distance from a high-dimensional feature vector to its category centroid in this application is a quantitative indicator used to measure the degree to which the comprehensive characteristics of a candidate sampling unit's soil microenvironment deviate from the typical characteristics of its category. It reflects the atypicality or internal heterogeneity of the sampling unit within its category; the larger the distance, the more unusual the sampling unit is, and the more unique the microenvironment information it may contain. The minimum distance to other category centroids in this application refers to a measure of... The quantitative index of similarity between the comprehensive characteristics of the soil microenvironment of each candidate sampling unit and the most similar typical characteristics of other categories is used to reflect the clarity or boundary of the sampling unit that distinguishes it from other types. The smaller the distance, the closer the characteristics of the sampling unit are to other types, and the stronger the spatial transition or mixing may be. The priority scoring model in this application refers to the algorithm rule that generates a comprehensive score based on the clustering results by integrating the calculation of intra-class distance and inter-class minimum distance. It is used to map the positional relationship of each candidate sampling unit in the feature space into a comparable and quantifiable priority value.

[0037] In specific implementation, the dynamic determination of target sampling units through the priority scoring model can be achieved in the following way: First, the priority scoring model is applied to all candidate sampling units, and the priority score corresponding to each unit is calculated and output sequentially; then, all candidate sampling units are sorted in descending order according to their priority scores to generate a unit sequence sorted from high to low sampling priority; finally, according to preset sampling constraints, such as the maximum number of sampling points M planned for this task, or a minimum priority score threshold S, units are selected sequentially from the top of the sorted sequence until the number of selected units reaches M or the priority score of the selected units is lower than the threshold S. At this time, these selected candidate sampling units are dynamically determined as target sampling units that need to be sampled for depth stratification in the current monitoring period; other methods can also be used in other embodiments, which are not limited here.

[0038] It should be noted that the target sampling unit in this application refers to a subset of units dynamically selected from all candidate sampling units based on a priority scoring model for actual deep stratified sampling. This ensures that sampling resources such as time and manpower are prioritized for locations that are most representative, unique, or critical in terms of soil spatial differences, thereby enabling the final soil samples to most efficiently characterize the soil microenvironment heterogeneity of the entire monitored area.

[0039] In step 104, a depth stratification sampling strategy is executed on the target sampling unit. The sampling depth range is adaptively determined according to the vertical variation trend of soil physical parameters, thereby obtaining representative soil samples.

[0040] In some embodiments, reference Figure 3 As shown, this figure is an exemplary flowchart of implementing a depth stratification sampling strategy in some embodiments of this application. The depth stratification sampling strategy is implemented on the target sampling unit, and the sampling depth range is adaptively determined based on the vertical variation trend of soil physical parameters to obtain representative soil samples. This can be achieved through the following steps: Vertical continuous detection of soil physical parameters is carried out at the designated location of the target sampling unit to obtain the vertical variation curve reflecting the parameter changes with depth; Based on the vertical variation curve, identify the characteristic depth points where the rate of change of soil properties exceeds a preset threshold, and define the sampling depth range based on the identified characteristic depth points; Based on the defined sampling depth range, stratified borehole sampling is performed at the designated location of the target sampling unit to obtain soil samples of the corresponding depth range.

[0041] In practice, the vertical continuous detection of soil physical parameters at a designated location in the target sampling unit, and the acquisition of vertical variation curves reflecting the parameters with depth, can be achieved in the following way: A soil profile conductivity meter based on the principle of electromagnetic induction is deployed at the center point of the target sampling unit, or a push-in profile probe rod integrating multiple sensors is used. During operation, the detection device is pressed down from the ground surface at a fixed depth interval, for example, every 5 centimeters, to a preset maximum exploration depth, such as 1.5 meters. At each interval, the volumetric water content, penetration resistance, and apparent conductivity of the soil are measured and recorded simultaneously. This yields a discrete data point sequence showing how physical parameters (water content, compaction characteristics, and electrical response characteristics) change with increasing depth. By performing smooth interpolation on this sequence, one or more continuous vertical variation curves are finally generated.

[0042] In specific implementation, the identification of characteristic depth points where the rate of change of soil properties exceeds a preset threshold based on the vertical variation curve, and the delineation of sampling depth intervals based on the identified characteristic depth points, can be achieved in the following way: First, select a vertical variation curve that best reflects the changes in soil layers, such as the soil apparent electrical conductivity curve, as the analysis object, and calculate the first derivative of the curve or the difference within the sliding window to quantify its rate of change with depth; then, determine the depth locations where the absolute value of the rate of change of soil properties exceeds the preset threshold, for example, the rate of change of electrical conductivity is greater than the preset threshold, as characteristic depth points where there is a sudden change in soil texture, bulk density, or composition. The preset thresholds can be set according to monitoring needs, such as conductivity change rate > 0.5 mS / m / cm and moisture content change rate > 10% / cm. Then, using the surface as the starting point, the depth range between two adjacent characteristic depth points is defined as a soil layer with relatively uniform properties. The depth range corresponding to the soil layers requiring focused study, such as suspected contaminated layers or cultivated layers determined according to the monitoring target, is defined as the sampling depth range for this sampling, for example, from 0.4 meters to 0.9 meters. The soil properties refer to the physicochemical properties of the soil, such as moisture content, density, conductivity, texture, structure, and porosity, which will not be elaborated here.

[0043] In specific implementation, based on the defined sampling depth range, stratified drilling sampling is performed at a designated location in the target sampling unit to obtain soil samples of the corresponding depth range. This can be achieved in the following way: at the same center point of the target sampling unit, a hydraulic or manual soil drill is used to drill a hole; when the drill bit reaches the starting depth of the defined sampling depth range, a clean inner liner sampling tube is installed, and drilling continues downward until the drill bit reaches the ending depth of the sampling depth range, thereby obtaining a section of undisturbed soil core that completely covers the target depth range; subsequently, this section of core sample is carefully pushed out of the sampling tube, and a sterile scraper or cutter is used to cut off all or a representative part of the section of core, for example, mixing the soil from the upper, middle and lower parts, and then placing it into a pre-labeled sample bag or sample box for sealed preservation; finally, the soil material in each sealed container is the soil sample obtained from the designated sampling depth range of the target sampling unit; other methods can also be used in other embodiments, which are not limited here.

[0044] It should be noted that, in this application, the vertical variation curve refers to a graphical or sequential data representation used to visually demonstrate how one or more soil physical parameters change with increasing soil depth in the vertical direction; the characteristic depth point in this application refers to a predetermined depth value that marks a substantial abrupt change or turning point in a certain physical or chemical property of the soil, and is used as a key boundary marker for dividing different homogeneous soil layers or identifying special soil layers; the sampling depth interval in this application refers to a continuous depth range determined based on one or more characteristic depth points for actually obtaining soil samples, which is used to clearly define the vertical spatial target of sampling, ensuring that the collected soil samples can represent a soil layer with relatively uniform characteristics in the vertical direction or with specified research significance, thereby achieving stratified and targeted sampling; the soil sample in this application refers to a soil material entity collected from the target sampling unit within a predetermined depth range strictly according to the sampling depth interval, which is used to physically carry and represent the soil composition, structure, and state information of the sampling unit within the target depth interval, and provides a physical carrier with clear spatial orientation for subsequent laboratory analysis.

[0045] In step 105, the obtained soil samples are associated with and labeled with their corresponding spatial locations, sampling depths and soil condition information to generate structured soil sampling data.

[0046] In some embodiments, the obtained soil samples are associated and labeled with their corresponding spatial location, sampling depth, and soil condition information to generate structured soil sampling data, which can be achieved through the following steps: Generate and assign globally unique sample identifiers to the acquired soil samples; The sample identifier is digitally associated and bound with the spatial location, sampling depth range information of the corresponding target sampling unit, and soil state information collected at the target sampling unit. The associated information obtained after binding is encapsulated into a complete structured data record, thereby forming structured soil sampling data.

[0047] In practice, generating and assigning a globally unique sample identifier to the acquired soil sample can be achieved in the following way: After each soil sample is packaged and placed into a sample bag, the identifier generation module is invoked immediately through a mobile terminal connected to the sample information management system. Based on pre-set encoding rules, such as combining the project code, sampling date, and sequence number, this module automatically generates a globally unique string as the sample identifier and drives a portable label printer to print a QR code label containing the identifier. The operator then affixes this QR code label to the outer surface of the corresponding sample bag, thereby completing the physical assignment and digital generation of the sample identifier.

[0048] In specific implementation, the digital association and binding of the sample identifier with the spatial location, sampling depth range information, and soil condition information collected at the target sampling unit can be achieved in the following way: First, on the association operation interface of the mobile terminal, scan the QR code on the sample bag to automatically read the sample identifier; then, the operator binds the following three pieces of information with the identifier by selecting or inputting on the interface: The first is the spatial location of the target sampling unit, i.e., the coordinates of the center point of the unit, which are automatically retrieved from the established candidate sampling unit spatial attribute database by unit number; the second is the sampling depth range information, i.e., the soil condition information collected at the target sampling unit in this sampling operation. The depth range, for example, 0.4-0.9 meters, is manually entered by the operator based on the pre-sampling detection record or imported from the detection equipment's historical records. The third item is soil condition information, which is the volumetric water content, soil compaction, and apparent electrical conductivity values ​​measured on-site and standardized at the target sampling unit. This soil condition information is obtained from the standardized result database through unit number association. The soil condition information refers to real-time status data such as soil moisture content, compaction characteristics, and electrical response characteristics collected on-site. The system uses these three types of information as attribute fields and establishes a one-to-one, irreversible association record with the sample identifier as the primary key in a relational database or structured file to complete the digital binding.

[0049] In specific implementation, the associated information obtained after binding is encapsulated into a complete structured data record, thereby forming structured soil sampling data. This can be achieved in the following way: After completing the digital binding, according to a predefined data structure template compatible with the laboratory information management system, such as using JSON Schema or a relational database table structure, all fields in a bound record, namely sample identifier, spatial location coordinates, sampling depth range, standardized physical parameter values, and necessary metadata such as sampling time and operator, are automatically assembled and encapsulated to generate an independent, uniformly formatted structured data record. This record can be uploaded to a cloud database in real time via an application programming interface or appended to a local data file. As the sampling work progresses, all such structured data records corresponding to all samples accumulate in the database, forming a complete data set that can be queried, statistically analyzed, and spatially displayed. This set constitutes the structured soil sampling data described in this application, providing a complete, standardized, and traceable data foundation for subsequent soil quality assessment, spatial distribution analysis, or pollution tracking. Other methods can also be used in other embodiments, which are not limited here.

[0050] It should be noted that, in this application, the sample identifier refers to a globally unique identification code assigned to each soil sample, used to uniquely identify and track each individual sample in the digital and physical worlds; the associated information obtained after binding in this application refers to a complete set of information formed by forcibly linking the sample identifier with its corresponding spatial location, sampling depth range, soil condition information, etc., through digital means; the structured data record in this application refers to a standardized data entry formed by encapsulating the associated information obtained after binding according to a predefined machine-readable format and specification, used to organize scattered and heterogeneous associated information into a data object with a unified format, clearly defined fields, and direct storage, processing, and exchange by a computer system; the structured soil sampling data in this application refers to a complete data set composed of structured data records corresponding to all samples, used as an integrated, standardized, spatially defined database to systematically carry all the results information of the entire monitoring and sampling task, and can provide a directly usable standardized data foundation for subsequent statistical analysis, spatial modeling, quality assessment, and long-term monitoring comparison.

[0051] In another aspect, in some embodiments, this application provides a soil monitoring and sampling system, referring to... Figure 4 The figure is a schematic diagram of the structure of a soil monitoring and sampling system according to some embodiments of this application. The soil monitoring and sampling system 400 includes: an acquisition module 401, a processing module 402, and an execution module 403, which are described below: The acquisition module 401 in this application is mainly used to acquire the spatial location and historical soil information of the area to be monitored; Processing module 402, in this application, is mainly used to divide the area to be monitored into grids based on the spatial location and historical soil information, forming candidate sampling units with spatial correlation. Within each candidate sampling unit, physical parameters characterizing the state of the soil surface and shallow layers are collected, and a high-dimensional feature vector reflecting the differences in soil microenvironment is generated. The processing module 402 described in this application is also used to construct a dynamic sampling priority scoring model that reflects the degree of spatial difference in soil based on all high-dimensional feature vectors, and to dynamically determine the target sampling unit through the priority scoring model. The processing module 402 described in this application is also used to perform a depth stratification sampling strategy on the target sampling unit, adaptively determine the sampling depth range according to the vertical variation trend of soil physical parameters, and then obtain a representative soil sample. The execution module 403 in this application is mainly used to associate and mark the obtained soil samples with the corresponding spatial location, sampling depth and soil state information to generate structured soil sampling data.

[0052] The foregoing has detailed examples of soil monitoring and sampling methods, systems, devices, and media provided in the embodiments of this application. It is understood that the corresponding apparatus includes hardware structures and / or software modules for performing each function in order to achieve the aforementioned functions. Those skilled in the art should readily recognize that, based on the units and algorithm steps described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0053] In some embodiments, this application also provides a computer device, the computer device including a memory and a processor, the memory for storing a computer program, and the processor for calling and running the computer program from the memory, so that the computer device performs the above-described soil monitoring and sampling method.

[0054] In some embodiments, reference Figure 5 The dashed lines in the figure indicate that the unit or module is optional. This figure is a schematic diagram of the structure of a computer device for implementing the soil monitoring and sampling method of this application. The soil monitoring and sampling method in the above embodiments can be achieved through... Figure 5 The computer device 500 shown is used to implement this, and the computer device 500 includes at least one processor 501, a memory 502 and at least one communication unit 505. The computer device 500 may be a terminal device, a server or a chip.

[0055] The processor 501 can be a general-purpose processor or a special-purpose processor. For example, the processor 501 can be a central processing unit (CPU). The CPU can be used to control the computer device 500, execute software programs, and process data from the software programs. The computer device 500 may also include a communication unit 505 for inputting (receiving) and outputting (transmitting) signals.

[0056] For example, computer device 500 may be a chip, communication unit 505 may be the input and / or output circuit of the chip, or communication unit 505 may be the communication interface of the chip, and the chip may be a component of terminal device, network device or other device.

[0057] For example, computer device 500 may be a terminal device or a server, and communication unit 505 may be a transceiver of the terminal device or the server, or communication unit 505 may be a transceiver circuit of the terminal device or the server.

[0058] The computer device 500 may include one or more memories 502 storing a program 504. The program 504 can be executed by a processor 501 to generate instructions 503, causing the processor 501 to perform the methods described in the above method embodiments according to the instructions 503. Optionally, the memory 502 may also store data (such as a target audit model). Optionally, the processor 501 may also read data stored in the memory 502, which may be stored at the same storage address as the program 504, or the data may be stored at a different storage address than the program 504.

[0059] The processor 501 and memory 502 can be configured separately or integrated together, for example, integrated on the system-on-chip (SOC) of the terminal device.

[0060] It should be understood that each step of the above method embodiment can be completed by hardware logic circuits or software instructions in the processor 501. The processor 501 can be a CPU, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, such as discrete gate, transistor logic devices, or discrete hardware components.

[0061] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0062] For example, in some embodiments, this application also provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the above-described soil monitoring and sampling method.

[0063] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0064] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A soil monitoring and sampling method, characterized in that, Includes the following steps: Obtain the spatial location and historical soil information of the area to be monitored; Based on the spatial location and historical soil information, the area to be monitored is divided into grids to form candidate sampling units with spatial correlation. Within each candidate sampling unit, physical parameters characterizing the state of the soil surface and shallow layers are collected, and high-dimensional feature vectors reflecting the differences in soil microenvironment are generated. A priority scoring model for dynamic sampling that reflects the degree of spatial difference in soil is constructed based on all high-dimensional feature vectors, and the target sampling unit is dynamically determined through the priority scoring model. A depth stratified sampling strategy is implemented for the target sampling unit, and the sampling depth range is adaptively determined according to the vertical variation trend of soil physical parameters, so as to obtain representative soil samples. The obtained soil samples are associated and labeled with their corresponding spatial location, sampling depth, and soil condition information to generate structured soil sampling data.

2. The method as described in claim 1, characterized in that, Based on the spatial location and historical soil information, the area to be monitored is divided into grids to form candidate sampling units with spatial correlation, specifically including: Based on the historical soil information, the spatial variation coefficients of key soil properties are determined, and then the grid division size is dynamically determined. Based on the area boundary defined by the spatial location and the grid division size, the area to be monitored is regularly divided into multiple continuous grid units; Each grid cell is assigned a unique identifier and its spatial geometric properties are calculated to form candidate sampling cells with spatial relationships.

3. The method as described in claim 1, characterized in that, Within each candidate sampling unit, physical parameters characterizing the state of the soil surface and shallow layers are collected, and high-dimensional feature vectors reflecting differences in the soil microenvironment are generated, specifically including: Collect physical parameters characterizing the surface and shallow soil conditions for each candidate sampling unit; The collected physical parameters are standardized and quality verified to obtain a standardized set of physical parameters; The standardized physical parameter set corresponding to each candidate sampling unit and its associated historical soil attribute data are fused and dimensionality reduced to generate a high-dimensional feature vector that reflects the differences in soil microenvironment.

4. The method as described in claim 1, characterized in that, The priority scoring model for dynamic sampling, which reflects the degree of spatial variability in soil, is constructed based on all high-dimensional feature vectors. Specifically, it includes: Unsupervised clustering analysis was performed based on all high-dimensional feature vectors to divide the soil microenvironment feature space into multiple categories and determine the centroid of each category. Based on the clustering results, the distance from the high-dimensional feature vector of each candidate sampling unit to the centroid of its class, as well as the minimum distance to the centroids of other classes, are calculated, thereby determining the priority scoring model for dynamic sampling that reflects the degree of spatial difference in soil.

5. The method as described in claim 1, characterized in that, A depth-stratified sampling strategy is implemented for the target sampling unit, adaptively determining the sampling depth range based on the vertical variation trend of soil physical parameters, thereby obtaining representative soil samples. Specifically, this includes: Vertical continuous detection of soil physical parameters is carried out at the designated location of the target sampling unit to obtain the vertical variation curve reflecting the parameter changes with depth; Based on the vertical variation curve, identify the characteristic depth points where the rate of change of soil properties exceeds a preset threshold, and define the sampling depth range based on the identified characteristic depth points; Based on the defined sampling depth range, stratified borehole sampling is performed at the designated location of the target sampling unit to obtain soil samples of the corresponding depth range.

6. The method as described in claim 1, characterized in that, The obtained soil samples are associated and labeled with their corresponding spatial location, sampling depth, and soil condition information to generate structured soil sampling data, specifically including: Generate and assign globally unique sample identifiers to the acquired soil samples; The sample identifier is digitally associated and bound with the spatial location, sampling depth range information of the corresponding target sampling unit, and soil state information collected at the target sampling unit. The associated information obtained after binding is encapsulated into a complete structured data record, thereby forming structured soil sampling data.

7. The method as described in claim 1, characterized in that, The physical parameters include at least the water content state, density characteristics, or electrical response characteristics.

8. A soil monitoring and sampling system, characterized in that, include: The acquisition module is used to acquire the spatial location and historical soil information of the area to be monitored; The processing module is used to divide the area to be monitored into grids based on the spatial location and historical soil information, forming candidate sampling units with spatial correlation. Within each candidate sampling unit, physical parameters characterizing the state of the soil surface and shallow layers are collected, and a high-dimensional feature vector reflecting the differences in soil microenvironment is generated. The processing module is also used to construct a dynamic sampling priority scoring model that reflects the degree of spatial difference in soil based on all high-dimensional feature vectors, and to dynamically determine the target sampling unit through the priority scoring model. The processing module is also used to execute a depth stratification sampling strategy on the target sampling unit, adaptively determine the sampling depth range according to the vertical variation trend of soil physical parameters, and thus obtain representative soil samples. The execution module is used to associate and label the obtained soil samples with their corresponding spatial location, sampling depth, and soil condition information to generate structured soil sampling data.

9. A computer device, characterized in that, The computer device includes a memory and a processor, the memory for storing computer programs, and the processor for calling and running the computer programs from the memory, causing the computer device to perform the soil monitoring and sampling method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions or code that, when executed on a computer, cause the computer to perform the soil monitoring and sampling method as described in any one of claims 1 to 7.

Images