A cloud computing-based groundwater exploration data processing method and system

Abstract

The application discloses a kind of groundwater exploration data processing method and system based on cloud computing, it is related to data processing technical field, including the following steps, acquisition and processing original survey data set, generate quality control voxel index body set and error field set;Quality control voxel index body set is carried out membership mapping and error weight fusion, generate processed water-bearing tendency scalar field and multi-resolution scalar field set, and establish the scale alignment mapping between adjacent resolution;For multi-resolution scalar field set, establish polygonal complex and generate filter complex sequence, and carry out discrete Morse reduction, obtain the reduced complex sequence, it is constructed as sawtooth sequence, and the sawtooth sequence is calculated sawtooth persistent homology, obtain topological feature trajectory set;Based on topological feature trajectory set, extract spatial structure object set and carry out multi-resolution fusion, generate groundwater three-dimensional topological spatial structure and structure consistency field, output groundwater three-dimensional topological model.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a cloud computing-based method and system for processing groundwater exploration data. Background Technology

[0002] Current 3D groundwater exploration modeling heavily relies on heterogeneous data such as borehole and geophysical data. However, due to significant differences in spatial data density and measurement accuracy, spatial interpolation is prone to generating error fields. Traditional 3D geological modeling often relies on single-grid resolution and fixed scalar thresholds for isosurface extraction, making it highly susceptible to interference from local high-frequency noise and interpolation edge effects. This results in frequent digital artifacts such as breaks or false connectivity in the extracted groundwater network. Furthermore, massive geological voxel grids face a severe curse of dimensionality when performing complex spatial connectivity analysis, and existing methods lack rigorous mechanisms for verifying the topological stability of the extraction results at different spatial scales. Consequently, it is difficult to automatically and robustly output a 3D spatial structure model with high confidence and true hydrogeological and physical significance without human intervention.

[0003] Currently, Chinese invention patent application CN202411479397.7 discloses a GIS data processing method and system for geological surveys and geochemical explorations, relating to the field of GIS data processing technology. It solves the technical problems of pre-generating a large number of rasterized tiles from vector and point cloud data, the storage and access of these large numbers of rasterized tiles causing system delays, and the chaotic management of access to various types of data, affecting overall data quality. By separating different types of data, it facilitates more efficient management and retrieval. Each data type has its specific application scenarios and processing methods; categorized storage helps to quickly locate the required data. Compressing raster and point cloud data and storing it on a cloud server can significantly reduce the storage space occupied and save site storage space, which is especially important for large-scale geographic information systems because these data are often very large. Meanwhile, storing vector data at the corresponding site facilitates timely retrieval of location data; however, related technologies struggle to extract significant topological features through persistent cohomology, lacking consistency and stability, and cannot effectively integrate error weights and membership information with the scale alignment mechanism of multi-resolution voxel scalar fields, resulting in low accuracy in groundwater 3D structure modeling. Summary of the Invention

[0004] The technical problem solved by this invention is that related technologies have difficulty extracting significant topological features through persistent cohomology of sawtooth patterns, lack consistency and stability, and cannot effectively integrate error weights and membership information with the scale alignment mechanism of multi-resolution voxel scalar fields, resulting in low accuracy in three-dimensional modeling of groundwater structures.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: A cloud computing-based groundwater exploration data processing method includes the following steps: Step S1: Obtain and process the original exploration data set to generate a quality control voxel index set and a quality control error field set; Step S2: Perform membership mapping and error weight fusion on the set of quality control voxel indicators to generate the processed water content scalar field and multi-resolution scalar field set, and establish scale alignment mapping between adjacent resolutions. Step S3: Construct cubic complexes for each resolution scalar field in the multi-resolution scalar field set and generate a filtered complex sequence, and perform discrete Morse reduction on the filtered complex sequence to obtain a reduced complex sequence; Step S4: Based on scale alignment mapping, construct the reduced complex sequence corresponding to each resolution into a zigzag sequence, and calculate the persistent homology of the zigzag sequence to obtain the set of topological feature trajectories; Step S5: Extract the set of spatial structure objects based on the set of topological feature trajectories and perform multi-resolution fusion to generate the three-dimensional topological spatial structure and structural consistency field of groundwater, and output the three-dimensional topological model of groundwater.

[0006] Preferably, step S1 specifically includes: Step S11: Obtain the original exploration data set, which includes borehole interpretation data and geophysical inversion body data; The borehole interpretation data includes the borehole spatial coordinates and corresponding hydrogeological parameters; The geophysical inversion data includes three-dimensional grid coordinates and corresponding physical property values; Step S12: Standardize the original exploration data set to obtain the exploration data set. The standardization process includes coordinate standardization, unit standardization, and error metadata standardization. Step S13: Map the survey data set to a predefined voxel grid through spatial interpolation to obtain a set of voxel index volumes; Based on the spatial distribution characteristics and inherent errors in the exploration data set, the comprehensive error value of each voxel grid in the predefined voxel grid is calculated on a voxel-by-voxel basis, generating an error field set aligned on a voxel-by-voxel basis with the voxel index set. The spatial distribution characteristics are the three-dimensional spatial distances between sampling points; Step S14: Perform quality control operations on the voxel index set and the error field set. The quality control operations include outlier determination and isolated body removal, resulting in a quality control voxel index set and a quality control error field set.

[0007] Preferably, step S2 specifically includes: Step S21: Based on the membership function, perform membership mapping on the quality control voxel indexes in the quality control voxel index set to obtain the water content membership set. Step S22: Generate a weight volume based on the quality control error field, and obtain the weight volume set. The calculation expression for the weight volume is as follows: ; in, To represent the weight volume value of the i-th type of physical index at spatial coordinates (x, y, z), As a predetermined positive number, Let be the error value after quality control corresponding to the i-th type of physical index at spatial coordinates (x, y, z). Index for physical indicator categories; Step S23: Calculate the water-bearing tendency scalar field based on the water-bearing membership set and the weight set; Step S24: Smooth and denoise the water-bearing tendency scalar field to obtain the processed water-bearing tendency scalar field; Step S25: Spatial downsampling is performed on the processed water content scalar field to generate a multi-resolution voxel mesh set and a multi-resolution scalar field set. The spatial coordinate correspondence rules in the spatial downsampling process are used as resampling relations, and scale alignment mapping between adjacent resolutions is established based on the resampling relations.

[0008] Preferably, the calculation expression for the scalar field of water content is as follows: ; in, Let be the scalar value of the water-bearing tendency at spatial coordinates (x, y, z). Let be the water content membership value of the i-th type of physical index at spatial coordinates (x, y, z).

[0009] Preferably, step S3 specifically includes: Step S31: Based on the multi-resolution scalar field set, construct cubic complexes on the voxel grids corresponding to each resolution in the multi-resolution voxel grid set, map the voxel units in the cubic complexes to three-dimensional cells, and use the resolution scalar field values ​​corresponding to the voxel units as the scalar values ​​of the three-dimensional cells. The common surfaces, common edges, and vertices between voxel units are mapped to the corresponding two-dimensional cells, one-dimensional cells, and zero-dimensional cells, respectively. Step S32: Using the upper super-level set filtering method, the scalar values ​​of the three-dimensional cell are propagated to the low-dimensional cell. The processing logic includes: For a target low-dimensional cell with dimension k < 3, extract the maximum value of the target low-dimensional cell and all its adjacent three-dimensional cells, and use the maximum value as the cell scalar value. The low-dimensional cell includes two-dimensional cell, one-dimensional cell, and zero-dimensional cell; Step S33: Compare the preset threshold sequence with the cell scalar value one by one, extract the set of cells whose cell scalar value is greater than or equal to the current preset threshold, and obtain the filtered complex sequence. The filtered complex corresponding to any threshold in the preset threshold sequence satisfies: ; in, To generate a filter sub-complex at a given threshold, For cell unit, It is a cubic complex. The scalar value of the cell. For a given threshold, The logic for filtering the superlevel set; Step S34: Perform discrete Morse reduction on the filtered complex sequence to obtain the reduced complex sequence and the corresponding Morse key cell set.

[0010] Preferably, step S4 specifically includes: Traverse the reduced complex sequence corresponding to each resolution scalar field in the multi-resolution scalar field set, establish a complex mapping between the reduced complex sequences corresponding to adjacent resolution scalar fields, and connect the reduced complex sequences corresponding to each resolution scalar field in resolution order to construct a zigzag sequence; The persistent homology of the sawtooth sequence is calculated to obtain a set of topological feature trajectories.

[0011] Preferably, the specific processing logic for calculating the persistent cohomology of the zigzag sequence includes: Traverse the zigzag sequence. When traversing to the first zigzag position of the zigzag sequence, a certain topological feature appears for the first time in the corresponding reduced complex. The cell scalar value associated with the first zigzag position is recorded as the birth scalar value of the topological feature. When traversing to the second zigzag position of the zigzag sequence, the topological feature disappears in the corresponding reduced complex, and the cell scalar value associated with the second zigzag position is recorded as the death scalar value of the topological feature. Based on the pairing criterion of persistent cohomology, the birth scalar value and death scalar value of the same topological feature are paired to form a topological feature trajectory. By traversing all topological features, a set of topological feature trajectories is obtained.

[0012] Preferably, step S5 specifically includes: Step S51: Under the logic of filtering significant topological feature trajectories, filter the set of significant topological feature trajectories from the set of topological feature trajectories; The logic for filtering salient topological feature trajectories specifically includes: If the same topological feature trajectory has a corresponding continuation relationship between the reduced complex sequences corresponding to adjacent resolutions through scale alignment mapping, then the topological feature trajectory is determined to have cross-resolution consistency. If a topological feature trajectory cannot be mapped by scale alignment, and there is a corresponding continuation relationship between the reduced complex sequences corresponding to adjacent resolutions, then the topological feature trajectory is removed from the set of topological feature trajectories. Extract the birth and death scalar values ​​corresponding to the topological feature trajectories with cross-resolution consistency, calculate the length of the survival scalar interval, and use the length as a persistence index; Topological feature trajectories with a persistence index greater than a preset stability threshold are selected as significant topological feature trajectories. Based on cross-resolution consistency and durability indices, the stability weights of each topological feature are calculated and mapped to a predefined voxel grid to generate a structural consistency field. Step S52: Obtain the extraction threshold based on the set of significant topological feature trajectories, and extract the set of spatial structure objects from the cubic complex corresponding to each resolution under the extraction threshold. The set of spatial structure objects includes the set of connected component spatial structure objects, the set of channel skeleton spatial structure objects, and the set of closed shell spatial structure objects. The processing logic includes: When dimension k=0, connect component labeling is performed on zero-dimensional cells that meet the extraction threshold, and a set of connected component spatial structure objects is output. When dimension k=1, combine the Morse key cell set and perform skeletonization operation on one-dimensional cells that meet the extraction threshold to output a set of channel skeleton space structure objects. When dimension k=2, boundary extraction is performed on the two-dimensional cell boundaries that meet the extraction threshold, and a set of closed shell space structure objects is output. Step S53: Using the set of significant topological feature trajectories, resolution fusion is performed on the set of spatial structure objects at each resolution. The fusion result is then verified by combining the set of Morse key cells to obtain the three-dimensional topological spatial structure of groundwater. The verification process includes redundancy elimination and connectivity preservation. Step S54: Spatial mapping is performed between the three-dimensional topological spatial structure of groundwater and the structural consistency field to output the three-dimensional topological model of groundwater.

[0013] Preferably, obtaining the extraction threshold based on the set of significant topological feature trajectories specifically includes: Read the birth and death scalar values ​​corresponding to the trajectories with significant topological features, and generate the survival scalar interval; The common stable interval is obtained by intersecting the survival scalar intervals of all resolutions. If the common stable interval is a non-empty set, then the extraction threshold is selected from the common stable interval according to the selection rules; If the common stable interval is an empty set, within the union of the survival scalar intervals at each resolution, select the scalar value covered by the most survival scalar intervals as the candidate extraction threshold. If there are multiple candidate extraction thresholds, then the extraction threshold will be selected from the candidate extraction thresholds according to the selection rules.

[0014] A cloud computing-based groundwater exploration data processing system includes an acquisition module, a mapping module, a filtering module, a calculation module, and an output module. The acquisition module is used to acquire and process the original exploration data set, and generate a quality control voxel index set and a quality control error field set. The mapping module is used to perform membership mapping and error weight fusion on the set of quality control voxel indicators, generate the processed water content scalar field and multi-resolution scalar field set, and establish scale alignment mapping between adjacent resolutions. The filtering module is used to construct cubic complexes for each resolution scalar field in the multi-resolution scalar field set and generate a filtered complex sequence, and to perform discrete Morse reduction on the filtered complex sequence to obtain a reduced complex sequence. The computation module is used to construct a zigzag sequence from the reduced complex sequence corresponding to each resolution based on the scale alignment mapping, and to calculate the persistent homology of the zigzag sequence to obtain a set of topological feature trajectories. The output module is used to extract a set of spatial structure objects based on the set of topological feature trajectories and perform multi-resolution fusion to generate a three-dimensional topological spatial structure and a structural consistency field of groundwater, and output a three-dimensional topological model of groundwater.

[0015] The beneficial effects of this invention are as follows: This invention utilizes a sawtooth persistent cohomology and discrete Morse reduction mechanism to achieve high robustness and adaptive analysis of the three-dimensional topological spatial structure of groundwater. By performing discrete Morse reduction on the filtered complex sequence for topological dimensionality reduction, cross-resolution object correspondence is established. Furthermore, topological consistency verification using topological splitting and merging events accurately isolates spatial structural object sets of different dimensions from massive local noise. The cross-resolution consistency metric and persistence index metric are fused and quantized into a structural consistency field, solving the modeling distortion problem in the process of spatial downsampling and multi-resolution scalar field fusion of exploration data sets. Attached Figure Description

[0016] Figure 1 A flowchart illustrating the steps of a cloud computing-based groundwater exploration data processing method according to an embodiment of the present invention; Figure 2This is a schematic diagram of the basic process of a cloud computing-based groundwater exploration data processing system provided in one embodiment of the present invention. Detailed Implementation

[0017] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0018] Example 1, referring to Figure 1 This paper provides a cloud computing-based method for processing groundwater exploration data, including the following steps: Step S1: Obtain and process the original exploration data set to generate a quality control voxel index set and a quality control error field set; Step S2: Perform membership mapping and error weight fusion on the set of quality control voxel indicators to generate the processed water content scalar field and multi-resolution scalar field set, and establish scale alignment mapping between adjacent resolutions. Step S3: Construct cubic complexes for each resolution scalar field in the multi-resolution scalar field set and generate a filtered complex sequence, and perform discrete Morse reduction on the filtered complex sequence to obtain a reduced complex sequence; Step S4: Based on scale alignment mapping, construct the reduced complex sequence corresponding to each resolution into a zigzag sequence, and calculate the persistent homology of the zigzag sequence to obtain the set of topological feature trajectories; Step S5: Extract the set of spatial structure objects based on the set of topological feature trajectories and perform multi-resolution fusion to generate the three-dimensional topological spatial structure and structural consistency field of groundwater, and output the three-dimensional topological model of groundwater.

[0019] This invention utilizes a sawtooth persistent cohomology and discrete Morse reduction mechanism to achieve high robustness and adaptive analysis of the three-dimensional topological spatial structure of groundwater. By performing discrete Morse reduction on the filtered complex sequence for topological dimensionality reduction, cross-resolution object correspondences are established. Furthermore, by using topological consistency verification through topological splitting and merging events, spatial structural object sets of different dimensions are accurately extracted from massive local noise, solving the modeling distortion problem in the process of spatial downsampling and multi-resolution scalar field fusion of exploration data sets.

[0020] In a specific embodiment, step S1 specifically includes: Step S11: Obtain the original exploration data set, which includes borehole interpretation data and geophysical inversion body data; Borehole interpretation data includes borehole spatial coordinates and corresponding hydrogeological parameters; Geophysical inversion data includes three-dimensional grid coordinates and corresponding physical property values; Step S12: Standardize the original exploration data set to obtain the exploration data set. The standardization process includes coordinate standardization, unit standardization, and error metadata standardization. Step S13: Map the survey data set to a predefined voxel grid through spatial interpolation to obtain a set of voxel index volumes; Based on the spatial distribution characteristics and inherent errors in the exploration data set, the comprehensive error value of each voxel grid in the predefined voxel grid is calculated on a voxel-by-voxel basis, generating an error field set aligned on a voxel-by-voxel basis with the voxel index set. The spatial distribution characteristics are the three-dimensional spatial distances between sampling points; Step S14: Perform quality control operations on the voxel index set and the error field set. The quality control operations include outlier determination and isolated body removal, resulting in the quality control voxel index set and the quality control error field set.

[0021] Specifically, hydrogeological parameters typically include water content and permeability coefficient; Physical property values ​​typically include resistivity and polarizability.

[0022] By conducting physical geological drilling in the target exploration area, combined with in-hole resistivity logging and in-hole hydrogeological tests, continuous original borehole logging curves and core physical logging information are obtained. Professional lithological interpretation software is used to comprehensively interpret this original information, extracting the water cut and permeability coefficients corresponding to different depths. Density electrical resistivity meters and ground-penetrating radars are deployed on the surface of the exploration area to transmit and receive physical response signals underground, acquiring observational data. Using topographically-constrained 3D smoothing inversion, the observed surface physical response signals are converted into a continuous 3D model reflecting the distribution of the subsurface medium. A 3D grid coordinate system covering the entire area, along with the corresponding resistivity and polarizability blocks, is output and read and imported into the system in the form of a standard 3D volumetric metadata file.

[0023] Standardization processing includes coordinate standardization, unit standardization, and error metadata standardization. Among them, coordinate standardization involves uniformly projecting and mapping the local engineering coordinate system, relative survey line coordinate system used in borehole interpretation data, and Yellow Sea elevation used in geophysical inversion data to the standard three-dimensional Cartesian coordinate system through a spatial coordinate transformation matrix, thereby eliminating the benchmark deviation of spatial position.

[0024] Units are standardized based on water content (%), permeability coefficient (m / d), and resistivity. Polarizability % has completely different physical dimensions and numerical magnitudes. The basic physical units are converted, and the converted basic physical units are subjected to dimensionless normalization processing to uniformly map their numerical range to the interval [0,1].

[0025] Error metadata standardization involves converting the inherent measurement error parameters of borehole logging instruments and the fitting residuals generated by geophysical electromagnetic inversion algorithms into Gaussian distribution standard deviations in a standard statistical sense, which serve as their respective inherent error benchmarks.

[0026] The predefined voxel grid is a regular three-dimensional discretized grid structure divided according to the spatial extent of the target exploration area in a standard three-dimensional Cartesian coordinate system. The overall dimension of the predefined voxel grid is preferably 256×256×128 to 1024×1024×512 voxel units; the spatial resolution of a single voxel unit is preferably 5 meters to 50 meters horizontally and 0.5 meters to 5 meters vertically. For example, the predefined voxel grid is preferably divided into a three-dimensional array containing 512×512×256 voxel units, and the physical size of a single voxel is set to 10m×10m×2m. This preferred grid can both preserve the high-frequency, fine stratigraphic variations in the vertical direction of borehole data and smooth the macroscopic trends in the horizontal direction of geophysical data, while its 2 N The grid dimension size ensures that when generating multi-resolution scalar field sets, the grids at each level achieve absolutely perfect integer-pixel scale alignment mapping, eliminating coordinate floating-point errors caused by resampling.

[0027] The comprehensive error value for each voxel in a predefined voxel grid is calculated on a voxel-by-voxel basis, specifically including: In spatial interpolation, voxel errors not only stem from inherent instrument errors but are also affected by spatial interpolation uncertainties caused by distance from sampling points. Therefore, for any target voxel in a predefined voxel grid, the nearest measured sampling point is first searched in the survey dataset, and the three-dimensional spatial distance between them is calculated as a spatial distribution feature. A spatial error transfer function based on distance attenuation is then constructed, and its calculation expression is: ;in For the comprehensive error, This is due to the inherent error of the most recent sampling point. For three-dimensional spatial distance, The preset interpolation space attenuation penalty coefficient is preferably 0.005. Through this function, the comprehensive error value of voxels farther from the measured point is amplified, thereby generating an error field set that is perfectly aligned with the voxel index set on a voxel-by-voxel basis.

[0028] Quality control operations include outlier detection and isolated object removal. Outlier detection employs a spatial neighborhood-based high-frequency abrupt change filtering to identify points in the voxel index volume with abnormal numerical spikes. Isolated object removal involves performing a 3D connected component labeling algorithm within the voxel mesh to calculate the volume of connected components formed by adjacent voxels with the same attribute range. If the volume of a connected component is less than a preset threshold for the number of small voxels, and it is a suspended fragment composed of 3 voxels, it is identified as an isolated object and removed.

[0029] In a specific embodiment, step S2 specifically includes: Step S21: Based on the membership function, perform membership mapping on the quality control voxel indexes in the quality control voxel index set to obtain the water content membership set. Step S22: Generate a weight volume based on the quality control error field, and obtain the weight volume set. The calculation expression for the weight volume is as follows: ; in, To represent the weight volume value of the i-th type of physical index at spatial coordinates (x, y, z), As a predetermined positive number, Let be the error value after quality control corresponding to the i-th type of physical index at spatial coordinates (x, y, z). Index for physical indicator categories; Step S23: Calculate the water-bearing tendency scalar field based on the water-bearing membership set and the weight set; Step S24: Smooth and denoise the water-bearing tendency scalar field to obtain the processed water-bearing tendency scalar field; Step S25: Spatial downsampling is performed on the processed water content scalar field to generate a multi-resolution voxel mesh set and a multi-resolution scalar field set. The spatial coordinate correspondence rules in the spatial downsampling process are used as resampling relations, and scale alignment mapping between adjacent resolutions is established based on the resampling relations.

[0030] Specifically, the membership functions are the Sigmoid function and the inverse Z-shaped function. For indicators that are positively correlated with water content, the Sigmoid function is used for mapping, so that the larger the physical value, the closer the water content membership degree is to 1. For indicators that are negatively correlated with water content, the inverse Z-shaped function is used for mapping, so that the lower the resistivity value, the closer the water content membership degree is to 1, while the mapping value of the high resistivity region of dense dry rock is close to 0.

[0031] Step S24, specifically smoothing and denoising the scalar field of water content tendency, includes: A preferred approach is three-dimensional bilateral filtering, which introduces both spatial distance weights and scalar value difference weights into the convolution kernel. This operation smooths out local high-frequency random noise introduced by interpolation algorithms or the data itself, while strictly preserving the abrupt geological boundaries between aquifers and impermeable layers. This denoising step is absolutely essential for subsequent topology calculations, as it can largely eliminate spurious local extrema caused by minute fluctuations in the scalar field.

[0032] In step S25, the spatial downsampling of the processed water content scalar field specifically includes: A three-dimensional non-overlapping sliding window pooling method is used to progressively downsample the base resolution voxel mesh. Taking a downsampling ratio of 2 as an example, a 2×2×2 three-dimensional sliding window is used to slide within the current scalar field with a step size of 2. Within the window, the average or maximum value of the water content scalar value of the eight high-dimensional voxel cells is extracted and assigned to a corresponding coarse-scale voxel cell in the next level of low-resolution mesh. This operation is repeated to generate a multi-resolution voxel mesh set and a corresponding multi-resolution scalar field set with progressively halved spatial resolution.

[0033] The spatial coordinate correspondence rule strictly records the algebraic relationship of the absolute spatial index mapping between parent and child nodes in the multi-resolution voxel mesh during the upper spatial downsampling process. Assuming the downsampling ratio is n=2, if the 3D coordinate index of a voxel in the high-resolution mesh is (x... fine, y fine ,z fine The unique voxel 3D coordinate index (x) in adjacent low-resolution grids is calculated proportionally. coarse, y coarse ,z coarse This algebraic correspondence rule is stored as a resampling relation in the system's mapping dictionary.

[0034] In a specific embodiment, the calculation expression for the water-bearing scalar field is as follows: ; in, Let be the scalar value of the water-bearing tendency at spatial coordinates (x, y, z). Let be the water content membership value of the i-th type of physical index at spatial coordinates (x, y, z). This is to perform a summation operation on all physical index categories involved in the calculation.

[0035] In a specific embodiment, step S3 specifically includes: Step S31: Based on the multi-resolution scalar field set, construct cubic complexes on the voxel grids corresponding to each resolution in the multi-resolution voxel grid set, map the voxel units in the cubic complexes to three-dimensional cells, and use the resolution scalar field values ​​corresponding to the voxel units as the scalar values ​​of the three-dimensional cells. The common surfaces, common edges, and vertices between voxel units are mapped to the corresponding two-dimensional cells, one-dimensional cells, and zero-dimensional cells, respectively. Step S32: Using the upper super-level set filtering method, the scalar values ​​of the three-dimensional cell are propagated to the low-dimensional cell. The processing logic includes: For a target low-dimensional cell with dimension k < 3, extract the maximum value of the target low-dimensional cell and all its adjacent three-dimensional cells, and use the maximum value as the cell scalar value. Low-dimensional cells include two-dimensional cells, one-dimensional cells, and zero-dimensional cells; Step S33: Compare the preset threshold sequence with the cell scalar value one by one, extract the set of cells whose cell scalar value is greater than or equal to the current preset threshold, and obtain the filtered complex sequence. The filtered complex corresponding to any threshold in the preset threshold sequence satisfies: ; in, To generate a filter sub-complex at a given threshold, For cell unit, It is a cubic complex. The scalar value of the cell. For a given threshold, The logic for filtering the superlevel set; Step S34: Perform discrete Morse reduction on the filtered complex sequence to obtain the reduced complex sequence and the corresponding Morse key cell set.

[0036] Specifically, the voxel units in the cubic complex are mapped to three-dimensional cells. This involves directly defining each solid cube pixel in a predefined voxel mesh as a topologically highest-dimensional fundamental geometric primitive, i.e., a three-dimensional cell. This mapping does not change the physical spatial location of the data; instead, it establishes an object-oriented topological index, giving it entity status in algebraic topological computation.

[0037] Mapping the common faces, edges, and vertices between voxel units to corresponding 2D, 1D, and 0D cells respectively involves extracting the common cross-section shared between two adjacent 3D cells and indexing and instantiating it as a 2D cell; extracting the common line segments where four adjacent 3D cells intersect and instantiating them as 1D cells; and extracting the common corner points where eight adjacent 3D cells intersect and instantiating them as 0D cells. Thus, the originally isolated discrete voxel array is rigorously reconstructed into a hierarchical primitive network containing points, edges, faces, volumes, and their interrelationships—a complete cubic complex.

[0038] The specific processing logic for propagating scalar values ​​of three-dimensional cells to lower-dimensional cells using the super-level set filtering method includes: A coplanar, edge-sharing, and point-sharing adjacency matrix is ​​established by traversing the mesh. For any target low-dimensional cell with dimension k < 3, all high-dimensional cells with higher dimensions that include that face are retrieved from its adjacency matrix. After extracting the initial scalar values ​​corresponding to these high-dimensional voxels, max pooling is performed, and the maximum value is hard-coded and assigned to the face. This process is repeated, with one-dimensional edges taking the maximum values ​​of all surrounding one-dimensional faces or three-dimensional voxels. This logic ensures that high-frequency extremum signals can be losslessly transmitted to the topological skeleton of the structure.

[0039] The dimension k takes values ​​of 0, 1, and 2.

[0040] To meet the strict mathematical prerequisites of topological filtering—that is, for any high-dimensional cell to exist at a certain threshold, all its low-dimensional boundary cells must exist beforehand or simultaneously at the same threshold—the initial water content scalar field only has values ​​on voxels. If it is not propagated to lower dimensions, it will lead to missing faces, edges, and points when truncating the model, thus causing the topological homology group calculation to collapse. By assigning the maximum value of its neighboring high-dimensional cells to low-dimensional cells, it can be ensured that in the subsequent top-down threshold truncating process, once a high-probability 3D cell appears, the faces, edges, and points surrounding it will necessarily appear synchronously, thus guaranteeing that the geometry extracted at any threshold is a closed and valid topological structure.

[0041] The preset threshold sequence is generated by extracting all real, non-repeating cell scalar values ​​in the multi-resolution scalar field and arranging them in descending order.

[0042] ; Given a water-bearing tendency threshold, the entire grid is traversed, and only cells with a scalar value F greater than or equal to the water-bearing tendency threshold are selected. These selected cells e combine to form a new, smaller subshape, which is the filter sub-complex. As the water-bearing tendency threshold gradually decreases, the volume of the filter sub-complex expands outward like rising water, and the spatial topological characteristics of the groundwater are fully recorded during this expansion process.

[0043] Discrete Morse reduction of the filtered complex sequence specifically includes: A discrete gradient vector field is constructed on the cubic complex. Adjacent cell pairs satisfying specific scalar value matching conditions are sought. If these cell pairs belong to free surfaces that do not cause topological changes in the homology group calculation, they are paired and eliminated simultaneously at the algebraic level. After multiple rounds of massive pairing and elimination, the core cells in the network that cannot be eliminated by pairing and cause fundamental changes in the topology are retained. These very few retained cells constitute the Morse key cell set, inheriting all the topological information of the original tens of millions of voxel grids. This reduces the computational cost of subsequent calculations of multi-resolution zigzag persistent homology from an exponential explosion to a linearly solvable level.

[0044] It should be noted that calculating the topological features of a 3D mesh is a typical combinatorial explosion problem. For a medium-sized exploration area, the resulting cubic complex contains billions of points, edges, faces, and volumes. Directly performing algebraic topological matrix reduction on such a massive boundary matrix would consume O(N^2) computational resources. 3 An exponential explosion could cause ordinary computers and even cloud servers to instantly experience memory overflow, leading to algorithm crashes.

[0045] This invention introduces a discrete Morse reduction mechanism between topological filtering and persistent homology computation. First, it constructs a discrete gradient vector field on a massive number of cubic complexes. Under the strict mathematical premise of topological equivalence—that is, never altering the essential properties of connected components, channels, and cavities—it seeks free surfaces and redundant cell pairs that do not cause changes to the topological structure, and then pairs and eliminates them on a large scale at the algebraic level. This directly reduces the computational complexity of homology algebra from an uncomputable exponential level to a linearly solvable level.

[0046] In a specific embodiment, step S4 specifically includes: Traverse the reduced complex sequence corresponding to each resolution scalar field in the multi-resolution scalar field set, establish a complex mapping between the reduced complex sequences corresponding to adjacent resolution scalar fields, and connect the reduced complex sequences corresponding to each resolution scalar field in resolution order to construct a zigzag sequence; The persistent homology of the sawtooth sequence is calculated to obtain a set of topological feature trajectories.

[0047] The set of topological feature trajectories is used to characterize the appearance, continuation, and disappearance of topological features in dimension k as the zigzag sequence progresses.

[0048] In a specific embodiment, the specific processing logic for calculating the persistent coherence of the sawtooth sequence includes: Traverse the zigzag sequence. When traversing to the first zigzag position of the zigzag sequence, a certain topological feature appears for the first time in the corresponding reduced complex. The cell scalar value associated with the first zigzag position is recorded as the birth scalar value of the topological feature. When traversing to the second zigzag position of the zigzag sequence, the topological feature disappears in the corresponding reduced complex, and the cell scalar value associated with the second zigzag position is recorded as the death scalar value of the topological feature. Based on the pairing criterion of persistent cohomology, the birth scalar value and death scalar value of the same topological feature are paired to form a topological feature trajectory. By traversing all topological features, a set of topological feature trajectories is obtained.

[0049] Specifically, when traversing the zigzag sequence, a certain topological feature first appears in the corresponding reduced complex when the first zigzag position is reached. The reason why the topological feature appears in the corresponding reduced complex even when traversing the zigzag sequence is because, in topological data analysis, the zigzag sequence represents a macroscopic evolutionary path, and the zigzag position is an index point on this path. The topological feature to be extracted is a specific geometric entity attached to a real spatial structure container. In this algorithm, this geometric container at a specific time point is the reduced complex.

[0050] It should be noted that in traditional 3D geological modeling, the generation of 3D structures relies on a single grid resolution and a single scalar threshold. However, groundwater exploration data is often accompanied by strong ambiguity and local high-frequency noise, and spatial interpolation is prone to oscillations at the boundaries. At a single scale, a tiny local data mutation can cause the extracted groundwater channels to break or generate false connectivity artifacts in originally impermeable aquitards. This invention overcomes the limitations of single-scale modeling by introducing the cutting-edge zigzag persistent homology from algebraic topology into geological modeling. The algorithm no longer fixates on a single grid but generates multi-resolution complex sequences from fine to coarse through spatial downsampling, and establishes a strict zigzag sequence between them.

[0051] During low-level computation, the algorithm establishes a geological authenticity filter by tracking the persistence of topological homology groups at different scales. Only structures that maintain topological connectivity and porosity characteristics without abrupt changes as the grid is continuously enlarged and reduced are identified as real macroscopic geological entities. Short-lived noise that only forms occasionally at a specific resolution is instantly identified and eliminated during cross-scale tracking. This method enables the modeling system to robustly output continuous, complete, and physically accurate groundwater network structures even when faced with poor low-level data containing a large amount of interpolation errors and inversion noise.

[0052] In a specific embodiment, step S5 specifically includes: Step S51: Under the logic of filtering significant topological feature trajectories, filter the set of significant topological feature trajectories from the set of topological feature trajectories; The logic for filtering salient topological feature trajectories specifically includes: If the same topological feature trajectory has a corresponding continuation relationship between the reduced complex sequences corresponding to adjacent resolutions through scale alignment mapping, then the topological feature trajectory is determined to have cross-resolution consistency. If a topological feature trajectory cannot be mapped by scale alignment, and there is a corresponding continuation relationship between the reduced complex sequences corresponding to adjacent resolutions, then the topological feature trajectory is removed from the set of topological feature trajectories. Extract the birth and death scalar values ​​corresponding to the topological feature trajectories with cross-resolution consistency, calculate the length of the survival scalar interval, and use the length as a persistence index; Topological feature trajectories with a persistence index greater than a preset stability threshold are selected as significant topological feature trajectories. Based on cross-resolution consistency and durability indices, the stability weights of each topological feature are calculated and mapped to a predefined voxel grid to generate a structural consistency field. Step S52: Obtain the extraction threshold based on the set of significant topological feature trajectories, and extract the set of spatial structure objects from the cubic complex corresponding to each resolution under the extraction threshold. The set of spatial structure objects includes the set of connected component spatial structure objects, the set of channel skeleton spatial structure objects, and the set of closed shell spatial structure objects. The processing logic includes: When dimension k=0, connect component labeling is performed on zero-dimensional cells that meet the extraction threshold, and a set of connected component spatial structure objects is output. When dimension k=1, combine the Morse key cell set, perform skeletonization operation on one-dimensional cells that meet the extraction threshold, and output a set of channel skeleton space structure objects. When dimension k=2, boundary extraction is performed on the two-dimensional cell boundaries that meet the extraction threshold, and a set of closed shell space structure objects is output. Step S53: Using the set of significant topological feature trajectories, resolution fusion is performed on the set of spatial structure objects at each resolution. The fusion result is then verified by combining the set of Morse key cells to obtain the three-dimensional topological spatial structure of groundwater. Verification processing includes redundancy elimination and connectivity preservation; Step S54: Spatial mapping is performed between the three-dimensional topological spatial structure of groundwater and the structural consistency field to output the three-dimensional topological model of groundwater.

[0053] Specifically, the preferred selection of the preset stability threshold is the value corresponding to the top 5% of all persistence indicators arranged in descending order.

[0054] The stability weights of each topological feature are calculated based on cross-resolution consistency and durability metrics, specifically including: The number of resolution levels at which the topological feature trajectory can exist continuously in the multi-resolution scalar field set is used as the cross-resolution consistency metric. After normalizing the cross-resolution consistency metric and the durability metric, a product operation is performed to obtain the stability weight of the topological feature. The spatial structure object corresponding to the topological feature in the predefined voxel grid is obtained. Based on the scale alignment mapping, the spatial structure object is projected back into the predefined voxel grid. The stability weight is assigned to all voxel units contained in the spatial structure object to obtain the structure consistency field. The structure consistency field is consistent with the spatial resolution of the predefined voxel grid.

[0055] In step S52, performing connected component labeling on zero-dimensional cells that meet the extraction threshold specifically includes: On the graph structure consisting of zero-dimensional cells that meet the extraction threshold and their associated one-dimensional cells, a depth-first search is performed to merge all zero-dimensional cells with path connections into the same set and assign them a unique component ID. The final output connected components physically delineate the boundaries of isolated underground water sacs or lenses.

[0056] Combining the Morse key cell set, the skeletonization operation for one-dimensional cells that meet the extraction threshold specifically includes: The stable manifold tracing algorithm in Discrete Morse theory is used. Starting from the saddle point in the Morse key cell, the algorithm traces along a one-dimensional cell path that meets a threshold, against the direction of the discrete gradient vector field, until the key extremum point is reached. These traced manifold paths directly form the network centerline, perfectly outputting the underground water channel skeleton without redundant thickness.

[0057] Boundary extraction for two-dimensional cell boundaries that meet the extraction threshold specifically includes: It iterates through all 2D cells that belong to the salient topological feature trajectory of the current target and geometrically enclose it to form a 2D cavity structure. It performs topological patch stitching in computer graphics to extract boundary 2D faces shared only by one 3D cell that meets a threshold and one 3D cell that does not. These patches are stitched together to output a tightly closed, seamless polygonal mesh used to characterize the outer impermeable shell of a confined aquifer or large karst cavity.

[0058] The verification process includes redundancy elimination and connectivity preservation; redundancy elimination involves performing voxel clipping based on key cell constraints. The edges of the fused 3D voxel blocks are inspected. If any convex, suspended voxel branches or rough halos do not contain any Morse key cells, they are identified as redundant expanded pixels due to low-resolution upsampling and are forcibly removed to ensure sharp and accurate model edges.

[0059] The connectivity is maintained in the scanned and fused model. If a channel skeleton that originally belonged to the same trajectory is found to have a fault of 1 to 2 voxels after mapping, the discrete gradient vector field stored at the bottom layer of the system is called. Morphological closing operation is performed at both ends of the fault along the gradient descent direction to forcibly bridge the fault gap and ensure the absolute physical connectivity of the groundwater flow topological channel.

[0060] In a specific embodiment, obtaining the extraction threshold based on the set of significant topological feature trajectories specifically includes: Read the birth and death scalar values ​​corresponding to the trajectories with significant topological features, and generate the survival scalar interval; The common stable interval is obtained by intersecting the survival scalar intervals of all resolutions. If the common stable interval is a non-empty set, then the extraction threshold is selected from the common stable interval according to the selection rules; If the common stable interval is an empty set, within the union of the survival scalar intervals at each resolution, select the scalar value covered by the most survival scalar intervals as the candidate extraction threshold. If there are multiple candidate extraction thresholds, then the extraction threshold will be selected from the candidate extraction thresholds according to the selection rules.

[0061] Specifically, if the common stable interval is an empty set, within the union of the survival scalar intervals at each resolution, the number of times each scalar value is included in the survival scalar interval is counted, and the scalar value that is covered by the most survival scalar intervals is selected as the candidate extraction threshold.

[0062] The selection rules are as follows: For each candidate scalar value, extract the set of spatial structure objects corresponding to the salient topological feature trajectory of the target from the cubic complex corresponding to each resolution of the multi-resolution scalar field set; Based on scale-aligned mapping, the set of spatial structure objects at each resolution is mapped to the reference resolution grid. Based on the voxel coverage intersection of the mapped spatial structure objects in the reference resolution grid, a cross-resolution object correspondence is established. The reference resolution grid is preferably the grid with the lowest spatial resolution among the multi-resolution voxel grid sets. Based on the cross-resolution object correspondence, the sum of the number of topology split events and the number of merging events corresponding to each candidate scalar value is counted, and topology consistency verification is performed; The process of mapping a single spatial structure object at the same resolution into two or more independent spatial structure objects in a reference resolution grid is defined as a topology splitting event. The process of mapping two or more independent spatial structure objects at the same resolution into a single spatial structure object in a reference resolution grid is defined as a topology merging event. Topology consistency verification includes ensuring that the number of connected components in the reference resolution grid remains unchanged when the topology feature dimension k=0. When the topological feature dimension k=1, the number of endpoints and the topological relationship between the bifurcation points of the corresponding channel skeleton spatial structure object in the reference resolution grid remain unchanged. When the topological feature dimension k=2, the corresponding closed shell space structure object maintains a closed two-dimensional shell boundary in the reference resolution grid. The candidate scalar value that minimizes the sum of the number of topology split events and the number of merge events and passes the topology consistency check is selected as the extraction threshold. If the candidate scalar values ​​that meet the above conditions are still not unique, the one with the largest value is selected as the final extraction threshold.

[0063] Example 2, refer to Figure 2 This paper presents a cloud computing-based groundwater exploration data processing system, which includes an acquisition module, a mapping module, a filtering module, a calculation module, and an output module. The acquisition module is used to acquire and process the raw exploration data set, and generate a quality control voxel index set and a quality control error field set; The mapping module is used to perform membership mapping and error weight fusion on the set of quality control voxel indicators, generate the processed water content scalar field and multi-resolution scalar field set, and establish scale alignment mapping between adjacent resolutions. The filtering module is used to construct cubic complexes for each resolution scalar field in the multi-resolution scalar field set and generate a filtered complex sequence, and to perform discrete Morse reduction on the filtered complex sequence to obtain a reduced complex sequence. The computation module is used to construct a zigzag sequence from the reduced complex sequence corresponding to each resolution based on the scale alignment mapping, and to calculate the persistent homology of the zigzag sequence to obtain a set of topological feature trajectories. The output module is used to extract a set of spatial structure objects based on the set of topological feature trajectories and perform multi-resolution fusion to generate a three-dimensional topological spatial structure and structural consistency field of groundwater, and output a three-dimensional topological model of groundwater.

[0064] This invention utilizes a sawtooth persistent cohomology and discrete Morse reduction mechanism to achieve high robustness and adaptive analysis of the three-dimensional topological spatial structure of groundwater. By performing discrete Morse reduction on the filtered complex sequence for topological dimensionality reduction, cross-resolution object correspondences are established. Furthermore, topological consistency verification using topological splitting and merging events accurately isolates spatial structural object sets of different dimensions from massive local noise. The cross-resolution consistency metric and persistence index are fused and quantized into a structural consistency field, solving the modeling distortion problem in the process of spatial downsampling and multi-resolution scalar field fusion of exploration data sets.

[0065] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0066] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the protection scope of the present invention.

Claims

1. A method for processing groundwater exploration data based on cloud computing, characterized in that, Includes the following steps: Step S1: Obtain and process the original exploration data set to generate a quality control voxel index set and a quality control error field set; Step S2: Perform membership mapping and error weight fusion on the set of quality control voxel indicators to generate the processed water content scalar field and multi-resolution scalar field set, and establish scale alignment mapping between adjacent resolutions. Step S3: Construct cubic complexes for each resolution scalar field in the multi-resolution scalar field set and generate a filtered complex sequence, and perform discrete Morse reduction on the filtered complex sequence to obtain a reduced complex sequence; Step S4: Based on scale alignment mapping, construct the reduced complex sequence corresponding to each resolution into a zigzag sequence, and calculate the persistent homology of the zigzag sequence to obtain the set of topological feature trajectories; Step S5: Extract the set of spatial structure objects based on the set of topological feature trajectories and perform multi-resolution fusion to generate the three-dimensional topological spatial structure and structural consistency field of groundwater, and output the three-dimensional topological model of groundwater.

2. The groundwater exploration data processing method based on cloud computing as described in claim 1, characterized in that, Step S1 specifically includes: Step S11: Obtain the original exploration data set, which includes borehole interpretation data and geophysical inversion body data; The borehole interpretation data includes the borehole spatial coordinates and corresponding hydrogeological parameters; The geophysical inversion data includes three-dimensional grid coordinates and corresponding physical property values; Step S12: Standardize the original exploration data set to obtain the exploration data set. The standardization process includes coordinate standardization, unit standardization, and error metadata standardization. Step S13: Map the survey data set to a predefined voxel grid through spatial interpolation to obtain a set of voxel index volumes; Based on the spatial distribution characteristics and inherent errors in the exploration data set, the comprehensive error value of each voxel grid in the predefined voxel grid is calculated on a voxel-by-voxel basis, generating an error field set aligned on a voxel-by-voxel basis with the voxel index set. The spatial distribution characteristics are the three-dimensional spatial distances between sampling points; Step S14: Perform quality control operations on the voxel index set and the error field set. The quality control operations include outlier determination and isolated body removal, resulting in a quality control voxel index set and a quality control error field set.

3. The groundwater exploration data processing method based on cloud computing as described in claim 2, characterized in that, Step S2 specifically includes: Step S21: Based on the membership function, perform membership mapping on the quality control voxel indexes in the quality control voxel index set to obtain the water content membership set. Step S22: Generate a weight volume based on the quality control error field, and obtain the weight volume set. The calculation expression for the weight volume is as follows: ； in, To represent the weight volume value of the i-th type of physical index at spatial coordinates (x, y, z), As a predetermined positive number, Let be the error value after quality control corresponding to the i-th type of physical index at spatial coordinates (x, y, z). Index for physical indicator categories; Step S23: Calculate the water-bearing tendency scalar field based on the water-bearing membership set and the weight set; Step S24: Smooth and denoise the water-bearing tendency scalar field to obtain the processed water-bearing tendency scalar field; Step S25: Spatial downsampling is performed on the processed water content scalar field to generate a multi-resolution voxel mesh set and a multi-resolution scalar field set. The spatial coordinate correspondence rules in the spatial downsampling process are used as resampling relations, and scale alignment mapping between adjacent resolutions is established based on the resampling relations.

4. The groundwater exploration data processing method based on cloud computing as described in claim 3, characterized in that, The calculation expression for the scalar field of water-bearing tendency is as follows: ； in, Let be the scalar value of the water-bearing tendency scalar field at spatial coordinates (x, y, z). Let be the water content membership value of the i-th type of physical index at spatial coordinates (x, y, z).

5. The groundwater exploration data processing method based on cloud computing as described in claim 4, characterized in that, Step S3 specifically includes: Step S31: Based on the multi-resolution scalar field set, construct cubic complexes on the voxel grids corresponding to each resolution in the multi-resolution voxel grid set, map the voxel units in the cubic complexes to three-dimensional cells, and use the resolution scalar field values ​​corresponding to the voxel units as the scalar values ​​of the three-dimensional cells. The common surfaces, common edges, and vertices between voxel units are mapped to the corresponding two-dimensional cells, one-dimensional cells, and zero-dimensional cells, respectively. Step S32: Using the upper super-level set filtering method, the scalar values ​​of the three-dimensional cell are propagated to the low-dimensional cell. The processing logic includes: For a target low-dimensional cell with dimension k < 3, extract the maximum value of the target low-dimensional cell and all its adjacent three-dimensional cells, and use the maximum value as the cell scalar value. The low-dimensional cell includes two-dimensional cell, one-dimensional cell, and zero-dimensional cell; Step S33: Compare the preset threshold sequence with the cell scalar value one by one, extract the set of cells whose cell scalar value is greater than or equal to the current preset threshold, and obtain the filtered complex sequence. The filtered complex corresponding to any threshold in the preset threshold sequence satisfies: ； in, To generate a filter sub-complex at a given threshold, For cell unit, It is a cubic complex. The scalar value of the cell. For a given threshold, The logic for filtering the superlevel set; Step S34: Perform discrete Morse reduction on the filtered complex sequence to obtain the reduced complex sequence and the corresponding Morse key cell set.

6. The groundwater exploration data processing method based on cloud computing as described in claim 5, characterized in that, Step S4 specifically includes: Traverse the reduced complex sequence corresponding to each resolution scalar field in the multi-resolution scalar field set, establish a complex mapping between the reduced complex sequences corresponding to adjacent resolution scalar fields, and connect the reduced complex sequences corresponding to each resolution scalar field in resolution order to construct a zigzag sequence; The persistent homology of the sawtooth sequence is calculated to obtain a set of topological feature trajectories.

7. The groundwater exploration data processing method based on cloud computing as described in claim 6, characterized in that, The specific processing logic for calculating the persistent cohomology of a sawtooth sequence includes: Traverse the zigzag sequence. When traversing to the first zigzag position of the zigzag sequence, a certain topological feature appears for the first time in the corresponding reduced complex. The cell scalar value associated with the first zigzag position is recorded as the birth scalar value of the topological feature. When traversing to the second zigzag position of the zigzag sequence, the topological feature disappears in the corresponding reduced complex, and the cell scalar value associated with the second zigzag position is recorded as the death scalar value of the topological feature. Based on the pairing criterion of persistent cohomology, the birth scalar value and death scalar value of the same topological feature are paired to form a topological feature trajectory. By traversing all topological features, a set of topological feature trajectories is obtained.

8. The groundwater exploration data processing method based on cloud computing as described in claim 7, characterized in that, Step S5 specifically includes: Step S51: Under the logic of filtering significant topological feature trajectories, filter the set of significant topological feature trajectories from the set of topological feature trajectories; The logic for filtering salient topological feature trajectories specifically includes: If the same topological feature trajectory has a corresponding continuation relationship between the reduced complex sequences corresponding to adjacent resolutions through scale alignment mapping, then the topological feature trajectory is determined to have cross-resolution consistency. If a topological feature trajectory cannot be mapped by scale alignment, and there is a corresponding continuation relationship between the reduced complex sequences corresponding to adjacent resolutions, then the topological feature trajectory is removed from the set of topological feature trajectories. Extract the birth and death scalar values ​​corresponding to the topological feature trajectories with cross-resolution consistency, calculate the length of the survival scalar interval, and use the length as a persistence index; Topological feature trajectories with a persistence index greater than a preset stability threshold are selected as significant topological feature trajectories. Based on cross-resolution consistency and durability indices, the stability weights of each topological feature are calculated and mapped to a predefined voxel grid to generate a structural consistency field. Step S52: Obtain the extraction threshold based on the set of significant topological feature trajectories, and extract the set of spatial structure objects from the cubic complex corresponding to each resolution under the extraction threshold. The set of spatial structure objects includes the set of connected component spatial structure objects, the set of channel skeleton spatial structure objects, and the set of closed shell spatial structure objects. The processing logic includes: When dimension k=0, connect component labeling is performed on zero-dimensional cells that meet the extraction threshold, and a set of connected component spatial structure objects is output. When dimension k=1, combine the Morse key cell set and perform skeletonization operation on one-dimensional cells that meet the extraction threshold to output a set of channel skeleton space structure objects. When dimension k=2, boundary extraction is performed on the two-dimensional cell boundaries that meet the extraction threshold, and a set of closed shell space structure objects is output. Step S53: Using the set of significant topological feature trajectories, resolution fusion is performed on the set of spatial structure objects at each resolution. The fusion result is then verified by combining the set of Morse key cells to obtain the three-dimensional topological spatial structure of groundwater. The verification process includes redundancy elimination and connectivity preservation. Step S54: Spatial mapping is performed between the three-dimensional topological spatial structure of groundwater and the structural consistency field to output the three-dimensional topological model of groundwater.

9. The groundwater exploration data processing method based on cloud computing as described in claim 8, characterized in that, The extraction threshold is obtained based on the set of salient topological feature trajectories, specifically including: Read the birth and death scalar values ​​corresponding to the trajectories with significant topological features, and generate the survival scalar interval; The common stable interval is obtained by intersecting the survival scalar intervals of all resolutions. If the common stable interval is a non-empty set, then the extraction threshold is selected from the common stable interval according to the selection rules; If the common stable interval is an empty set, within the union of the survival scalar intervals at each resolution, select the scalar value covered by the most survival scalar intervals as the candidate extraction threshold. If there are multiple candidate extraction thresholds, then the extraction threshold will be selected from the candidate extraction thresholds according to the selection rules.

10. A cloud computing-based groundwater exploration data processing system, applied in any one of the cloud computing-based groundwater exploration data processing methods as described in claims 1-9, characterized in that, It includes a data acquisition module, a mapping module, a filtering module, a calculation module, and an output module; The acquisition module is used to acquire and process the original exploration data set, and generate a quality control voxel index set and a quality control error field set. The mapping module is used to perform membership mapping and error weight fusion on the set of quality control voxel indicators to generate the processed water content scalar field and multi-resolution scalar field set, and to establish scale alignment mapping between adjacent resolutions. The filtering module is used to construct cubic complexes for each resolution scalar field in the multi-resolution scalar field set and generate a filtered complex sequence, and to perform discrete Morse reduction on the filtered complex sequence to obtain a reduced complex sequence. The computation module is used to construct a zigzag sequence from the reduced complex sequence corresponding to each resolution based on the scale alignment mapping, and to calculate the persistent homology of the zigzag sequence to obtain a set of topological feature trajectories. The output module is used to extract a set of spatial structure objects based on the set of topological feature trajectories and perform multi-resolution fusion to generate a three-dimensional topological spatial structure and structural consistency field of groundwater, and output a three-dimensional topological model of groundwater.

Images