Digital Twin-Based Groundwater Infiltration Recharge Assessment Method and Equipment

Abstract

This invention discloses a method and equipment for assessing groundwater infiltration and recharge based on digital twins, belonging to the field of groundwater digital twin assessment technology. The method includes: traversing the target area for multi-source sensor data acquisition; performing groundwater infiltration simulation to generate a full-process infiltration dataset and obtain an infiltration coupling model; extracting geological structure parameters of the target area and constructing a three-dimensional geological structure model; associating and mapping the infiltration coupling model with the three-dimensional geological structure model to obtain dynamic simulation results; performing deviation analysis, automatic correction and optimization, and generating a groundwater infiltration and recharge assessment report. This invention solves the technical problems of insufficient multi-source data fusion, low spatiotemporal simulation accuracy, and weak model dynamic correction capability in existing groundwater infiltration and recharge assessment technologies. It achieves dynamic simulation of groundwater infiltration and recharge based on multi-source data coupling and automatic correction of the digital twin, thus improving the spatiotemporal accuracy of groundwater infiltration and recharge assessment.

Description

Technical Field

[0001] This invention relates to the field of digital twin assessment technology for groundwater, specifically to a method and equipment for assessing groundwater infiltration and recharge based on digital twins. Background Technology

[0002] In the field of groundwater infiltration and recharge assessment, existing technologies often face problems such as low efficiency in fusion of multi-source monitoring data (such as water level, soil moisture content, geological parameters, etc.) and insufficient accuracy in spatiotemporal scale simulation. Traditional models are difficult to effectively couple the dynamic interaction between hydrological processes and geological structures, and lack an automatic model correction mechanism based on real-time monitoring data. As a result, the assessment results cannot accurately reflect the infiltration and recharge process under complex geological conditions, making it difficult to meet the demand for high-precision and real-time assessment in refined water resource management.

[0003] Existing technologies suffer from technical problems such as insufficient fusion of multi-source data, low spatiotemporal simulation accuracy, and weak dynamic correction capabilities in groundwater infiltration recharge assessment. Summary of the Invention

[0004] This application provides a method and equipment for assessing groundwater infiltration and recharge based on digital twins, which is used to address the technical problems in existing groundwater infiltration and recharge assessments, such as insufficient fusion of multi-source data, low spatiotemporal simulation accuracy, and weak dynamic correction capability of the model.

[0005] In view of the above problems, this application provides a method and equipment for assessing groundwater infiltration recharge based on digital twins.

[0006] The first aspect of this application provides a method for assessing groundwater infiltration recharge based on digital twins, the method comprising:

[0007] Multi-source sensor data acquisition is performed across the target area to obtain a multi-source dynamic monitoring dataset. Groundwater infiltration simulation is then performed based on this dataset to generate a full-process infiltration dataset. This dataset is then coupled to obtain an infiltration coupling model. Geological structure parameters of the target area are extracted from the multi-source dynamic monitoring dataset to construct a three-dimensional geological structure model. The infiltration coupling model and the three-dimensional geological structure model are mapped together to construct a digital twin of the groundwater system. Simulation and calculation are performed based on the digital twin to obtain dynamic simulation results. Deviation analysis is conducted between the dynamic simulation results and real-time monitoring data. The digital twin is then automatically corrected and optimized based on the deviation results to generate a groundwater infiltration recharge assessment report.

[0008] A second aspect of this application provides an electronic device, comprising: a memory for storing executable instructions; and a processor for implementing a digital twin-based groundwater infiltration recharge assessment method when executing the executable instructions stored in the memory.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] Multi-source sensor data acquisition is performed across the target area to obtain a multi-source dynamic monitoring dataset. Groundwater infiltration simulation is conducted to generate a full-process infiltration dataset. This dataset is then coupled to obtain an infiltration coupling model. Geological structure parameters of the target area are extracted to construct a three-dimensional geological structure model. The infiltration coupling model and the three-dimensional geological structure model are mapped to construct a digital twin of the groundwater system. Simulation and calculation are performed based on the digital twin to obtain dynamic simulation results. Deviation analysis is conducted, and the digital twin is automatically corrected and optimized based on the deviation results to generate a groundwater infiltration recharge assessment report. This approach achieves the technical effect of realizing dynamic simulation of groundwater infiltration recharge with multi-source data coupling and automatic correction of the digital twin, thereby improving the spatiotemporal accuracy of groundwater infiltration recharge assessment. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A schematic flowchart illustrating the groundwater infiltration recharge assessment method based on digital twins provided in this application embodiment;

[0013] Figure 2 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0014] Explanation of reference numerals in the attached drawings: Input device 201, processor 202, memory 203, output device 204. Detailed Implementation

[0015] This application provides a method and equipment for assessing groundwater infiltration and recharge based on digital twins, which addresses the technical problems in existing groundwater infiltration and recharge assessments, such as insufficient fusion of multi-source data, low spatiotemporal simulation accuracy, and weak model dynamic correction capabilities.

[0016] 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 a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0017] Example 1, as Figure 1 As shown, this application provides a method for assessing groundwater infiltration recharge based on digital twins, the method comprising:

[0018] Step S100: Traverse the target area to collect data from multiple sources using multi-source sensors and obtain a multi-source dynamic monitoring dataset.

[0019] Specifically, when conducting multi-source sensing data acquisition across the target area, the hydrogeological zoning map of the target area is first retrieved. Remote sensing data is then acquired via remote sensing satellites or drones to obtain a set of remote sensing images containing information such as surface water distribution and vegetation cover. Based on the hydrogeological zoning map, the changes in stratigraphic lithology and structural zoning are analyzed to determine key spatial boundaries for topography, soil types, etc., dividing the target area into multiple monitoring grid units according to these boundaries. The remote sensing image set is then mapped to each monitoring grid unit. Image recognition technology is used to match surface features, activating multiple sensor node groups deployed within the grid, such as rainfall sensors, soil moisture sensors, and groundwater monitoring wells. Rainfall sensors are used to collect meteorological parameters such as precipitation intensity and temperature, while soil moisture sensors are used to acquire soil physical parameters such as soil moisture content and porosity. Groundwater monitoring wells are used to collect dynamic groundwater monitoring data such as water level and water quality, while surface hydrological sensors are used to collect surface hydrological parameters such as river flow and water level. A monitoring connectivity graph is constructed with the center of each monitoring grid unit as the node. The minimum path between all nodes is calculated by traversing the connectivity graph. Based on the path, the acquisition sequence of the sensor node group is planned, and the device movement trajectory is generated. The acquisition clocks of each type of sensor are spatiotemporally aligned with the geographic coordinates to construct a unified spatiotemporal reference. The movement trajectory is executed according to the reference. When the sensor node group reaches the grid center, it is triggered to synchronously collect surface hydrological parameters, meteorological parameters, soil physical parameters, and groundwater dynamic monitoring data to form an initial monitoring dataset. The initial dataset is spatiotemporally aligned, and algorithms such as Kriging interpolation are used to imput missing data, finally generating a multi-source dynamic monitoring dataset containing multi-dimensional parameters.

[0020] Step S200: Based on the multi-source dynamic monitoring dataset, perform groundwater infiltration simulation to generate a full infiltration process dataset. Couple the full infiltration process dataset to obtain an infiltration coupling model.

[0021] Specifically, when simulating groundwater infiltration based on a multi-source dynamic monitoring dataset, the soil inversion algorithm is first used to invert the time-series soil moisture content data of the target area by combining soil physical parameters and surface hydrological parameters. Meteorological parameters such as precipitation intensity and temperature are then introduced, and water movement analysis is performed on the soil moisture content time-series data based on the Richards equation to construct soil moisture movement parameters such as soil moisture diffusivity and permeability coefficient. Based on these soil moisture movement parameters, a soil hydraulic parameter field characterizing the spatial distribution of soil hydraulic properties is obtained through spatial interpolation. Surface infiltration boundary conditions are then set based on this soil hydraulic parameter field, and the HYDRUS numerical model is used to simulate groundwater infiltration from the surface to the soil moisture content. The entire process of the water layer is simulated to generate a full-process infiltration dataset containing information such as changes in soil moisture content and water transport paths at different time periods. Based on the full-process infiltration dataset, the spatiotemporal distribution parameters of soil moisture content are obtained by analyzing the soil infiltration process, and the infiltration flux parameters are obtained by analyzing the groundwater infiltration process. The groundwater recharge rate is calculated by combining the spatiotemporal distribution parameters of soil moisture content and the infiltration flux parameters, and the infiltration flux parameters and the groundwater recharge rate are spatiotemporally matched to construct a two-way feedback coupling mechanism in which surface water infiltration affects soil water distribution and soil water infiltration recharges groundwater. By iteratively correcting this coupling mechanism, an infiltration coupling model that couples the physical processes of surface water, soil water, and groundwater is finally obtained.

[0022] Step S300: Extract geological structure parameters of the target area based on the multi-source dynamic monitoring dataset and construct a three-dimensional geological structure model.

[0023] Specifically, when extracting geological structure parameters of the target area and constructing a three-dimensional geological structure model based on a multi-source dynamic monitoring dataset, the following steps are taken: First, geological analysis is performed on the physical exploration data (such as seismic wave velocity and resistivity sounding data) in the multi-source dynamic monitoring data to obtain basic information such as rock layer interfaces and fault distribution. Then, a multi-level constrained geological inversion framework is constructed, using borehole core data and well logging curves as hard constraints, combined with regional geological data to form soft constraints, to invert and constrain the geological analysis data. Following this inversion framework, multi-source data are integrated to construct a three-dimensional cube containing geological structure parameters such as lithology, porosity, and permeability. Using the geological structure parameter cube as input, GOCAD geological modeling software is used to divide geological strata using sequence stratigraphy methods. Combined with tectonic geological features, a three-dimensional geological structure model reflecting the true geological structure of the target area is constructed, achieving accurate characterization of stratigraphic spatial distribution, lithological variations, and other characteristics.

[0024] Step S400: Associate and map the infiltration coupling model with the three-dimensional geological structure model to construct a digital twin of the groundwater system, and perform simulation and calculation based on the digital twin to obtain dynamic simulation results.

[0025] Specifically, when mapping the infiltration coupling model to the three-dimensional geological structure model, the infiltration coupling model is first regularized to construct a finite difference network based on grid cells. Simultaneously, the three-dimensional geological structure model is unstructured to construct a tetrahedral network adapted to complex geological interfaces. The finite difference network is traversed and labeled in conjunction with the tetrahedral network to form multiple label sets containing node coordinates and attribute parameters. By calculating the Euclidean distance from the vertices of the tetrahedral network to the finite difference network, a distance weight matrix reflecting the degree of spatial location association is constructed. Based on the distance weight matrix and the label sets, the mapping calculation is performed on the finite difference network and the tetrahedral network to establish a hybrid spatial index mapping relationship, realizing the spatiotemporal association binding of hydrological processes and geological structures, thereby constructing a digital twin of the groundwater system. Based on this digital twin, the infiltration coupling model is driven to simulate and extrapolate the water cycle, obtain the vertical water flow, and obtain the total regional recharge by spatial integration of the vertical water flow and the cumulative recharge over time by temporal integration. At the same time, the three-dimensional geological structure model is driven, and the total regional recharge and the cumulative recharge over time are synchronized to the geological model for auxiliary simulation. Finally, the groundwater infiltration recharge covering different spatiotemporal scales is obtained and added to the dynamic simulation results to achieve a multi-dimensional quantitative characterization of the groundwater infiltration recharge process.

[0026] Step S500: Perform deviation analysis based on the dynamic simulation results and real-time monitoring data, automatically correct and optimize the digital twin based on the deviation results, and generate a groundwater infiltration and recharge assessment report.

[0027] Specifically, when performing deviation analysis based on dynamic simulation results and real-time monitoring data, real-time monitoring data of surface hydrology and soil physics in the target area are captured by edge computing nodes. This data is then compared with the dynamic simulation results output by the digital twin to perform spatiotemporal deviation analysis, generating a spatiotemporal deviation tensor containing deviation values ​​in longitude, latitude, and time dimensions. The spatiotemporal deviation tensors are arranged in descending order of absolute deviation value to construct a simulation deviation sequence. Deviation indicators are set according to geological and hydrological standards. When the deviation indicator exceeds a preset threshold (e.g., replenishment error exceeds 15%), a tiered correction strategy is triggered, first addressing the limited areas with high deviation. The differential network nodes undergo weight adjustment, and the geological parameters of the tetrahedral network are iteratively inverted. After executing the correction strategy, a digital twin optimization body is generated, which drives the water cycle and geological structure coupling simulation to form a closed-loop feedback of data acquisition-simulation-correction. Finally, based on the optimized digital twin body, the spatial distribution characteristics (such as the geographical distribution of high and low recharge areas) and temporal variation trends (such as the recharge fluctuation curves of the rainy and dry seasons) of groundwater infiltration recharge are extracted. Combined with hydrogeological analysis, a groundwater infiltration recharge assessment report containing visualization charts and data reports is generated to provide a basis for water resource management decisions.

[0028] In one possible implementation, step S100 further includes:

[0029] Step S110: Retrieve the hydrogeological zoning map of the target area, perform remote sensing acquisition on the target area, and obtain a set of remote sensing images.

[0030] Step S120: Perform a change analysis based on the hydrogeological zoning map to determine the key boundary lines for spatial division.

[0031] Step S130: Divide the target area into multiple monitoring grid units according to the spatial division key boundary lines.

[0032] Step S140: Map the remote sensing image set to the multiple monitoring grid units for traversal matching, and activate multiple sensor node groups according to the matching results.

[0033] Step S150: Multi-source sensing data is collected by the multiple sensor node groups according to a unified spatiotemporal reference to generate an initial monitoring dataset.

[0034] Step S160: Perform spatiotemporal alignment on the initial monitoring dataset, and perform missing data imputation processing on the aligned dataset to generate the multi-source dynamic monitoring dataset.

[0035] Specifically, a hydrogeological zoning map of the target area is retrieved from a geographic information database. This map details the distribution of strata lithology, geological structural features, and hydrogeological unit divisions within the area. Simultaneously, remote sensing operations are conducted on the target area using remote sensing satellites or drones equipped with multispectral imaging devices to collect remote sensing image data covering visible light, near-infrared, and other bands. After preprocessing such as radiometric and geometric corrections, a remote sensing image set containing spatial features such as surface vegetation cover, water body distribution, and topography is formed, providing basic geographic information support for subsequent regional division and monitoring node deployment.

[0036] Based on the retrieved hydrogeological zoning maps, change analysis was conducted by overlaying topographic slope data, soil type distribution data, and land cover type data. Areas with a topographic slope greater than 5° were extracted as boundaries of topographic abrupt changes. Soil type boundaries were delineated based on significant differences in soil permeability, such as between clay and sand. Simultaneously, vegetation cover abrupt change zones with a Normalized Difference in Vegetation Index (NDVI) exceeding 0.2 were identified. By combining the aforementioned slope change thresholds, soil permeability differences, and vegetation cover abrupt change characteristics, key spatial boundary lines reflecting the spatial differentiation patterns of hydrogeological parameters were determined.

[0037] Based on the determined key spatial boundaries, the spatial analysis function of a Geographic Information System (GIS) was used to divide the target area into multiple monitoring grid units with homogeneous hydrogeological characteristics. During the division process, constraints were set forth by topographic slope abrupt change lines >5°, soil permeability difference boundaries (such as the boundary between clay and sand), and vegetation cover abrupt change zones (NDVI difference >0.2) to ensure relative consistency in topography, soil type, and surface cover characteristics within each grid unit. Simultaneously, the regularity of the grid shape and monitoring efficiency were considered, ultimately forming a monitoring grid unit system covering the entire target area.

[0038] Leveraging the spatial overlay analysis capabilities of a Geographic Information System (GIS), preprocessed remote sensing image sets are spatially mapped to multiple pre-defined monitoring grid units. Image recognition algorithms are then used to traverse and match surface features (such as topography, vegetation cover type, and water distribution) within each grid unit. The Normalized Difference Vegetation Index (NDVI) in the remote sensing images is used to identify vegetation cover areas, spectral features are used to distinguish water and land boundaries, and topographic slope data is combined to divide different geomorphic units. Based on the matched surface feature information, the corresponding sensor node groups within the grid units are automatically activated. For example, soil moisture sensor groups are activated in vegetation areas with NDVI values ​​higher than 0.5, water level monitoring sensor groups are activated near water body boundaries, and slope runoff monitoring sensor groups are activated in areas with a topographic slope greater than 5°, achieving on-demand activation and precise deployment of sensor node groups.

[0039] A monitoring connectivity graph of the target area is constructed using the center position of each monitoring grid unit as nodes. Graph theory algorithms are used to calculate the minimum value of all nodes, resulting in a set of minimum paths composed of multiple optimal acquisition paths. Based on this path set, the acquisition sequence of each sensor node group is analyzed to generate the movement trajectory of the equipment within the target area. The clock systems of multiple sensor node groups are synchronized and calibrated with the Global Positioning System (GPS), and a unified spatiotemporal reference is constructed in conjunction with geographic coordinates to ensure the time accuracy and spatial positioning accuracy of data acquisition. Multiple sensor node groups execute the equipment movement trajectory according to the unified spatiotemporal reference. When they reach the center position of each monitoring grid unit, they are triggered to synchronously collect surface hydrological parameters (such as river flow and water level), meteorological parameters (such as precipitation intensity and temperature), soil physical parameters (such as water content and porosity), and groundwater dynamic monitoring data (such as water level and water quality) at fixed points. The collected multi-source data are initially organized according to time series and spatial location to generate an initial monitoring dataset containing the original monitoring data.

[0040] When performing spatiotemporal alignment on the initial monitoring dataset, the surface hydrology, meteorology, soil physics, and groundwater dynamic data collected by different sensor node groups are first spatially calibrated and time-series synchronized based on GPS coordinates and a unified timestamp, eliminating spatiotemporal misalignment caused by device clock deviations and positioning errors. For missing values ​​in the aligned data, Kriging interpolation combined with spatiotemporal correlation is used for interpolation. First, the spatial weight matrix of valid data around the missing point is calculated, and then the missing time dimension values ​​are filled by time series trend prediction. For abnormal data points, the Kalman filter algorithm is used for smoothing. Finally, the interpolated and corrected multi-dimensional monitoring data is reorganized according to a unified spatiotemporal grid to generate a multi-source dynamic monitoring dataset covering a complete spatiotemporal sequence, providing high-precision data support for subsequent groundwater infiltration simulation and digital twin modeling.

[0041] In one possible implementation, step S150 further includes:

[0042] Step S151: Construct a monitoring connectivity graph of the target area using the center positions of the multiple monitoring grid units as nodes.

[0043] Step S152: Traverse the monitored connected graph and calculate the minimum value for all nodes to obtain multiple minimum value paths.

[0044] Step S153: Analyze the acquisition sequence of the multiple sensor node groups based on the multiple minimum paths to generate the device movement trajectory.

[0045] Step S154: Align the multiple sensor node groups in time and space to construct a unified time and space reference.

[0046] Step S155: Execute the device movement trajectory for the multiple sensor node groups according to the unified spatiotemporal base. When the device reaches the center position of each monitoring grid unit, trigger the multiple sensor node groups to perform fixed-point sensing and acquisition to obtain the initial monitoring dataset.

[0047] Specifically, using the center locations of multiple monitoring grid units as nodes, the geographic coordinates of each node are extracted through a geographic information system (GIS). The Euclidean distance between nodes, or the actual travel distance considering the influence of terrain, is calculated to construct a monitoring connectivity graph of the target area. This connectivity graph represents the spatial relationships between the monitoring grid units in the form of a graph theory model. Nodes in the graph correspond to the grid center locations, and the edge weights reflect the distance or travel difficulty between nodes. This provides a topological foundation for subsequent data acquisition path planning for sensor node groups, ensuring efficient path optimization calculations based on the graph model.

[0048] Using path optimization algorithms from graph theory, the constructed monitoring connected graph is traversed. The minimum path for all nodes is calculated by solving the shortest Hamiltonian path problem. Based on the node weights of the monitoring connected graph (i.e., the spatial distance or terrain resistance parameter of the center position of each monitoring grid cell), Dijkstra's algorithm is used to find the shortest closed path that passes through each node exactly once. In the process, multiple minimum paths that meet different constraints (such as prioritizing flat terrain areas and equipment energy consumption thresholds) are generated to ensure that the sensing device can cover all monitoring grid cells with minimal movement cost, providing an optimal set of paths for subsequent data acquisition sequence planning and trajectory generation.

[0049] Based on multiple obtained minimum paths and combined with real-time obstacle monitoring data (such as the distribution of surface obstacles obtained through remote sensing images or IoT sensors), the acquisition sequence of multiple sensor node groups is dynamically analyzed. The acquisition time efficiency and energy consumption cost under different paths are calculated according to the node traversal order and path length of each minimum path. Secondly, real-time obstacle information is introduced, and path planning algorithms (such as the A* algorithm) are used to optimize the initial minimum paths for obstacle avoidance. When a real-time obstacle (such as a sudden flood or construction area) is detected on a certain path, the system automatically switches to an alternative minimum path or generates a local detour path. Finally, considering the movement characteristics of the sensor node group's equipment type (such as drones or ground monitoring vehicles) and the data acquisition timing requirements, the optimized path sequence is integrated into a device movement trajectory containing real-time obstacle avoidance logic. This trajectory not only satisfies the shortest path principle but also dynamically avoids obstacles encountered during monitoring, ensuring the continuity and security of multi-source sensor data acquisition.

[0050] The acquisition clocks of multiple sensor node groups are synchronized with the standard time source of global navigation satellite systems (such as GPS and BeiDou) to ensure that the timestamp error of each device does not exceed 10 milliseconds. Simultaneously, the positioning module of each sensor node group is calibrated based on the WGS84 coordinate system or a dedicated geographic coordinate system for the target area, and spatial coordinate accuracy is improved to sub-meter level through differential positioning technology. Based on this, a unified spatiotemporal reference framework incorporating time and spatial dimensions (longitude / latitude / elevation) is established. This provides a standardized spatiotemporal reference system for subsequently collected multi-source data such as surface hydrology and soil physics, eliminating data spatiotemporal misalignment caused by clock deviations and positioning errors, and laying the foundation for multi-source data fusion analysis and digital twin modeling.

[0051] Under a unified spatiotemporal reference, devices equipped with multiple sensor node groups (such as unmanned vehicles and drones) strictly follow the generated movement trajectory, using a real-time positioning system to calibrate the deviation between their current position and the trajectory nodes. When the device reaches the center of each monitoring grid unit, it triggers the sensor node group according to the unified time reference, simultaneously collecting surface hydrological parameters (such as river flow and water level), meteorological parameters (such as precipitation intensity, temperature, and wind speed), soil physical parameters (such as water content, porosity, and permeability), and groundwater dynamic monitoring data (such as water level, water quality, and water temperature). During the collection process, all sensor data are labeled with timestamps accurate to the second and sub-meter spatial coordinates. After initial encapsulation by edge computing units, the data is stored to form an initial monitoring dataset containing multi-dimensional raw monitoring data, providing the basic input for subsequent spatiotemporal alignment and data interpolation.

[0052] In one possible implementation, step S200 further includes:

[0053] Step S210: Based on the multi-source dynamic monitoring dataset, perform soil inversion on the target area to obtain soil moisture content time series data.

[0054] Step S220: Introduce precipitation intensity data, perform water movement analysis on the soil moisture content time series data based on the precipitation intensity data, and construct soil water movement parameters.

[0055] Step S230: Calculate the spatial distribution of the soil moisture movement parameters to obtain the soil hydraulic parameter field.

[0056] Step S240: Set the surface infiltration boundary conditions according to the soil hydraulic parameter field, and perform a full-process simulation of groundwater infiltration based on the surface infiltration boundary conditions to generate a full-process infiltration dataset.

[0057] Specifically, based on parameters such as soil bulk density, porosity, surface temperature, and precipitation from a multi-source dynamic monitoring dataset, a random forest regression model was used to retrieve soil data for the target area. First, the multi-source data underwent standardization preprocessing to eliminate the influence of dimensions. Then, a random forest model was constructed using the training set data. By integrating the prediction results of multiple decision trees, variance was reduced, improving retrieval accuracy. During model training, parameters such as node splitting criteria and the number of trees were optimized to minimize the mean square error between predicted and measured soil moisture content values. After evaluating the model using a validation set, the optimized random forest model was applied to the entire target area. Combined with a spatiotemporal interpolation algorithm, time-series soil moisture content data with a temporal resolution of 1 hour and a spatial resolution of 10 meters was obtained. This data accurately reflects the dynamic changes in soil moisture over different time periods, providing reliable data support for subsequent soil moisture movement analysis.

[0058] Precipitation intensity data (including precipitation rate and duration at different times) of the target area is introduced and spatiotemporally matched with soil moisture content time series data. Based on the Richards equation, the change of soil moisture content with precipitation process is analyzed for water movement. By calculating parameters such as water infiltration rate and matrix potential gradient in the soil, soil moisture diffusivity and permeability coefficient are constructed as soil moisture content changes, thereby quantitatively characterizing the migration law and physical properties of soil moisture during precipitation infiltration.

[0059] Based on soil moisture movement parameters (such as permeability coefficient and soil moisture diffusivity), spatial distribution is calculated using Kriging interpolation. First, a semi-variogram analysis is performed on discrete parameter points to fit nugget values, ranges, and sill values ​​to characterize spatial variability. Then, based on this, unbiased optimal estimation is performed on unsampled points in the target area, generating a continuous two-dimensional soil hydraulic parameter field. Simultaneously, validation is performed using inverse distance weighted interpolation. By setting a distance decay index and a search neighborhood range, parameters at different soil depths are interpolated in layers, ultimately constructing a three-dimensional soil hydraulic parameter field that exhibits variability in both horizontal and vertical directions, visually demonstrating the spatial heterogeneity of soil permeability and moisture retention capacity.

[0060] Based on the generated soil hydraulic parameter field, surface infiltration boundary conditions are set, including parameters such as initial soil moisture content, surface water depth, and infiltration rate. These are then imported into the HYDRUS hydrological model as boundary inputs to dynamically simulate the entire process of groundwater infiltration from the surface to the aquifer. During the simulation, the coupling effects of multiple physical processes, such as precipitation infiltration, surface runoff, soil moisture redistribution, and groundwater recharge, are fully considered. Real-time calculations of physical quantities such as soil moisture content distribution, water transport paths, and infiltration flux at different times are performed. Finally, a dataset containing both time series and spatial distribution of the entire infiltration process is generated, providing complete physical process data support for the subsequent construction of infiltration coupling models.

[0061] In one possible implementation, step S200 further includes:

[0062] Step S250: Perform soil infiltration analysis based on the infiltration process dataset to obtain the spatiotemporal distribution parameters of soil moisture content.

[0063] Step S260: Perform groundwater infiltration analysis based on the infiltration full-process dataset to obtain infiltration flux parameters.

[0064] Step S270: Calculate the infiltration recharge rate based on the spatiotemporal distribution parameters of soil moisture content and the infiltration flux parameters.

[0065] Step S280: Spatiotemporally match the infiltration flux parameters with the groundwater recharge rate to construct a two-way feedback coupling mechanism.

[0066] Step S290: Iteratively correct the bidirectional feedback coupling mechanism to construct the infiltration coupling model.

[0067] Specifically, the soil moisture content data in the infiltration process dataset is analyzed from multiple dimensions. By extracting the moisture content monitoring values ​​at different time points and soil layers at different depths, a dynamic distribution model is constructed using a spatiotemporal interpolation algorithm. This transforms the discrete moisture content data into a continuous spatiotemporal distribution field, thereby obtaining spatiotemporal distribution parameters of soil moisture content that can characterize the changing trend of soil moisture content over time (such as moisture fluctuations before and after precipitation) and the vertical stratification and horizontal differentiation characteristics in space. This provides a quantitative basis for in-depth analysis of the water transport patterns during soil infiltration.

[0068] Relevant data on the groundwater infiltration stage are extracted from the infiltration process dataset. Using hydrodynamic theories such as Darcy's law, the transport process of water in soil pores is quantitatively analyzed. By calculating the water content of soil per unit cross-sectional area per unit time, the infiltration flux parameter that can characterize the intensity of groundwater infiltration is obtained. This parameter comprehensively reflects the influence of soil permeability, hydraulic gradient and other factors on the infiltration process, providing key data support for assessing groundwater recharge capacity.

[0069] A mass conservation algorithm based on the finite difference method is employed for infiltration recharge calculation. The target area is divided into regular grid cells according to the spatial dimension, and a calculation step size is set in the time dimension. A water balance equation is established for each grid cell: Groundwater recharge rate = (soil moisture content at the end of the time period - soil moisture content at the beginning of the time period) × grid volume / time step size + infiltration flux × grid cross-sectional area. The spatiotemporal distribution parameter of soil moisture content is used to calculate the water change within the grid cell, and the infiltration flux parameter serves as the vertical water input. By iteratively calculating the recharge rate of each grid cell at different time steps, the groundwater recharge rate distribution field of the entire target area is finally generated. This algorithm balances computational accuracy and efficiency, and can accurately reflect the dynamics of groundwater recharge under spatiotemporally heterogeneous conditions.

[0070] Spatiotemporal grid indexing technology was used to match infiltration flux parameters with groundwater recharge rates. The target area was divided into a three-dimensional grid with uniform spatiotemporal resolution, with a time step of 1 hour and a spatial grid precision of 10m × 10m × 5m. By establishing a spatiotemporal correlation matrix, the infiltration flux parameters of each grid cell were determined. (Unit: m) 3 / (m 2The infiltration flux (·h) is correlated with the groundwater recharge rate (unit: m / h) for the corresponding time period. Based on this, a feedback control algorithm is introduced. When the infiltration flux of a certain grid cell increases by 10%, the impact of this change on the groundwater level is calculated using Darcy's law. This allows for the adjustment of the infiltration boundary conditions in subsequent time periods, forming a two-way feedback link between infiltration flux, recharge rate, soil moisture content, and infiltration flux. Ultimately, a two-way feedback coupling mechanism with dynamic equilibrium characteristics is constructed, enabling the coordinated evolution and mutual constraint of the two at the spatiotemporal scales.

[0071] An iterative optimization framework based on a genetic algorithm is employed, embedding a bidirectional feedback coupling mechanism into the model parameter correction process. First, a 12-dimensional parameter vector is defined, including soil permeability coefficient and water-holding curve parameters. The feedback mechanism calculates the deviation between infiltration flux and recharge rate under the current parameters, using the Nash efficiency coefficient and root mean square error as objective functions. A new generation of parameter populations is generated through selection, crossover, and mutation operations. In each iteration, the feedback mechanism adjusts the infiltration boundary conditions based on the latest parameters, resimulates, and evaluates the deviation. Iteration terminates when the parameter optimization margin is less than 1% for five consecutive generations or the NSE is greater than 0.95. Finally, through this adaptive iterative correction, an infiltration coupling model that accurately reflects the soil-groundwater interaction process is constructed, enabling dynamic simulation of groundwater infiltration and recharge under different precipitation conditions.

[0072] In one possible implementation, step S300 further includes:

[0073] Step S310: Perform geological analysis based on the multi-source dynamic monitoring dataset to obtain a physical exploration dataset.

[0074] Step S320: Apply inversion constraints based on the physical exploration dataset to construct a multi-level constrained geological inversion framework.

[0075] Step S330: Perform data fusion according to the multi-level constrained geological inversion framework to construct a geological structure parameter cube, wherein the geological structure parameter cube contains geological structure parameters.

[0076] Step S340: Use the geological structure parameters as input to perform geological modeling and construct the three-dimensional geological structure model.

[0077] Specifically, the geological sensing data such as seismic waves, resistivity, and geomagnetism in the multi-source dynamic monitoring dataset are preprocessed, environmental noise is removed by bandpass filtering, effective geological signals are identified by feature extraction algorithms, and calibration is performed in conjunction with core drilling data. After outliers are removed, a standardized physical exploration dataset is generated. This dataset contains parameters such as seismic wave velocity, resistivity distribution, and geomagnetic intensity at different depths, which can accurately reflect the geological interfaces and lithological distribution characteristics of the target area.

[0078] Based on a geophysical exploration dataset, a multi-level geological inversion framework was constructed using a hierarchical constraint strategy. First, the dielectric constant distribution of shallow (0–50 m) strata was inverted using ground-penetrating radar (GPR) profile data. By analyzing the relationship between radar wave reflection characteristics and dielectric constant, a high-resolution dielectric model of the shallow strata was established. Second, the shear wave velocity (Vs) field of the mid-deep (50–300 m) strata was inverted based on micromotion monitoring data. The shear wave velocity structure of the subsurface medium was inverted using surface wave dispersion characteristics, obtaining velocity stratification information of the mid-deep strata. Finally, the resistivity characteristics of deep (>300 m) lithology were calibrated using borehole resistivity logging data, and the correspondence between deep lithology and resistivity was established by combining prior geological knowledge. By coupling the shallow dielectric constant, mid-deep shear wave velocity field, and deep lithological resistivity characteristics with multi-level constraints, a multi-level constrained geological inversion framework from shallow to deep was constructed, providing multi-scale constraints for subsequent geological structure parameter inversion.

[0079] Based on the multi-level constrained geological inversion framework, a weighted fusion algorithm is used to fuse physical exploration data of different depths and types (such as shallow strata dielectric constant, mid-deep shear wave velocity field, and deep lithological resistivity). By setting depth weights and data reliability weights, multi-source data are mapped to a unified three-dimensional spatial grid, forming a geological structure parameter cube with a resolution of 1m×1m×1m. This cube contains geological structure parameters such as lithology, porosity, permeability, and shear wave velocity, realizing the three-dimensional spatial discretization and quantitative expression of geological parameters, and providing accurate parameter support for subsequent three-dimensional geological modeling.

[0080] Using parameters such as lithology, porosity, and permeability within a geological structure parameter cube as input, the parameters are interpolated in three-dimensional space using the GOCAD geological modeling software via the Kriging interpolation algorithm to generate a continuous geological attribute volume. Then, combined with the interpretation results of seismic reflection phase axes of geological interfaces, the stratigraphic interfaces are constructed using triangular meshing technology, ultimately forming a three-dimensional geological structure model that includes faults, strata, lithological distribution, and physical property parameters. This model can intuitively display the geological spatial distribution characteristics and physical property distribution patterns of the target area.

[0081] In one possible implementation, step S400 further includes:

[0082] Step S410: Based on the infiltration coupling model, perform regularization processing to construct a finite difference network.

[0083] Step S420: Based on the three-dimensional geological structure model, perform unstructured processing to construct a tetrahedral network.

[0084] Step S430: Traverse the finite difference network and combine it with the tetrahedral network to identify the identifiers and obtain multiple identifier sets.

[0085] Step S440: Calculate the distance weights from the vertices of the tetrahedral network to the finite difference network, and construct the distance weight matrix.

[0086] Step S450: Based on the distance weight matrix and the multiple identifier sets, perform mapping calculations on the finite difference network and the tetrahedral network to construct a hybrid spatial index mapping relationship.

[0087] Step S460: Based on the hybrid spatial index mapping relationship, the infiltration coupling model and the three-dimensional geological structure model are spatiotemporally associated and bound to construct the digital twin of the groundwater system.

[0088] Specifically, the computational domain of the infiltration coupling model is discretized in a regular manner. The three-dimensional space is divided into uniform cubic grid cells according to a uniform spatial step size (e.g., 10m × 10m × 5m). Each grid cell has the same geometric size and topology in the X, Y, and Z axes, thereby constructing a regular finite difference network. This provides a standardized spatial framework for the subsequent numerical calculation of the groundwater infiltration process, ensuring the consistency and repeatability of the calculation.

[0089] For complex geological interfaces and heterogeneous characteristics in 3D geological structure models, an unstructured mesh generation algorithm is used to discretize the geological body into a series of tetrahedral elements of varying shapes and sizes. During this process, the side lengths of the tetrahedra are automatically adjusted based on the curvature of the geological interfaces and the gradient of lithological variations. Smaller tetrahedra are used in areas with complex geological structures (such as fault zones and stratigraphic pinch-outs) to improve modeling accuracy, while larger tetrahedra are used in areas with simple geological structures to reduce computational load. Ultimately, a tetrahedral network that can accurately fit the morphology of geological interfaces and adapt to spatial variations in geological parameters is constructed.

[0090] For each regular grid cell in the finite difference network, calculate its spatial bounding box extent. Then, traverse the tetrahedral network to retrieve all tetrahedral cells that are completely or partially located within the bounding box. Extract the unique ID (identity identifier) ​​of these tetrahedral cells and mark them onto the corresponding finite difference grid cells to form a mapping relationship between finite difference grid cells and tetrahedral cell IDs. Repeat this process until all finite difference grid cells are marked, and finally obtain multiple identifier sets containing spatial mapping relationships, providing basic data for the spatial association of the two types of networks in the future.

[0091] For each vertex in the tetrahedral network, calculate its Euclidean distance to the center of each grid cell in the finite difference network. Then, convert the distance into a weighting coefficient using a Gaussian kernel function (closer distances result in higher weights). The weighting formula is as follows: Where w represents the weight coefficient, used to quantify the spatial association strength between tetrahedral network vertices and finite difference network mesh cells; the closer the distance, the greater the weight. d represents the Euclidean distance from a tetrahedral network vertex to the center of a mesh cell in the finite difference network, used to measure the spatial distance between the two types of network nodes. σ is a scale parameter used to control the rate at which the weight decays with distance, affecting the distribution range and sensitivity of the distance weight. e is the natural constant, the base of the natural exponential function, used in this formula to construct the exponential decay relationship between distance and weight. Arranging the weight coefficients of all tetrahedral vertices and finite difference mesh cells in rows and columns, a distance-weight matrix of dimension N×M is constructed (N is the number of tetrahedral vertices, M is the number of finite difference mesh cells). This matrix quantifies the spatial association strength between the two types of network nodes, providing weight parameters for subsequent mapping calculations.

[0092] Using a pre-constructed distance weight matrix and multiple identifier sets, a mapping calculation is performed between finite difference networks and tetrahedral networks. For each finite difference grid cell, based on the tetrahedral cell ID contained in its identifier set and the corresponding weight coefficients in the distance weight matrix, a weighted average method is used to calculate the attribute transfer relationship between the grid cell and the tetrahedral cell. This establishes a hybrid spatial index mapping relationship that reflects both spatial location association and attribute similarity, achieving a bidirectional mapping between regularized finite difference networks and unstructured tetrahedral networks, and providing an efficient data index structure for the spatiotemporal association of the two types of models.

[0093] By employing a hybrid spatial index mapping relationship and a spatiotemporal dual-dimensional association technique, the infiltration coupling model and the three-dimensional geological structure model are bound together. First, the finite difference mesh elements of the infiltration coupling model and the tetrahedral elements of the three-dimensional geological structure model are established spatially through index mapping. Then, based on the time step, hydrological process parameters (such as infiltration flux and recharge rate) are mapped to corresponding geological spatial locations according to the time series, forming a three-dimensional association network integrating space, time, and attributes. Specifically, a data interaction interface is developed, and spatiotemporal synchronization thresholds are set (spatial error ≤ 5 meters, time error ≤ 1 hour). When the model is running, the mapping relationship in the hybrid index is retrieved in real time, driving the bidirectional transmission of hydrological parameters and geological attributes. Ultimately, a digital twin that dynamically reflects the groundwater infiltration and recharge process is constructed, achieving high-precision simulation of the physical processes of the groundwater system.

[0094] In one possible implementation, step S400 further includes:

[0095] Step S470: Based on the digital twin, drive the infiltration coupling model to perform water cycle simulation and deduction to obtain the vertical water flow; perform spatial integration based on the vertical water flow to obtain the total regional recharge; perform temporal integration based on the vertical water flow to obtain the cumulative recharge over a period of time; drive the three-dimensional geological structure model based on the digital twin, synchronize the total regional recharge and the cumulative recharge over a period of time to the three-dimensional geological structure model for auxiliary simulation to obtain groundwater infiltration recharge with multiple spatiotemporal scales; add the groundwater infiltration recharge with multiple spatiotemporal scales to the dynamic simulation results.

[0096] Specifically, based on the constructed digital twin, the lithological parameters (such as permeability and porosity) of the three-dimensional geological structure model are spatiotemporally matched with the hydrological process parameters (such as precipitation intensity and evaporation) of the infiltration coupling model through a hybrid spatial index mapping relationship. The Richards equation is solved by the finite difference method to calculate the vertical water transport of each grid cell per unit time, generating a high-precision vertical flow field and realizing the dynamic quantification of the groundwater infiltration process.

[0097] Based on the vertical water flux field, and according to the spatial grid division of the three-dimensional geological structure model, spatial integration is performed on the vertical water flux of all grid cells in the target area or a specified sub-region. By traversing the mapping relationship between the finite difference network and the tetrahedral network, the flux value of each grid cell is obtained, in meters (m). 3 / (m 2 ·d) Multiply the product by its corresponding three-dimensional spatial volume weight, then sum them up along the horizontal and vertical directions to obtain the total supply amount covering the entire region or a specified sub-region (unit: m). 3 This enables the quantitative conversion from single-point flux to regional total flux.

[0098] Based on time-series data of vertical flux fields, the vertical flux is accumulated and integrated for each time step within specific time periods such as day, month, and year. By traversing the time-dimensional data recorded in the digital twin, the flux values ​​of each time step within the corresponding time period are superimposed in chronological order. Simultaneously, the area weight of the spatial grid is incorporated for calculation, ultimately yielding the cumulative replenishment (unit: m³) within a specific time period. 3 This enables the quantification of the time dimension from instantaneous flux to total amount over a period of time, accurately reflecting the cumulative effect of groundwater infiltration and recharge at different time scales.

[0099] Through a hybrid spatial indexing mapping mechanism using digital twins, the total regional recharge is allocated to the tetrahedral mesh (e.g., 10m×10m×5m resolution) of the 3D geological structure model according to the volume weight of geological units. Simultaneously, the cumulative recharge over a period is mapped to the corresponding geological strata according to time series (e.g., hourly / daily / monthly). Multi-scale constraints are constructed by combining the permeability tensor field and porosity distribution. A parallel computing framework is used to achieve bidirectional data transfer between hydrological parameters (e.g., recharge intensity) and geological properties (e.g., permeability coefficient). A spatiotemporal synchronization algorithm (time step error ≤ 0.1h, spatial location error ≤ 0.5m) drives the 3D seepage field simulation, generating groundwater infiltration recharge datasets covering different geological units (e.g., unconfined layers / confined layers) and different time scales (e.g., rainstorm events / seasonal changes). This achieves multi-dimensional quantification from regional total to local flux, and from long-term trends to instantaneous responses.

[0100] Based on a digital twin-driven infiltration coupling model and a three-dimensional geological structure model, the vertical water flow at different geological strata is first calculated using the finite difference method (e.g., the vertical water flow of the sand and gravel layer under a rainstorm scenario is 5.2–8.7 m). 3 / (m 2 ·d)) Then, spatial integration is performed on the vertical water flow to obtain the total regional recharge (e.g., the total regional recharge in a 24-hour rainstorm scenario is 15600m). 3 Simultaneously, time integration is performed to obtain the cumulative supply amount for a given period (e.g., the annual cumulative supply amount of 35,600 m³ in a seasonal supply scenario). 3 Next, the lithological parameters in the three-dimensional geological structure model (such as the permeability coefficient of silty clay layer of 0.85 m / d and the permeability coefficient of moderately weathered rock layer of 1.24 m / d) are spatiotemporally correlated with the above-mentioned recharge data. At different spatial resolutions from 10m×10m to 100m×100m and at different time scales from hourly to yearly, the data of vertical water flow, total regional recharge, and cumulative recharge over time are mapped to the grid nodes and tetrahedral vertices of the digital twin. Finally, the data are integrated into the dynamic simulation results in the form of three-dimensional cloud maps and spatiotemporal distribution curves.

[0101] In one possible implementation, step S500 further includes:

[0102] Step S510: Capture the real-time monitoring data of the target area in real time, perform deviation analysis between the dynamic simulation results and the real-time monitoring data, and obtain the deviation results, which include a spatiotemporal deviation tensor.

[0103] Step S520: Perform descending order analysis based on the spatiotemporal deviation tensor to construct a simulated deviation sequence, and set deviation indices according to the simulated deviation sequence.

[0104] Step S530: When the deviation index exceeds the preset deviation threshold, a graded correction strategy is triggered.

[0105] Step S540: Execute the hierarchical correction strategy to automatically correct and optimize the digital twin, and construct an optimized digital twin.

[0106] Step S550: Re-simulate and extrapolate based on the digital twin optimization body to form a data loop, evaluate according to the data loop, and generate the groundwater infiltration recharge assessment report.

[0107] Specifically, by deploying sensor arrays (such as groundwater level sensors and soil water potential sensors) in the target area, monitoring data such as water level, water content, and seepage pressure are collected in real time. The data is transmitted to a digital twin platform using IoT communication protocols. The dynamic simulation results are spatiotemporally registered with the real-time monitoring data. For each monitoring point, the deviation between the simulated and measured values ​​in the X, Y, Z spatial coordinates and time series is calculated. Then, a spatiotemporal deviation tensor containing spatial location deviation components (ΔX, ΔY, ΔZ) and time series deviation components (Δt) is constructed. This tensor quantifies the error distribution in different dimensions in matrix form, providing a multi-dimensional data foundation for subsequent deviation analysis.

[0108] The deviation values ​​of each dimension in the spatiotemporal deviation tensor (such as spatial location error, time series error, and attribute value error) are sorted in descending order to form a simulated deviation sequence. This sequence visually presents the distribution gradient of the deviation and the key error points. By statistically analyzing the cumulative distribution characteristics of the deviation values ​​in the sequence (such as the proportion of the top 10% of deviation values ​​to the total error), and in conjunction with standards in the field of groundwater simulation, deviation indicators that can characterize the overall simulation accuracy are set. Root mean square error (RMSE) and mean absolute error (MAE) are used to quantitatively clarify the degree of deviation between the simulation results and actual monitoring data.

[0109] The system compares the set deviation indicators (such as root mean square error (RMSE) and mean absolute error (MAE)) with preset deviation thresholds (such as RMSE ≤ 3% and MAE ≤ 5%) in real time. When the deviation indicator exceeds the threshold, the system automatically identifies the degree of deviation and triggers a graded correction strategy. This strategy is divided into multi-level response mechanisms (such as mild, moderate, and severe correction levels) based on the magnitude of the deviation. Different levels correspond to different correction priorities and parameter adjustment ranges. For example, for mild deviations, only local mesh parameters are optimized, while for severe deviations, the model structure reconstruction process is initiated to ensure that the correction strategy matches the error scale and achieves dynamic control of simulation accuracy.

[0110] Based on the triggered hierarchical correction strategy, the corresponding correction process is initiated according to the degree of deviation: for slight deviation, the local parameters (such as permeability coefficient and porosity) of the infiltration coupling model in the digital twin are fine-tuned using the particle swarm optimization algorithm; for moderate deviation, the mesh accuracy of the three-dimensional geological structure model is locally refined using the Kriging interpolation method (e.g., refining the 10m×10m mesh to 5m×5m), and the boundary conditions are updated using Kalman filtering; for severe deviation, the model reconstruction mechanism is initiated, the tetrahedral network topology is regenerated based on real-time monitoring data, and the model parameter set is globally optimized using a genetic algorithm, ultimately constructing a digital twin optimized body with a higher degree of matching with the actual monitoring data, realizing multi-level correction from parameter adjustment to structural optimization.

[0111] Based on the digital twin optimization model, the infiltration coupling model and the three-dimensional geological structure model are re-driven to carry out water cycle simulation. The newly generated simulation results and real-time monitoring data are then subjected to spatiotemporal deviation analysis to form a data closed loop of monitoring-deviation analysis-correction-resimulation. By evaluating the convergence trend of deviation indicators in the closed loop (e.g., RMSE decreases to below the threshold after three consecutive corrections), parameter stability (e.g., the fluctuation range of optimized parameters is ≤2%), and model prediction accuracy (e.g., the spatial correlation coefficient with measured data is ≥0.9), a groundwater infiltration recharge assessment report is automatically generated, which includes multi-temporal and spatial scale recharge distribution, visualization of the correction process, quantification of uncertainty, and prediction of future trends. This provides dynamic and reliable decision support for water resource management.

[0112] Example 2, Figure 2 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention, showing a block diagram of an exemplary electronic device suitable for implementing the embodiments of the present invention. Figure 2 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention. This electronic device is in the form of a general-purpose computing device, and its components may include, but are not limited to, an input device 201, a processor 202, a memory 203, and an output device 204. The processor 202 may be one or more; the processor 202 executes various functional applications and data processing of the computer device by running software programs, instructions, and modules stored in the memory 203, thereby realizing the aforementioned groundwater infiltration recharge assessment method based on digital twins.

[0113] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0114] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0115] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A method for assessing groundwater infiltration recharge based on digital twinning, characterized in that, The method comprises: traversing the target area to perform multi-source sensing collection to obtain a multi-source dynamic monitoring data set; based on the multi-source dynamic monitoring data set, simulating groundwater infiltration to generate an infiltration full-process data set, and coupling the infiltration full-process data set to obtain an infiltration coupling model; based on the multi-source dynamic monitoring data set, extracting geological structure parameters of the target area to construct a three-dimensional geological structure model; associating and mapping the infiltration coupling model and the three-dimensional geological structure model to construct a digital twin of the groundwater system, and performing simulation and deduction calculation according to the digital twin to obtain a dynamic simulation result; performing deviation analysis according to the dynamic simulation result and real-time monitoring data, automatically correcting and optimizing the digital twin according to the deviation result, and generating a groundwater infiltration recharge evaluation report; associating and mapping the infiltration coupling model and the three-dimensional geological structure model to construct a digital twin of the groundwater system, the method comprising: based on the infiltration coupling model, performing regular processing to construct a finite difference network; based on the three-dimensional geological structure model, performing unstructured processing to construct a tetrahedral network; traversing the finite difference network in combination with the tetrahedral network to perform identification to obtain a plurality of identification sets; calculating the distance weight of the vertex of the tetrahedral network to the finite difference network to construct a distance weight matrix; performing mapping calculation on the finite difference network and the tetrahedral network according to the distance weight matrix in combination with the plurality of identification sets to construct a hybrid space index mapping relationship; based on the hybrid space index mapping relationship, associating and binding the infiltration coupling model and the three-dimensional geological structure model in time and space to construct the digital twin of the groundwater system; performing simulation and deduction calculation according to the digital twin to obtain a dynamic simulation result, the method comprising: based on the digital twin, driving the infiltration coupling model to perform water cycle simulation and deduction to obtain vertical water flow flux; based on the vertical water flow flux, performing spatial integration to obtain regional recharge total amount; based on the vertical water flow flux, performing time integration to obtain time period cumulative recharge amount; based on the digital twin, driving the three-dimensional geological structure model, synchronizing the regional recharge total amount and the time period cumulative recharge amount to the three-dimensional geological structure model to perform auxiliary simulation, and obtaining groundwater infiltration recharge amount with multiple time and space scales; adding the groundwater infiltration recharge amount with multiple time and space scales to the dynamic simulation result.

2. The digital-twin-based groundwater infiltration recharge assessment method of claim 1, wherein, traversing the target area to perform multi-source sensing collection to obtain a multi-source dynamic monitoring data set, the method comprising: calling a hydrogeological zoning map of the target area, performing remote sensing collection on the target area to obtain a remote sensing image set; performing change analysis according to the hydrogeological zoning map to determine spatial division key boundary lines; dividing the target area into a plurality of monitoring grid units according to the spatial division key boundary lines; mapping the remote sensing image set to the plurality of monitoring grid units for traversal matching, and activating a plurality of sensing node groups according to the matching result; performing multi-source sensing collection according to a unified time and space reference through the plurality of sensing node groups to generate an initial monitoring data set; The initial monitoring data set is spatio-temporally aligned, and missing data is interpolated based on the aligned data set to generate the multi-source dynamic monitoring data set.

3. The digital-twin-based groundwater infiltration recharge assessment method of claim 2, wherein, The method comprises the following steps: A monitoring connected graph of the target area is constructed with the center positions of the plurality of monitoring grid units as nodes; Minimum value calculation is performed on all nodes by traversing the monitoring connected graph to obtain a plurality of minimum value paths; Based on the plurality of minimum value paths, the acquisition sequence of the plurality of sensor node groups is analyzed to generate a device movement trajectory; The plurality of sensor node groups are spatio-temporally aligned to construct a unified spatio-temporal reference; According to the unified spatio-temporal reference, the device movement trajectory of the plurality of sensor node groups is executed, and when the center position of each monitoring grid unit is reached, the plurality of sensor node groups are triggered to perform point sensing acquisition to obtain the initial monitoring data set.

4. The digital-twin-based groundwater infiltration recharge assessment method of claim 1, wherein, Based on the multi-source dynamic monitoring data set, groundwater infiltration simulation is performed to generate an infiltration whole-process data set, the method comprising: Based on the multi-source dynamic monitoring data set, soil inversion is performed on the target area to obtain soil moisture content time series data; Rainfall intensity data is introduced, and water movement analysis is performed on the soil moisture content time series data based on the rainfall intensity data to construct soil water movement parameters; According to the soil water movement parameters, spatial distribution calculation is performed to obtain soil hydraulic parameter fields; According to the soil hydraulic parameter fields, surface infiltration boundary conditions are set, and based on the surface infiltration boundary conditions, the whole process of groundwater infiltration is simulated to generate an infiltration whole-process data set.

5. The digital twin based groundwater infiltration recharge assessment method of claim 4, wherein, The method comprises the following steps: Based on the infiltration whole-process data set, soil infiltration analysis is performed to obtain soil moisture content spatio-temporal distribution parameters; Based on the infiltration whole-process data set, groundwater infiltration analysis is performed to obtain infiltration flux parameters; According to the soil moisture content spatio-temporal distribution parameters and the infiltration flux parameters, infiltration recharge calculation is performed to generate groundwater recharge rate; The infiltration flux parameters and the groundwater recharge rate are spatio-temporally matched to construct a two-way feedback coupling mechanism; According to the two-way feedback coupling mechanism, iterative correction is performed to construct the infiltration coupling model.

6. The digital twin-based groundwater infiltration recharge assessment method of claim 1, wherein, Based on the multi-source dynamic monitoring data set, geological structure parameters of the target area are extracted to construct a three-dimensional geological structure model, the method comprising: Based on the multi-source dynamic monitoring data set, geological analysis is performed to obtain physical exploration data sets; According to the physical exploration data sets, inversion constraints are constructed to build a multi-level constraint geological inversion framework; According to the multi-level constraint geological inversion framework, data fusion is performed to construct a geological structure parameter cube, which contains geological structure parameters; The geological structure parameters are input to perform geological modeling to construct the three-dimensional geological structure model.

7. The digital-twin-based groundwater infiltration recharge assessment method of claim 1, wherein, According to the dynamic simulation results and real-time monitoring data, deviation analysis is performed, and the digital twin is automatically corrected and optimized according to the deviation results to generate a groundwater infiltration recharge evaluation report, the method comprising: The real-time monitoring data of the target area is captured in real time, the dynamic simulation result is analyzed for deviation with the real-time monitoring data, a deviation result is obtained, and the deviation result includes a space-time deviation tensor; a descending order analysis is performed based on the space-time deviation tensor, a simulation deviation sequence is constructed, and a deviation index is set according to the simulation deviation sequence; when the deviation index exceeds a preset deviation threshold, a hierarchical correction strategy is triggered; the hierarchical correction strategy is executed to automatically correct and optimize the digital twin, and a digital twin optimization body is constructed; the simulation deduction is performed again according to the digital twin optimization body, a data closed loop is formed, an evaluation is performed according to the data closed loop, and an evaluation report of the groundwater infiltration recharge is generated.

8. An electronic device, comprising: The electronic device comprises: a memory for storing executable instructions; a processor for executing the executable instructions stored in the memory, realizing the digital twin-based groundwater infiltration recharge evaluation method in any one of claims 1 to 7.

Images