Groundwater overexploitation area dynamic evaluation method and system based on water quantity tracking

Abstract

This invention discloses a method and system for dynamic assessment of groundwater over-extraction areas based on water volume tracking, belonging to the field of groundwater assessment technology. The method includes: acquiring over-extraction area distribution data within a target area based on groundwater extraction volume, surface water infiltration volume, and meteorological evaporation; constructing a groundwater cycle model to determine zonal control boundaries and optimize control strategies; performing water balance analysis to generate an over-extraction risk evolution map within a time sliding window; and setting up a water flow scheduling network associated with the over-extraction risk evolution map for water resource allocation control. This invention solves the technical problems in existing technologies where the assessment of groundwater over-extraction areas lacks dynamism and the accuracy of water resource allocation is insufficient, leading to a lack of rationality and timeliness in over-extraction control. It achieves dynamic assessment and precise water resource allocation of groundwater over-extraction areas, improving the rationality and timeliness of over-extraction control.

Description

Technical Field

[0001] This invention relates to the field of groundwater assessment technology, specifically to a dynamic assessment method and system for groundwater over-extraction areas based on water volume tracking. Background Technology

[0002] In the field of groundwater over-extraction area management, with the increasing demand for water resources due to socio-economic development, problems such as land subsidence and aquifer depletion caused by excessive groundwater extraction are becoming increasingly prominent. Current technologies rely heavily on static data for traditional groundwater over-extraction assessments, making it difficult to capture real-time dynamic changes in groundwater extraction, surface water infiltration, and meteorological evaporation in the target area. This results in delayed updates to over-extraction area distribution data, failing to accurately reflect the actual situation of key indicators such as land subsidence rate and aquifer storage capacity at the same time point. Furthermore, existing water balance analyses often ignore the impact of abnormal land subsidence zones on over-extraction risks, leading to a lack of precise risk guidance in water resource allocation and regulation. This hinders the dynamic and refined management of over-extraction and fails to meet the current needs for sustainable management of groundwater over-extraction areas.

[0003] The existing technology lacks dynamism in the assessment of groundwater over-extraction areas and has insufficient accuracy in water resource allocation, resulting in technical problems such as a lack of rationality and timeliness in over-extraction control. Summary of the Invention

[0004] This application provides a method and system for dynamic assessment of groundwater over-extraction areas based on water volume tracking, which addresses the technical problems in existing technologies such as the lack of dynamism in groundwater over-extraction area assessment, insufficient accuracy in water resource allocation, and consequently, a lack of rationality and timeliness in over-extraction control.

[0005] In view of the above problems, this application provides a method and system for dynamic assessment of groundwater over-extraction areas based on water volume tracking.

[0006] The first aspect of this application provides a method for dynamic assessment of groundwater over-extraction areas based on water volume tracking, the method comprising:

[0007] Within the target area, based on groundwater extraction, surface water infiltration, and meteorological evaporation, over-extraction zone distribution data, including ground subsidence rate and aquifer storage variables, are obtained at the same time stamp. A groundwater cycle model is constructed based on the groundwater extraction and water demand database, and dynamic simulation and prediction are performed using the over-extraction zone distribution data to determine zoning control boundaries and optimized regulation strategies. Water balance analysis is conducted based on the surface water infiltration and meteorological evaporation to generate an over-extraction risk evolution map within a time sliding window. Based on the zoning control boundaries and optimized regulation strategies, a water flow scheduling network is established and mapped to the over-extraction risk evolution map for water resource allocation and regulation.

[0008] A second aspect of this application provides a dynamic assessment system for groundwater over-extraction areas based on water volume tracking, the system comprising:

[0009] The over-extraction area distribution data acquisition module is used to acquire over-extraction area distribution data, including ground subsidence rate and aquifer storage variables, under the same time stamp within the target area, based on groundwater extraction, surface water infiltration, and meteorological evaporation. The dynamic simulation and prediction module is used to construct a groundwater cycle model based on the groundwater extraction and water demand database, and perform dynamic simulation and prediction using the over-extraction area distribution data to determine zoning control boundaries and optimize control strategies. The water balance analysis module is used to perform water balance analysis based on the surface water infiltration and meteorological evaporation, generating an over-extraction risk evolution map within a time sliding window. The water resource allocation and control module is used to set up a water flow scheduling network associated with the over-extraction risk evolution map based on the zoning control boundaries and optimized control strategies, and to perform water resource allocation and control.

[0010] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0011] Within the target area, based on groundwater extraction, surface water infiltration, and meteorological evaporation, data on the distribution of over-extraction zones at the same time stamp are obtained. A groundwater cycle model is constructed to determine zonal control boundaries and optimize control strategies. Water balance analysis is performed based on the surface water infiltration and meteorological evaporation to generate an over-extraction risk evolution map within a time-sliding window. A water flow scheduling network mapped to the over-extraction risk evolution map is established for water resource allocation and control. This achieves the technical effect of dynamic assessment and precise water resource allocation of groundwater over-extraction zones, improving the rationality and timeliness of over-extraction control. Attached Figure Description

[0012] 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.

[0013] Figure 1 A schematic diagram of the dynamic assessment method for groundwater over-extraction areas based on water volume tracking provided in this application embodiment;

[0014] Figure 2 A schematic diagram of the structure of a dynamic assessment system for groundwater over-extraction areas based on water volume tracking, provided in an embodiment of this application.

[0015] Figure labeling: Module 10 for acquiring data on the distribution of over-extraction areas, Module 20 for dynamic simulation and prediction, Module 30 for water balance analysis, and Module 40 for water resource allocation and regulation. Detailed Implementation

[0016] This application provides a dynamic assessment method and system for groundwater over-extraction areas based on water volume tracking, which addresses the technical problems in existing technologies such as the lack of dynamism in groundwater over-extraction area assessment, insufficient accuracy in water resource allocation, and consequently, a lack of rationality and timeliness in over-extraction control.

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only 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.

[0018] Example 1, as Figure 1 As shown, this application provides a dynamic assessment method for groundwater over-extraction areas based on water volume tracking, the method comprising:

[0019] Step S100: Within the target area, based on groundwater extraction, surface water infiltration, and meteorological evaporation, obtain over-extraction zone distribution data including ground subsidence rate and aquifer storage variables under the same timestamp.

[0020] Specifically, within the target area, a multi-source water tracking network integrating multiple underground sampling wells, surface water monitoring nodes, and meteorological observation nodes is first established. This network is used to obtain groundwater extraction, surface water infiltration, and meteorological evaporation. Groundwater extraction is obtained by summing up the pumping volume of each extraction point in real time through the metering equipment of the underground sampling wells. Surface water infiltration is obtained by calculating the infiltration volume of water bodies such as rivers and lakes by the surface water monitoring nodes in combination with regional hydrogeological parameters. Meteorological evaporation is obtained by correcting the evaporation pan data of the meteorological observation nodes with regional meteorological models. Subsequently, groundwater extraction, surface water infiltration, and meteorological evaporation were marked with a unified timestamp to ensure consistency of the three types of data in the time dimension. Then, combined with the geological structure parameters of the target area, the above water data were analyzed through a hydrological calculation model to derive the ground subsidence rate and aquifer storage variable under the same timestamp. The ground subsidence rate was obtained through correlation analysis between the data monitored by the stratified subsidence meter and the water volume change, and the aquifer storage variable was obtained based on the porosity and water level changes of the aquifer. Finally, the ground subsidence rate, aquifer storage variable, and corresponding spatial location information were integrated to form the over-extraction area distribution data containing these key data.

[0021] Step S200: Based on the groundwater extraction volume and water demand database, construct a groundwater cycle model, and combine it with the over-extraction area distribution data to perform dynamic simulation and prediction, determine the zoning control boundary and optimize the control strategy.

[0022] Specifically, the process begins by collecting water demand data from various sectors within the target area, including agriculture, industry, and residential use, to establish a water demand database covering water type, water usage time period, and water volume. Then, groundwater extraction data is used as input parameters, and combined with the water distribution characteristics in the water demand database, a groundwater cycle model is constructed to simulate the entire process of groundwater recharge, runoff, and discharge. Subsequently, over-extraction zone distribution data, including land subsidence rate and aquifer storage variables, are imported into this groundwater cycle model. Simulation scenarios at different time scales, such as monthly, quarterly, and annual, are set up to dynamically simulate and predict future groundwater level changes and the expansion trend of over-extraction in the target area. Based on the simulation prediction results and combined with the regional water resource carrying capacity, lightly over-extracted, moderately over-extracted, and severely over-extracted zones are identified, and the zoning control boundaries for each zone are determined. Simultaneously, preliminary control strategies, including extraction quotas and requirements for improving water efficiency, are formulated. To improve the accuracy of the results, remote sensing image nodes and InSAR nodes will be deployed to monitor the surface deformation trend and surface water area changes in the target area, and dynamically correct the zoning control boundaries. In addition, a digital twin will be established based on the geological structure, hydrological parameters and historical mining data of the target area. Candidate control strategies will be input into a mixed control space consisting of zoning mining quotas, zoning water diversion priorities and water allocation ratios between adjacent zoning areas. The simulation results will be compared with the monitoring data. Through reward function-driven reinforcement learning, with the goal of minimizing over-extraction risk and maximizing water resource utilization efficiency, the optimal solution in the action space will be iteratively explored to finally determine the precise zoning control boundaries and optimized control strategies.

[0023] Step S300: Based on the surface water infiltration and meteorological evaporation, perform water balance analysis and generate an over-extraction risk evolution map within a time sliding window.

[0024] Specifically, based on the obtained surface water infiltration and meteorological evaporation data, a water balance analysis is conducted in conjunction with the groundwater recharge and extraction situation in the target area. The difference between the total groundwater recharge and total discharge in the area over a certain period is calculated to determine the surplus or deficit status of the groundwater system. Using a cosine similarity algorithm, the current period's land subsidence data is matched with historical data from the same period and historical high-risk periods to accurately identify abnormal land subsidence intervals. For these abnormal intervals, a risk-sensitive feature under an attention mechanism is set to quantify the contribution of surface water infiltration and meteorological evaporation to the water balance. Subsequently, a fixed-duration time sliding window is set up. Within each time step, the over-extraction risk index, calculated from indicators such as comprehensive water deficit rate, ground subsidence rate, and aquifer reservoir attenuation magnitude, is continuously updated by combining the latest water balance analysis results and risk-sensitive characteristic data. Finally, the over-extraction risk index updated at each time step is mapped to the GIS system according to spatial location, presenting the over-extraction risk level at different locations and time nodes in the target area in an intuitive form, and generating an over-extraction risk evolution map within the time sliding window.

[0025] Step S400: Based on the zoning control boundary and optimized regulation strategy, set up a water flow scheduling network that is associated with the over-extraction risk evolution map to carry out water resource allocation regulation.

[0026] Specifically, based on the established zoning control boundaries and optimized regulation strategies, the work involves linking the over-extraction risk evolution map generated within a time-sliding window. This involves spatially matching the over-extraction risk levels (low, medium, and high) of the target area at different times and locations presented in the over-extraction risk evolution map with the zoning control boundaries. This clarifies the risk distribution characteristics and core water supply-demand contradictions of each control zone (slightly, moderately, and severely over-extracted) at different time points. Next, combining the zoning extraction quotas, zoning water diversion priorities, and water allocation ratios between adjacent zones as specified in the optimized regulation strategies, a water flow scheduling network dynamically linked to the over-extraction risk evolution map is designed. For example, for control zones with high over-extraction risk and sufficient surface water availability, a dedicated surface water transfer channel is planned to enhance replenishment. For adjacent low-risk and high-risk zones, cross-zone water allocation channels are established to achieve water resource surplus and deficit complementarity. During the operation of the water flow scheduling network, the scale, direction, and timing of water transfer will be dynamically adjusted based on the real-time updates of the over-extraction risk evolution map, i.e., the risk level of a certain control zone increases or decreases. This ensures that water resource allocation is always precisely matched with the needs of over-extraction risk control, and ultimately achieves the allocation and regulation of water resources in the target area, effectively alleviating the problem of groundwater over-extraction.

[0027] In one possible implementation, step S100 further includes:

[0028] Step S110: Set up a multi-source water tracking network, which integrates multiple underground collection wells, surface water monitoring nodes and meteorological observation nodes.

[0029] Step S120: Determine the groundwater extraction volume, surface water infiltration volume, and meteorological evaporation volume based on the multi-source water tracking network.

[0030] Specifically, within the target area for dynamic assessment of groundwater over-extraction zones, a comprehensive, interconnected multi-source water tracking network is constructed. The core of this network lies in integrating three types of key monitoring nodes to achieve comprehensive collection of regional water-related data. Multiple underground sampling wells are deployed based on the hydrogeological conditions and groundwater extraction distribution characteristics of the target area, ensuring coverage of major extraction areas, recharge areas, and transition zones. Each well is equipped with high-precision metering equipment and a water level monitoring device to capture real-time pumping volume and groundwater level changes at extraction points. Surface water monitoring nodes are primarily deployed along riverbanks, around lakes, and along reservoir banks within the target area, where surface water is abundant. Auxiliary monitoring points are also added in major infiltration areas, such as areas with exposed permeable rock layers and vegetated areas, collecting surface water runoff data and infiltration-related parameters using flow monitoring instruments and soil moisture sensors. Meteorological observation nodes are deployed in conjunction with the distribution of regional meteorological stations, with multiple small automatic weather stations evenly distributed throughout the target area to monitor meteorological elements such as precipitation, temperature, humidity, and wind speed. The three types of nodes are connected to the data transmission network through wireless communication modules to form a unified multi-source water volume tracking network, enabling real-time aggregation and sharing of monitoring data.

[0031] Based on the established multi-source water tracking network integrating multiple underground sampling wells, surface water monitoring nodes, and meteorological observation nodes, targeted methods are used to determine groundwater extraction, surface water infiltration, and meteorological evaporation: For groundwater extraction, high-precision metering equipment equipped in each underground sampling well within the multi-source water tracking network is used to record the instantaneous and cumulative pumping volumes at each extraction point in real time. The extraction data from each sampling well is then aggregated to the data processing center via network data transmission, and the total groundwater extraction volume for the target area is calculated statistically. For surface water infiltration, river runoff data collected by surface water monitoring nodes within the network is used. The data includes lake water level changes and soil moisture content. Combined with preset hydrogeological parameters for the target area, such as aquifer permeability coefficient and soil porosity, the infiltration rate between surface water and soil and aquifer is calculated using hydrodynamic formulas. This allows for the estimation of surface water infiltration during a specific time period. For meteorological evaporation, real-time meteorological data such as temperature, humidity, wind speed, and solar radiation intensity are obtained from meteorological observation nodes in the network. A standardized evaporation calculation model and a modified Penman-Montis formula are used to process the meteorological data. Error correction is performed by combining the baseline evaporation data from evaporation pan monitoring. Finally, the meteorological evaporation of the target area is determined.

[0032] In one possible implementation, step S200 further includes:

[0033] Step S210: Deploy remote sensing image nodes and InSAR nodes to monitor the surface deformation trend and surface water area changes in the target area, and dynamically correct the zoning control boundary.

[0034] Specifically, to address the need for dynamic correction of zoning control boundaries in the target area, remote sensing image nodes and InSAR nodes are first deployed within the region. The remote sensing image nodes periodically acquire high-resolution satellite remote sensing images or UAV aerial images to capture spatial distribution changes of surface water bodies such as rivers, lakes, and reservoirs within the target area, accurately identifying the expansion or contraction of surface water areas. The InSAR nodes utilize synthetic aperture radar interferometry to continuously monitor minute surface deformations in the target area. Through interferometric processing of multiple radar images, surface deformation field data is generated, thereby extracting the rate and extent of surface subsidence or uplift and clarifying surface deformation trends. Subsequently, the surface water area change data acquired by the remote sensing image nodes and the surface deformation trend data acquired by the InSAR nodes are correlated with the preliminary zoning control boundaries determined using a groundwater cycle model. The analysis focuses on comparing whether surface deformation within light, moderate, and severe over-extraction zones matches the degree of over-extraction, and whether changes in surface water area affect water resource replenishment around the over-extraction areas. If significant surface subsidence exceeding model predictions is found within an over-extraction zone, or if the shrinkage of surface water area leads to a substantial decrease in the zone's recharge, the original zone control boundary will be expanded or adjusted based on these actual monitoring data. If surface deformation in a region tends to stabilize and surface water recharge is sufficient, the control boundary range will also be adjusted accordingly. This achieves dynamic optimization of the zone control boundary, ensuring that it accurately matches the actual hydrogeological conditions and over-extraction development trend of the target area.

[0035] In one possible implementation, step S210 further includes:

[0036] Step S211: Based on the remote sensing image nodes and InSAR nodes, verify the coupling between the surface deformation funnel and the water level contour lines, and identify the funnel boundary.

[0037] Step S212: Set the funnel boundary feature vector through the funnel boundary.

[0038] Step S213: Based on the funnel boundary feature vector, the partition control boundary is expanded and modified.

[0039] Specifically, the work utilizes deployed remote sensing image nodes and InSAR nodes as monitoring platforms. Remote sensing image nodes acquire high-resolution surface images to help capture surface features related to groundwater, such as surface water distribution and vegetation growth within the target area. InSAR nodes employ synthetic aperture radar interferometry to continuously monitor the target area, generating high-precision surface deformation data by processing multiple radar images. This allows for the accurate location of surface deformation funnels formed by groundwater extraction—specifically, depressions with significant localized surface subsidence. Simultaneously, groundwater level monitoring data acquired through a multi-source water tracking network is used to create contour maps of the target area, clearly showing the spatial distribution differences in aquifer levels. From these maps, an inverted cone-shaped depression caused by groundwater extraction is identified. This area exhibits the core characteristics of a funnel boundary with a significantly higher water level than the surrounding area, a hydraulic gradient pointing towards the depression center, and groundwater converging towards the center. Subsequently, the surface deformation funnel identified by the InSAR node was coupled with the inverted cone-shaped depression area in the water level contour map for verification. By comparing the degree of overlap between the two in terms of spatial location and morphological range, as well as the consistency between the surface deformation trend and the water level decline trend, the accuracy of the inverted cone-shaped depression area was further confirmed. Finally, the boundary of the funnel was accurately identified, and the specific spatial range of the inverted cone-shaped depression area of ​​the aquifer water level corresponding to the boundary was clarified.

[0040] The work was based on the identified funnel boundary, which is the spatial boundary of an inverted cone-shaped depression formed by groundwater extraction. Its core characteristics are a significantly higher water level depth than the surrounding area, a hydraulic gradient pointing towards the center of the depression, and groundwater converging towards the center. First, core indicator parameters characterizing the key attributes of this funnel boundary were extracted. These parameters specifically include the boundary closure area, the central water level depth, and the boundary expansion rate. After extracting and quantifying these core indicator parameters, they were integrated according to a pre-defined dimensional order to construct a mathematical vector that comprehensively and accurately describes the unique characteristics of the funnel boundary. This completes the setting of the funnel boundary feature vector, providing a precise quantitative basis for subsequent adjustments to the expansion and change of the zonal control boundary based on this feature vector.

[0041] Based on the established funnel boundary feature vectors, including the boundary closure area, central water level depth, and boundary expansion rate, the expansion correction of the zonal control boundary is achieved using a support vector regression algorithm combined with a spatial interpolation algorithm. First, multiple sets of funnel boundary feature vectors acquired in historical periods and corresponding manually verified zonal control boundary expansion data are used as training samples and input into the support vector regression model. The penalty parameter C and kernel function parameter γ of the model are optimized using a grid search method to construct a mapping relationship model between funnel boundary features and control boundary expansion magnitude. This allows the model to accurately predict the theoretical expansion magnitude of the zonal control boundary corresponding to a newly input funnel boundary feature vector. Subsequently, combined with GIS spatial data of the target area, the expansion magnitude predicted by the support vector regression model is used as a constraint. Kriging interpolation is then used to interpolate the spatial grid cells surrounding the original zonal control boundary to determine whether each grid cell belongs to the expansion area that needs to be included in the control boundary. Grid cells with an expansion probability higher than a preset threshold (e.g., 0.8) in the interpolation results are determined to be within the range that needs to be included in the control boundary. Finally, the spatial coordinates corresponding to these grid units are integrated to generate the corrected zoning control boundary. This ensures that the corrected boundary can accurately match the actual diffusion trend of over-extraction risk based on the funnel boundary feature vector, thereby improving the dynamic adaptability of the control boundary.

[0042] In one possible implementation, step S212 further includes:

[0043] Step S2121: Extract the core index parameters of the funnel boundary, including the boundary closure area, the central water level depth, and the boundary expansion rate.

[0044] Step S2122: Set the feature vector of the funnel boundary using the core index parameters of the funnel boundary.

[0045] Specifically, the core indicator parameters are extracted by taking the identified funnel boundary as the target. For the boundary closure area, using GIS spatial analysis tools, the vector graphics of the funnel boundary are imported into the system. Spatial topology calculation is then used to calculate the area of ​​the inverted cone-shaped depression enclosed by the boundary, accurately obtaining the specific value of the boundary closure area to reflect the spatial coverage scale of the funnel. For the central water level depth, combined with water level monitoring data from underground collection wells in the multi-source water tracking network, the geometric center of the inverted cone-shaped depression area of ​​the funnel is first determined using a spatial interpolation algorithm. Then, real-time water level data from this center location and surrounding adjacent underground collection wells are retrieved to calculate the groundwater level depth at this center location, quantifying the development depth of the funnel and the degree of over-extraction impact. For the boundary expansion rate, surface deformation data monitored by InSAR nodes and water level correlation feature data captured by remote sensing image nodes at different time points are integrated. By comparing the spatial position changes of the funnel boundary at adjacent time points, linear regression analysis is used to calculate the average distance the funnel boundary expands outward per unit time, thereby determining the boundary expansion rate and judging whether the funnel's development trend is continuous expansion, stabilization, or contraction, providing a comprehensive and accurate quantitative indicator basis for subsequent feature vector settings.

[0046] Based on the extracted core index parameters of the funnel boundary, a feature vector for the funnel boundary is set. The specific implementation method is as follows: First, the three types of core index parameters are preprocessed. The Min-Max standardization method is used to transform the boundary closure area, the central water level depth, and the boundary expansion rate to the [0, 1] interval to eliminate the influence of the difference in the dimensions of different parameters on the construction of the feature vector. For example, the boundary closure area of ​​a certain funnel is 25 km². 2 The standardized calculation is performed using the formula "(Actual area - Minimum funnel area) / (Maximum funnel area - Minimum funnel area)". Next, relying on the feature vector generation module in the data processing system, the pre-processed parameter values ​​are sequentially used as the three components of a vector, following a fixed dimensional order of "Boundary closure area - Center water level depth - Boundary expansion rate", constructing an ordered numerical array in the form of [Standardized closure area value, Standardized water level depth value, Standardized expansion rate value]. Finally, the format of this numerical array is compared with that of historical funnel boundary feature vectors of similar types to confirm that there are no abnormal missing values ​​or values ​​exceeding reasonable ranges, thus completing the setting of the funnel boundary feature vector.

[0047] In one possible implementation, step S200 further includes:

[0048] Step S220: Establish a digital twin based on the geological structure, hydrological parameters and historical mining data of the target area.

[0049] Step S230: Based on the digital twin, input candidate control strategies, compare simulation results with monitoring data, and obtain the optimized control strategy through reinforcement learning.

[0050] Specifically, geological structural data, such as lithology, aquifer thickness, aquitard distribution, and fault location, of the target area are obtained through geological drilling and ground-penetrating radar detection. This data is then transformed into a three-dimensional geological vector model using GIS technology. Secondly, relying on a multi-source water tracking network, hydrological parameters such as groundwater level, surface water infiltration, and meteorological evaporation are collected in real time via sensors. Historical monitoring data is then compiled to form a time-series hydrological database. Thirdly, groundwater extraction records from various sectors, including agriculture, industry, and domestic use, over the past 10-30 years are collected in the target area, including well coordinates, extraction volume, and extraction time periods, to construct a structured historical extraction database. Subsequently, the three-dimensional geological vector model, time-series hydrological database, and historical extraction database are imported into a digital twin modeling platform. The finite difference method is embedded, and the model output is compared with historical monitoring data. Model parameters, such as permeability coefficient and specific yield, are adjusted to reduce deviations. Finally, a digital twin highly consistent with the physical environment and hydrological processes of the target area is constructed. This twin can receive real-time updates from monitoring data and dynamically simulate changes in the groundwater system.

[0051] The scope and form of candidate control strategies are clearly defined. The action space associated with the candidate control strategies is defined as a mixed control space composed of zonal extraction quotas, zonal water diversion priorities, and water allocation ratios between adjacent zonal zones. Each dimension of the action space corresponds to specific control parameters, such as setting the extraction quota of a certain agricultural over-extraction area at 400,000 cubic meters per year, prioritizing urban water diversion over suburban areas, and setting the water allocation ratio between adjacent A / B zones at 1:0.8, thus forming multiple sets of candidate control strategies with different parameter combinations. Next, these candidate control strategies are input into a digital twin one by one to obtain the simulation results corresponding to each strategy, including key indicator data such as the groundwater level change trend of each zone in the target area, the over-extraction risk index, and water resource utilization efficiency. Subsequently, real-time monitoring data from a multi-source monitoring network, including underground sampling wells, surface water monitoring nodes, meteorological observation nodes, remote sensing image nodes, and InSAR nodes, are collected. The simulation results are compared with the monitoring data, and the deviation between the two is quantified by calculating indicators such as mean square error and absolute error. The smaller the deviation, the higher the adaptability of the candidate strategy to the actual situation. Building upon this foundation, a reinforcement learning algorithm is introduced to minimize over-extraction risk. Based on whether the over-extraction risk index is below a preset safety threshold and with maximizing water resource utilization efficiency as the multi-objective optimization guideline, a reward function is designed. When the simulation results of a candidate strategy deviate little from the monitoring data and meet the conditions of low over-extraction risk and high water resource utilization efficiency, the strategy is given a high positive reward; conversely, if the deviation is large or the optimization objective is not achieved, a penalty is imposed. Through iterative exploration of the action space using the reinforcement learning algorithm, the parameter combinations of candidate control strategies are continuously adjusted. Simulation verification and deviation correction are repeatedly performed in a digital twin, gradually selecting the strategy with the smallest deviation between the simulation results and actual monitoring data that optimally balances over-extraction risk and water resource utilization efficiency, ultimately yielding the optimized control strategy.

[0052] In one possible implementation, step S230 further includes:

[0053] Step S231: The action space associated with the candidate control strategy is defined as a mixed control space consisting of the zoned mining quota, the zoned water diversion priority, and the water allocation ratio of adjacent zones.

[0054] Step S232: Driven by the reward function, with the goal of minimizing over-extraction risk and maximizing water resource utilization efficiency as the multi-objective optimization guide, the optimal solution of the action space is explored iteratively through reinforcement learning to obtain the optimized control strategy.

[0055] Specifically, to accurately define the operational scope of candidate control strategies, the associated action space is defined as a mixed control space including zoned extraction quotas, zoned water diversion priorities, and water allocation ratios between adjacent zones. Zoned extraction quotas target different over-extraction levels within the target area, such as light, moderate, and severe over-extraction zones, setting upper limits on groundwater extraction per unit time for each zone. For example, the extraction quota for lightly over-extracted zones could be set at 250,000-350,000 cubic meters per quarter, and for severely over-extracted zones at 80,000-150,000 cubic meters per quarter, using quantitative indicators to limit the extraction intensity of each zone. Zoned water diversion priorities are tiered based on each zone's water use attributes, over-extraction risk level, and water scarcity status. For example, urban areas with concentrated domestic water use and over-extraction zones are prioritized. Areas with extremely high risk and requiring urgent water replenishment are designated as Level 1 water diversion priority, ensuring priority for their surface water or external water transfer supply. Areas with slight over-extraction, primarily for agricultural irrigation, are designated as Level 2 or 3 priority, with limited water resources allocated rationally. The water allocation ratio between adjacent zones is determined for geographically adjacent and hydrologically connected control zones, defining the range of water resources that can be allocated between them. For example, when Zone A has a surface water surplus while Zone B is severely over-extracted, the water allocation ratio from Zone A to Zone B is set at 12% to 28%, achieving complementary water resources within the region. These three dimensions together constitute a hybrid control action space covering extraction control, water diversion scheduling, and inter-regional allocation, providing a clear and quantifiable parameter range and operational direction for subsequent reinforcement learning algorithms to explore optimal control strategies.

[0056] Guided by the dual core principles of minimizing over-extraction risk and maximizing water resource utilization efficiency, this research employs a multi-objective optimization approach. A reward function drives the reinforcement learning process, iteratively exploring the optimal solution in the action space to obtain optimized control strategies. First, a multi-dimensional reward function is constructed: the over-extraction risk index is compared to a preset safety threshold. If the index falls below the threshold in the simulation results, a positive basic reward is assigned, with additional reward weight added for every 10% decrease. Simultaneously, water resource utilization efficiency is compared to a preset high-efficiency threshold. If it exceeds the threshold, a positive basic reward is also assigned, with additional reward weight added for every 5% increase in efficiency. If both over-extraction risk and water resource utilization efficiency meet the targets, a superimposed reward mechanism is triggered. Subsequently, reinforcement learning iterations are initiated: in the initial stage, candidate control strategies are randomly selected from the action space, namely, zonal extraction quotas, zonal water diversion priorities, and adjacent zonal water allocation ratios. These strategies are input into a digital twin to generate simulation results, and the corresponding reward values ​​are calculated based on the reward function. Based on the reward value feedback, an ε-greedy algorithm is used to adjust the strategy parameters, such as reducing the extraction quota in over-extraction areas, increasing the water diversion priority in water-scarce areas, and optimizing the allocation ratio between adjacent zonal areas. New candidate strategies are generated and simulated again for verification. Through iteration, the scope of exploration is gradually narrowed. When the reward value tends to stabilize and reaches the preset optimal threshold in multiple consecutive iterations, the exploration is stopped, and the corresponding candidate control strategy is determined as the optimized control strategy to ensure that it can maximize the efficiency of water resource utilization while effectively reducing the risk of over-extraction.

[0057] In one possible implementation, step S300 further includes:

[0058] Step S310: Match ground subsidence anomaly intervals using cosine similarity and set risk-sensitive features under the attention mechanism.

[0059] Step S320: Based on the surface water infiltration and meteorological evaporation, and combined with the risk-sensitive characteristics, quantify their contribution to the water balance, and set the over-extraction risk evolution map.

[0060] Specifically, ground subsidence data from multi-source monitoring devices within the target area is first collected, including high-precision surface deformation data captured by InSAR nodes and ground subsidence rate data from over-extraction area distribution data. This data is then organized into standardized data vectors by timestamp. Simultaneously, based on the geological conditions, historical subsidence patterns, and groundwater extraction background of the target area, a baseline subsidence vector reflecting normal ground subsidence status is pre-defined. Subsequently, a cosine similarity algorithm is used to calculate the similarity value between the actual ground subsidence vector and the baseline subsidence vector for each area. If the similarity value for a certain area is lower than a set threshold, such as 0.7, it indicates a significant deviation from normal subsidence conditions, and this area is identified as an abnormal ground subsidence zone. The lower the similarity value, the higher the degree of abnormality. After identifying the abnormal regions, an attention mechanism is introduced. Based on factors such as the degree of subsidence anomaly, whether the spatial distribution density is concentrated to form a subsidence funnel, and the spatial correlation with groundwater extraction wells, different attention weights are assigned to different abnormal regions: regions with high anomaly, close to the core mining area and showing an accelerating subsidence trend are assigned high weights, while regions with low anomaly, far from the mining area and showing a stabilizing subsidence trend are assigned low weights. In this way, a risk-sensitive feature under the attention mechanism that can highlight high-risk areas is constructed.

[0061] By utilizing surface water monitoring nodes and meteorological observation nodes within a multi-source water tracking network, surface water infiltration and meteorological evaporation data at different time stamps were collected for the target area. After data cleaning, extreme values ​​were removed, and missing values ​​were filled in to form a basic water quantity dataset. Next, risk-sensitive features under the obtained attention mechanism were introduced, including attention weights for each region, ranging from 0 to 1. Higher weights correspond to areas with high levels of land subsidence anomalies and a high potential for over-extraction. A "risk-water contribution quantification model" was constructed: This model uses surface water infiltration and meteorological evaporation as input variables, with risk-sensitive feature weights as adjustment factors. After formula correction, infiltration contribution = surface water infiltration × (1 + risk-sensitive feature weight), and evaporation contribution = meteorological evaporation × (1 + risk-sensitive feature weight). This quantifies the actual impact of these two water quantity elements on water balance in different regions. For regions with high risk-sensitive feature weights, the mitigating effect of surface water infiltration on over-extraction risk and the aggravating effect of meteorological evaporation on over-extraction risk are amplified, ensuring that water balance analysis better reflects regional risk differences. Subsequently, based on the corrected infiltration and evaporation contributions, a time-sliding window was calculated for each zone, with a water balance difference within a 3-month window, i.e., corrected infiltration contribution - corrected evaporation contribution - groundwater extraction. A mapping model between the water balance difference and the over-extraction risk index was then constructed: by collecting historical water balance data of the target area and corresponding over-extraction risk levels (e.g., low, medium, and high risk), a linear regression algorithm was used to fit the relationship between the two, determining the correspondence rules between the difference interval and the risk index, such as a difference ≤ -100,000 m³. 3 Corresponding risk index: 0.8-1.0, -100,000 m 3 <Difference ≤ 0 corresponds to 0.4-0.8, and difference > 0 corresponds to 0-0.4. Based on this rule, the over-extraction risk index for each partition within each window is calculated. Finally, the over-extraction risk indices for each window are sorted by time series and linked to the partitioned spatial grid of the GIS system. Through color-coded mapping (e.g., green for 0-0.4, yellow for 0.4-0.8, and red for 0.8-1.0), the spatial distribution and changing trends of over-extraction risk in each partition at different times are visually presented, completing the setting of the over-extraction risk evolution map.

[0062] In one possible implementation, step S320 further includes:

[0063] Step S321: Update the over-extraction risk index in a rolling manner within each time step using a time sliding window.

[0064] Step S322: Map the over-extraction risk index obtained from the rolling update to the GIS system to obtain the over-extraction risk evolution map.

[0065] Specifically, by combining the changing cycle of groundwater over-extraction risk in the target area with the data update frequency, the parameters of the time sliding window are determined. The window duration is set to 3 months to cover quarterly hydrological changes, such as fluctuations in precipitation and irrigation water use, and the time step is set to 1 month to ensure timely capture of risk dynamics each month. Subsequently, when the window starts rolling at each time step, the latest basic data within that window period is automatically collected, including real-time monitoring values ​​of surface water infiltration and meteorological evaporation newly added by the multi-source water tracking network, as well as updated data of risk-sensitive characteristics, such as the attention weights of abnormal land subsidence intervals after adjustment. Based on this latest data, the corrected water balance difference for each zone within the window is recalculated, and then, according to the preset "water balance difference-over-extraction risk index mapping model," the over-extraction risk index for each zone in the current period is obtained. Meanwhile, to avoid sudden changes in the risk index due to single data fluctuations, a weighted rolling mechanism is adopted, assigning 60% weight to the new data in the current period and 20% weight to the historical data in the previous two windows. The final over-extraction risk index for the current period is obtained by weighted averaging, realizing smooth updates of the index over time and ensuring that it can continuously and accurately reflect the dynamic changes in the over-extraction risk of each region.

[0066] The over-extraction risk index data obtained through rolling updates is standardized in format. The index is categorized according to the administrative divisions or grid units of the target area, such as 1km×1km grids, ensuring that each spatial unit corresponds to a unique over-extraction risk index value. Simultaneously, basic geographic information for each spatial unit is supplemented, such as latitude and longitude range, watershed, and main land use type. Subsequently, the data interaction interface with the GIS system is activated, and the standardized spatial unit-over-extraction risk index data is imported into the GIS platform in batches. Data association rules are established, treating the over-extraction risk index as attribute data and spatially associating it with spatial vector layers of the target area in the GIS system, such as administrative boundary layers and hydrological zoning layers, so that each vector graphic unit can load the corresponding risk index attribute. Next, a visualization mapping scheme for the risk index is designed: based on the over-extraction risk index value range of 0-1, it is divided into three levels: low risk (0-0.4), medium risk (0.4-0.8), and high risk (0.8-1.0), each matched with a gradient of green, yellow, and red colors, with higher risk levels showing higher color saturation, thus visually distinguishing the risk differences between different areas. Finally, by using the spatiotemporal rendering function of the GIS system, the over-extraction risk index data at different time steps are integrated according to the time series to generate a dynamic over-extraction risk evolution map. This map can switch the risk distribution status of different periods through the time axis control, clearly showing the changing trend of over-extraction risk of each spatial unit over time.

[0067] Example 2 is based on the same inventive concept as the dynamic assessment method for groundwater over-extraction areas based on water volume tracking in the previous examples, such as... Figure 2As shown, this application provides a dynamic assessment system for groundwater over-extraction areas based on water volume tracking. The system and method embodiments in this application are based on the same inventive concept. The system includes:

[0068] The over-extraction area distribution data acquisition module 10 is used to acquire over-extraction area distribution data, including ground subsidence rate and aquifer storage variables, under the same timestamp within the target area, based on groundwater extraction, surface water infiltration, and meteorological evaporation.

[0069] The dynamic simulation and prediction module 20 is used to construct a groundwater cycle model based on the groundwater extraction volume and water demand database, and to perform dynamic simulation and prediction in conjunction with the over-extraction area distribution data to determine the zoning control boundary and optimize the control strategy.

[0070] The water balance analysis module 30 is used to perform water balance analysis based on the surface water infiltration and meteorological evaporation, and generate an over-extraction risk evolution map within a time sliding window.

[0071] The water resources allocation and regulation module 40 is used to set up a water flow scheduling network that is associated with the over-extraction risk evolution map based on the zoning control boundary and the optimized regulation strategy, and to carry out water resources allocation and regulation.

[0072] Furthermore, the system is also used to implement the following functions:

[0073] A multi-source water volume tracking network is set up, which integrates multiple underground collection wells, surface water monitoring nodes, and meteorological observation nodes; based on the multi-source water volume tracking network, the groundwater extraction volume, surface water infiltration volume, and meteorological evaporation volume are determined.

[0074] Furthermore, the system is also used to implement the following functions:

[0075] Deploy remote sensing image nodes and InSAR nodes to monitor the surface deformation trend and changes in surface water area of ​​the target area, and dynamically correct the zoning control boundary.

[0076] Furthermore, the system is also used to implement the following functions:

[0077] Based on the remote sensing image nodes and InSAR nodes, the surface deformation funnel and water level contour lines are coupled and verified to identify the funnel boundary; a funnel boundary feature vector is set through the funnel boundary; and the expansion and change correction of the zonal control boundary is performed based on the funnel boundary feature vector.

[0078] Furthermore, the system is also used to implement the following functions:

[0079] Extract the core index parameters of the funnel boundary, including the boundary closure area, the central water level depth, and the boundary expansion rate; and set the feature vector of the funnel boundary using the core index parameters of the funnel boundary.

[0080] Furthermore, the system is also used to implement the following functions:

[0081] A digital twin is established based on the geological structure, hydrological parameters, and historical mining data of the target area. Based on the digital twin, candidate control strategies are input, and the simulation results are compared with the monitoring data. The optimized control strategy is obtained through reinforcement learning.

[0082] Furthermore, the system is also used to implement the following functions:

[0083] The action space associated with the candidate regulation strategy is defined as a mixed control space consisting of zoned extraction quotas, zoned water diversion priorities, and water allocation ratios between adjacent zones. Driven by a reward function, with the goal of minimizing over-extraction risk and maximizing water resource utilization efficiency as the multi-objective optimization orientation, the optimal solution of the action space is explored iteratively through reinforcement learning to obtain the optimized regulation strategy.

[0084] Furthermore, the system is also used to implement the following functions:

[0085] By matching ground subsidence anomaly intervals using cosine similarity, risk-sensitive features under the attention mechanism are set; based on the surface water infiltration and meteorological evaporation, and combined with the risk-sensitive features, their contribution to water balance is quantified, and the over-extraction risk evolution map is set.

[0086] Furthermore, the system is also used to implement the following functions:

[0087] The over-extraction risk index is updated in a rolling manner within each time step using a time sliding window; the over-extraction risk index obtained from the rolling update is mapped to the GIS system to obtain the over-extraction risk evolution map.

[0088] 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.

[0089] 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.

[0090] 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 dynamic assessment method for groundwater over-extraction areas based on water volume tracking, characterized in that, The method includes: Within the target area, based on groundwater extraction, surface water infiltration, and meteorological evaporation, data on the distribution of over-extraction areas, including ground subsidence rate and aquifer storage variables, are obtained under the same timestamp. Based on the aforementioned groundwater extraction volume and water demand database, a groundwater cycle model is constructed, and dynamic simulation and prediction are performed in conjunction with the aforementioned over-extraction area distribution data to determine the zoning control boundaries and optimize the control strategy. Collect water demand data from various industries within the target area, establish a water demand database covering water use type, water use time period, and water use volume, use the obtained groundwater extraction volume as input parameter, and combine the water use distribution characteristics in the water demand database to construct a groundwater cycle model that can simulate the entire process of groundwater recharge, runoff, and discharge; import the obtained over-extraction area distribution data including land subsidence rate and aquifer storage variables into the groundwater cycle model, set simulation scenarios at different time scales, and dynamically simulate and predict the future groundwater level changes and over-extraction expansion trend in the target area; based on the simulation prediction results and combined with the regional water resource carrying capacity, divide the area into slightly over-extraction areas, moderately over-extraction areas, and severely over-extraction areas, determine the zoning control boundaries of each zone, and formulate control strategies that include extraction quotas and water use efficiency improvement requirements; Based on the surface water infiltration and meteorological evaporation, a water balance analysis was performed to generate an over-extraction risk evolution map within a time sliding window. Using a cosine similarity algorithm, the current land subsidence data is matched with historical land subsidence data from the same period and historical high-risk periods to identify abnormal land subsidence intervals. For these abnormal intervals, risk-sensitive features under an attention mechanism are set to quantify the contribution of surface water infiltration and meteorological evaporation to the water balance. A fixed-duration time sliding window is set. Within each time step, the over-extraction risk index calculated from the comprehensive water deficit rate, land subsidence rate, and aquifer reservoir decay amplitude is continuously updated by combining the latest water balance analysis results and risk-sensitive feature data. The over-extraction risk index updated at each time step is mapped to the GIS system according to spatial location, presenting the over-extraction risk level at different locations and time nodes in the target area, and generating an over-extraction risk evolution map within the time sliding window. Based on the zoning control boundaries and optimized regulation strategies, a water flow scheduling network is set up and mapped to the over-extraction risk evolution map to carry out water resource allocation and regulation.

2. The dynamic assessment method for groundwater over-extraction areas based on water volume tracking as described in claim 1, characterized in that, The method includes: A multi-source water tracking network is set up, which integrates multiple underground collection wells, surface water monitoring nodes and meteorological observation nodes; Based on the multi-source water tracking network, the groundwater extraction volume, surface water infiltration volume, and meteorological evaporation volume are determined.

3. The dynamic assessment method for groundwater over-extraction areas based on water volume tracking as described in claim 1, characterized in that, The method for determining zone control boundaries and optimizing control strategies further includes: Deploy remote sensing image nodes and InSAR nodes to monitor the surface deformation trend and changes in surface water area of ​​the target area, and dynamically correct the zoning control boundary.

4. The dynamic assessment method for groundwater over-extraction areas based on water volume tracking as described in claim 3, characterized in that, The method for dynamically correcting the partition control boundary includes: Based on the remote sensing image nodes and InSAR nodes, the surface deformation funnel and water level contour lines are coupled and verified to identify the funnel boundary. A feature vector for the funnel boundary is set through the funnel boundary; Based on the funnel boundary feature vector, the zoning control boundary is expanded and modified.

5. The dynamic assessment method for groundwater over-extraction areas based on water volume tracking as described in claim 4, characterized in that, The method for setting a funnel boundary feature vector through the funnel boundary includes: Extract the core index parameters of the funnel boundary, including the boundary closure area, the central water level depth, and the boundary expansion rate; The feature vector of the funnel boundary is set using the core index parameters of the funnel boundary.

6. The dynamic assessment method for groundwater over-extraction areas based on water volume tracking as described in claim 1, characterized in that, The method for determining zone control boundaries and optimizing control strategies further includes: A digital twin is established based on the geological structure, hydrological parameters, and historical mining data of the target area; Based on the digital twin, candidate control strategies are input, and the simulation results are compared with the monitoring data. The optimized control strategy is obtained through reinforcement learning.

7. The dynamic assessment method for groundwater over-extraction areas based on water volume tracking as described in claim 6, characterized in that, The optimized control strategy is obtained through reinforcement learning, and the method further includes: The action space associated with the candidate control strategy is defined as a mixed control space consisting of zoned mining quota, zoned water diversion priority, and water allocation ratio between adjacent zones. Driven by a reward function, and guided by multiple objectives of minimizing over-extraction risk and maximizing water resource utilization efficiency, the optimal solution of the action space is explored iteratively through reinforcement learning to obtain the optimized control strategy.

8. A dynamic assessment system for groundwater over-extraction areas based on water volume tracking, characterized in that, The system is used to implement the dynamic assessment method for groundwater over-extraction areas based on water volume tracking as described in any one of claims 1-7, and the system comprises: The over-extraction area distribution data acquisition module is used to acquire over-extraction area distribution data, including ground subsidence rate and aquifer storage variables, under the same timestamp within the target area, based on groundwater extraction, surface water infiltration, and meteorological evaporation. The dynamic simulation and prediction module is used to construct a groundwater cycle model based on the groundwater extraction volume and water demand database, and to perform dynamic simulation and prediction in conjunction with the over-extraction area distribution data to determine the zoning control boundaries and optimize the control strategy. The water balance analysis module is used to perform water balance analysis based on the surface water infiltration and meteorological evaporation, and generate an over-extraction risk evolution map within a time sliding window. The water resources allocation and regulation module is used to set up a water flow scheduling network that is associated with and mapped to the over-extraction risk evolution map based on the zoning control boundary and the optimized regulation strategy, and to carry out water resources allocation and regulation.

Images