Ecological threshold extraction and monitoring early warning method and system based on multi-source data

Abstract

The invention discloses an ecological threshold extraction and monitoring early warning method and system based on multi-source data, and relates to the technical field of environment monitoring. The ecological threshold extraction and monitoring early warning method and system based on the multi-source data comprises the following steps: S1, collecting a multi-source data set and performing normalization processing, and identifying an ecological health steady state to form a historical health sample; s2, constructing a three-dimensional aquifer by using a historical healthy sample, and simulating a flow field to divide risk levels; s3, obtaining an ecological sensitive area threshold value based on the multi-source data set and the risk level, and carrying out early warning response; s4, performing early warning treatment according to the divided water resource safety area, establishing a three-dimensional hydrogeological model, and dynamically correcting a water level threshold value; and S5, performing comprehensive evaluation based on the corrected dynamic threshold, executing measures and generating a report closed loop. According to the invention, the accuracy of ecological threshold extraction, monitoring early warning and water resource management is effectively improved, and the problems of false steady state misjudgment and early warning misjudgment and missed judgment of underground water monitoring in the prior art are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of environmental monitoring, in particular to an ecological threshold extraction and monitoring and early warning method and system based on multi-source data. BACKGROUND

[0002] With the increasing demand for ecological protection and the increasing complexity of environmental changes, traditional ecological monitoring systems mainly rely on single data sources and static thresholds for monitoring and early warning. There are obvious limitations in spatial coverage, data fusion, real-time response and comprehensive evaluation, which cannot fully reflect the dynamic changes of complex ecological environment. Especially in water resource management, ecological degradation prediction and ecological restoration evaluation, the existing technology has not effectively integrated multi-source data, resulting in that ecological monitoring cannot timely respond to dynamic adjustment of environmental changes and hydrological conditions. There are deficiencies in dynamic threshold adjustment and prediction accuracy, especially in high-frequency changing ecological systems, the response mechanism is particularly weak.

[0003] For example, the invention patent with publication number CN117109665A discloses a river ecological environment data online monitoring method and system, which relates to the technical field of river environment monitoring. The water body environment module monitors various water quality parameters in the water body in real time, the water quality data module monitors water quality indicators in real time, the biodiversity module monitors the information of fish, benthic animals and aquatic plant biodiversity in real time, the river function module obtains the dynamic situation of the river and the flood regulation function of the river, and the data processing module integrates the preprocessed environmental parameters, water quality index data, biological data and function data into a central database. The data analysis module obtains river data based on the central database, which includes environmental parameters, water quality index data, biological data and function data. The monitoring system comprehensively analyzes multi-source data to evaluate the health status and sustainability of the river ecosystem, and the analysis is more comprehensive, providing support and decision-making for river management.

[0004] For example, the invention patent with publication number CN119860819A discloses an environmental monitoring method and system based on multi-source data, which includes: dividing the area to be monitored into several environmental monitoring areas, determining the pore dynamic parameters of the environmental monitoring areas to screen the pore enrichment areas, constructing a feature vector set according to the direction similarity of several wind direction characteristic vectors and determining whether there is an interaction influence area, extracting the wind power parameters of each feature vector in the feature vector set, determining the range of the interaction influence area, obtaining the water-soil ecological coupling value of the interaction influence area to determine whether to issue an early warning prompt of environmental abnormality in the interaction influence area, and then, dynamically adjusting the monitoring focus according to the specific conditions of the soil in the area combined with the actual environmental impact factors, improving the capture accuracy of the monitoring area and the reliability of the environmental monitoring.

[0005] In the prior art, the existing system often cannot effectively reflect the three-dimensional hydraulic connection of groundwater and the coupling of ecological water demand under the condition that the aquifer structure is complex and the recharge process changes rapidly. Especially in the oasis edge and downstream dry-up sensitive area, it is easy to appear "water level is acceptable but ecology has degenerated" false steady state misjudgment, thereby affecting the management of groundwater resources and the evaluation of ecological protection.

[0006] Therefore, in view of the above problems, there is an urgent need for an ecological threshold extraction and monitoring and early warning method and system based on multi-source data. SUMMARY

[0007] Technical problems to be solved In view of the deficiencies of the prior art, the present application provides an ecological threshold extraction and monitoring and early warning method and system based on multi-source data, which solves the problems of false steady state misjudgment and early warning misjudgment and missed judgment of groundwater monitoring in the prior art.

[0008] Technical scheme To achieve the above purpose, the present application realizes the following technical scheme: an ecological threshold extraction and monitoring and early warning method and system based on multi-source data, comprising the following steps: S1, collecting a multi-source data set, performing NTP millisecond-level synchronization and normalization processing, and constructing a unified database to identify an ecological health steady state time period to form a historical health sample; S2, using the historical health sample, constructing a three-dimensional aquifer structure by a layered interpolation method, simulating a groundwater flow field and identifying a recharge area to obtain groundwater depth partition results and divide different risk levels; S3, combining the multi-source data set and the risk level, obtaining an ecological sensitive area threshold value through a regression model, and combining an abnormal trigger condition and a well point level reliable value to perform early warning response; S4, combining the multi-source data set to delineate a water resource safety zone, performing early warning disposal, establishing a three-dimensional hydrogeological model, and dynamically correcting the water level threshold; S5, combining the corrected dynamic threshold, performing comprehensive evaluation, generating a periodical report, and writing back the management results to form a closed loop.

[0009] Further, the specific steps of collecting a multi-source data set, performing NTP millisecond-level synchronization and normalization processing, and building a unified database are as follows: relying on the natural resource comprehensive observation system in the lower reaches of Heihe River to cooperatively collect a multi-source data set, the multi-source data set including: a well point monitoring data set, a geophysical exploration data set, a remote sensing ecological index data set, a surface water and meteorological data set, and an irrigation and water taking management data set; unifying the multi-source data set through millisecond-level time synchronization, data calibration, continuity guarantee, reliability label and spatial consistency, normalizing the time frequency, amplitude and spatial scale according to the sampling frequency of every ten days or every month, and constructing a unified database.

[0010] Further, the specific steps of identifying the ecological health steady state time period to form the historical health sample are as follows: through rolling analysis of a fixed time window, automatically identify the time period when the region is in ecological health steady state; save the data in all multi-source data sets in the time period in ecological health steady state as historical health samples; obtain the missing rate by comparing the number of effective collection records of each monitoring well in the well point monitoring data set in a uniform sampling period with the number of records that should be collected in the period; obtain the drift rate by comparing the water level observation value in the well point monitoring data set with the water level in the ecological health steady state time period and performing smoothness comparative analysis; obtain the abnormal rate by comparing the water level in the well point monitoring data set with the remote sensing ecological index data set in a uniform sampling period and comparing the proportion of data that is abnormal or inconsistent with adjacent time windows; the complement of the missing rate multiplied by the missing rate weight coefficient is the integrity credible contribution item, the complement of the drift rate multiplied by the drift rate weight coefficient is the stability credible contribution item, and the complement of the abnormal rate multiplied by the abnormal rate weight coefficient is the consistency credible contribution item; the integrity credible contribution item, the stability credible contribution item, and the consistency credible contribution item are summed to obtain the well point level credible value; compare the well point level credible value with the allowed minimum credibility threshold in real time, and when the well point level credible value is higher than the allowed minimum credibility threshold, it is determined that the monitoring data quality meets the conditions for abnormal identification and trigger judgment, and enters the abnormal trigger; if the well point level credible value is lower than the allowed minimum credibility threshold, the abnormal trigger is not immediately executed, the well point monitoring data is marked as to-be-checked data, and the supplementary measurement or consistency comparison with adjacent well point data and remote sensing data is triggered to avoid misjudgment or omission due to insufficient data quality.

[0011] Further, using the historical health sample, the specific steps of constructing a three-dimensional aquifer structure, simulating a groundwater flow field, and identifying a recharge area by a hierarchical interpolation method are as follows: obtain the well depth, aquifer horizon, and filter pipe position of the well point monitoring data set, and combine the groundwater depth contour surface, hydraulic slope direction, and low resistance anomaly area distribution in the geophysical exploration data set to obtain the three-dimensional structure of the aquifer depth by stratified interpretation and spatial stratified Kriging interpolation method; output the regional aquifer thickness distribution, depth fluctuation, and permeability difference zone identification results; take the constructed three-dimensional aquifer depth structure as a spatial boundary framework, set boundary conditions according to hydrogeological unit division and groundwater flow field distribution characteristics: take the hydraulic connection of the river section as a constant water head boundary; take the regional first-order divide line and the outer edge of the desertification area as a no-flow boundary; fuse the surface water and meteorological data set and the irrigation and water taking management data set to establish a groundwater flow direction simulation through the water level depth gradient field, fuse the farmland irrigation rotation irrigation record, groundwater water taking data, and meteorological driving to perform sensitivity attribution analysis and construct a multi-source driving contribution rate decomposition model, and through multivariate regression, principal component analysis, and impulse response decomposition statistical method, decompose the groundwater water quantity change into contribution items of different driving mechanisms.

[0012] Furthermore, the specific steps for obtaining groundwater depth zoning results and classifying different risk levels are as follows: Groundwater depth data is obtained through underground resistivity, ground vibration, gravity anomalies, and geological exploration. Spatial structure consistency verification and anomaly identification are performed to predict groundwater depth and obtain iso-depth values ​​for different depth sections. Combined with actual water level and well depth data from wellpoint monitoring datasets and compared with iso-depth surface layers obtained from geophysical exploration, groundwater depth zoning is formed. Spatial interpolation technology is used to perform three-dimensional reconstruction of the depth values, resulting in groundwater depth zoning results. Using these groundwater depth zoning results, ecologically sensitive areas are divided into different risk levels.

[0013] Furthermore, combining multi-source datasets and risk levels, the thresholds for ecologically sensitive areas are derived through regression models. The specific steps for early warning response, combining anomaly triggering conditions and well-point level confidence values, are as follows: Key ecological water level thresholds are extracted from zoning results such as well-point monitoring data, remote sensing ecological indicators, and groundwater depth; for different types of vegetation, ecologically sensitive area thresholds are derived through regression models, generating ecological water level thresholds and boundary thresholds; when the data in the collected multi-source datasets meet the anomaly triggering conditions and the well-point level confidence value is higher than the minimum confidence threshold, the system automatically triggers anomaly identification and reporting actions; anomaly warning information is pushed, entering the early warning response stage; anomaly event labels and triggering reasons are recorded, and the digital twin computing module and threshold verification mechanism are activated.

[0014] Furthermore, by combining multi-source datasets to delineate water resource safety zones, the specific steps for implementing early warning and response are as follows: Different levels of water resource safety zones are delineated based on groundwater level changes and groundwater depth data; these levels include: safe zones, warning zones, and risk zones; when entering the early warning response phase, different early warning and response measures are implemented based on different triggering conditions; early warning responses include: yellow zone response, orange zone response, and red zone response. After completing the corresponding early warning and response measures, detailed labeling records are required, and a traceability mechanism is established.

[0015] Furthermore, the specific steps for establishing a three-dimensional hydrogeological model and dynamically correcting the water level threshold are as follows: A comprehensive simulation of groundwater flow, infiltration, recharge, and discharge processes is performed using a multidimensional dataset. A three-dimensional hydrogeological model is established through spatial interpolation and dynamic simulation. Based on real-time monitoring data and the simulation results of the three-dimensional hydrogeological model, the water level threshold is reviewed and corrected: the benchmark water level threshold is used as the benchmark threshold term; the product of the water quantity deficit anomaly field coefficient and the water quantity deficit anomaly field is used as the water quantity deviation contribution term; and the product of the ecological response gap field coefficient and the ecological response gap field is used as the ecological response gap field contribution term. The ecological gap correction term, the product of the structural sensitivity field coefficient and the structural sensitivity field are used as the structural vulnerability modulation term, the product of the model-observation bias coefficient and the model-observation bias is used as the model correction compensation term, and the product of the certainty penalty coefficient and the certainty penalty term is used as the uncertainty suppression term. The six terms—the baseline threshold term, the water volume deviation contribution term, the ecological gap correction term, the structural vulnerability modulation term, the model correction compensation term, and the uncertainty suppression term—are summed to obtain the corrected dynamic threshold. Based on the corrected dynamic threshold, water resource management is optimized in real time, and the water level threshold is updated periodically.

[0016] Furthermore, based on the revised dynamic threshold, a comprehensive assessment is conducted to generate periodic reports, and the specific steps for writing back the governance results to form a closed loop are as follows: A comprehensive assessment mechanism is constructed based on the revised dynamic threshold to periodically assess the effectiveness of water resource use and ecological restoration; the threshold trigger delay value is multiplied by the response weight coefficient as the timeliness term, the comprehensive judgment value is multiplied by the effect weight coefficient as the effectiveness term, the hydrological sequence value is multiplied by the stability weight coefficient as the steady-state term, and the correction ratio is multiplied by the learning weight coefficient as the enhancement term. The sum of the four terms—timeliness, effectiveness, steady-state, and enhancement—is obtained to obtain the comprehensive assessment value; the comprehensive assessment value is compared with the decision threshold in real time. When the comprehensive assessment value is greater than the decision threshold, emergency response measures are taken to control water resource use in a timely manner; when the comprehensive assessment value is less than or equal to the decision threshold, it indicates that the system is in a relatively stable state, and routine management measures are taken; the status of the integrated monitoring well network, ecological health indicators, and safety zones are visualized, and weekly, monthly, and quarterly reports for individual wells and zones are automatically generated. The implementation results of the limited extraction and water conveyance governance measures are automatically written back, forming a complete feedback loop.

[0017] Furthermore, the second aspect of this invention provides an ecological threshold extraction and monitoring early warning system based on multi-source data, applied to a method for ecological threshold extraction and monitoring early warning based on multi-source data, comprising: a groundwater multi-source data acquisition and management module, used to acquire multi-source datasets, perform NTP millisecond-level synchronization and normalization processing, construct a unified database, and identify ecological health steady-state time periods to form historical health samples; a three-dimensional hydrogeological digital twin module, used to utilize historical health samples to construct a three-dimensional aquifer structure through hierarchical interpolation, simulate the groundwater flow field and identify recharge areas, obtain groundwater burial depth zoning results, and classify different risk levels; an ecological effective water level threshold extraction module, used to combine multi-source datasets and risk levels, derive ecologically sensitive area thresholds through a regression model, and conduct early warning responses based on abnormal triggering conditions and well-point level confidence values; a dynamic water resource safety zone hierarchical early warning module, used to delineate water resource safety zones based on multi-source datasets, execute early warning measures, establish a three-dimensional hydrogeological model, and dynamically correct water level thresholds; and a closed-loop management and evaluation module, used to conduct comprehensive evaluation based on the corrected dynamic thresholds, generate periodic reports, and write back the management results to form a closed loop.

[0018] Beneficial effects The present invention has the following beneficial effects: (1) This invention, by introducing three-dimensional hydrogeological digital twin technology and integrating multi-source datasets, can accurately reconstruct the three-dimensional distribution and flow patterns of groundwater. This multi-source data-based fusion assessment mechanism greatly improves the accuracy of reflecting the dynamic relationship between groundwater resources and the ecological environment, and avoids the limitations of traditional well-point monitoring.

[0019] (2) This invention, through a dynamic water resource safety zone delineation method combined with a modified dynamic threshold, enables the implementation of real-time water resource management optimization strategies in different water resource risk zones. By using real-time updated water level thresholds and ecological health assessments, it ensures that water resources prioritize ecological restoration and agricultural irrigation needs, effectively supporting the sustainable development of ecosystems.

[0020] (3) This invention combines real-time monitoring data with a corrected dynamic water level threshold to form a comprehensive assessment value. It can dynamically assess the impact of various environmental changes on groundwater resources, respond promptly to ecological changes, and provide more targeted and efficient decision support. By comparing the assessment value with historical health samples in real time, corresponding water resource management measures can be automatically triggered, thereby effectively reducing waste in water resource utilization.

[0021] (4) This invention can automatically detect abnormal changes in the groundwater system and issue early warnings by continuously monitoring and analyzing real-time data. For example, when the water level reaches a sensitive threshold, the system will automatically activate an early warning and take measures. At the same time, it can evaluate the response effect of each early warning event in real time through automated report generation and closed-loop measures, providing feedback for future water resource management optimization, reducing manual intervention, and improving the accuracy of decision-making and execution efficiency.

[0022] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0023] Figure 1 This is a flowchart of the ecological threshold extraction and monitoring early warning method based on multi-source data of the present invention; Figure 2 This is a structural diagram of the ecological threshold extraction and monitoring early warning system based on multi-source data of the present invention; Figure 3 This is a distribution map of groundwater level monitoring points in 2024, as per the present invention. Figure 4 This is a distribution map of groundwater level rise and fall in 2024, as presented in this invention. Figure 5 This is a map showing the shallow groundwater isohyetal lines and depths in 2024, as presented in this invention. Figure 6 This is a spatial distribution map of ecological and environmental problems in the key area of ​​the lower reaches of the Heihe River, as presented in this invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Please see Figures 1-6This invention provides a technical solution: an ecological threshold extraction and monitoring early warning method and system based on multi-source data, comprising: S1, collecting multi-source datasets, performing NTP millisecond-level synchronization and normalization processing, constructing a unified database, and identifying historical health samples by identifying ecological health steady-state time periods; S2, using historical health samples, constructing a three-dimensional aquifer structure through hierarchical interpolation, simulating the groundwater flow field and identifying recharge areas, obtaining groundwater depth zoning results, and classifying different risk levels; S3, combining multi-source datasets and risk levels, deriving ecologically sensitive area thresholds through regression models, and conducting early warning responses by combining abnormal triggering conditions and well-point-level confidence values; S4, delineating water resource safety zones based on multi-source datasets, implementing early warning measures, establishing a three-dimensional hydrogeological model, and dynamically correcting water level thresholds; S5, combining the corrected dynamic thresholds, conducting comprehensive evaluations, generating periodic reports, and writing back the governance results to form a closed loop.

[0026] Specifically, the steps for collecting multi-source datasets and performing NTP millisecond-level synchronization and normalization processing to build a unified database are as follows: Relying on the comprehensive natural resource observation system of the lower reaches of the Heihe River, multi-source datasets are collected collaboratively through well network monitoring, geophysical exploration, remote sensing inversion, hydrological and meteorological observation, and water intake and management records. A dynamic sensing network of all elements coupled with groundwater and ecology is constructed. All data sources are existing project results. The multi-source datasets include: well point monitoring dataset, geophysical exploration dataset, remote sensing ecological index dataset, surface water and meteorological dataset, and irrigation and water intake management dataset.

[0027] The wellpoint monitoring dataset includes hydraulic parameters, water quality parameters, dynamic stability, and wellpoint attributes, reflecting the dynamic changes of shallow groundwater and forming the core basis for threshold extraction. Hydraulic parameters include water level depth, hydraulic head elevation, and water level change rate, providing direct indicators for determining recharge and over-extraction pressure. Water quality parameters include conductivity, conductivity change rate, and pH value, distinguishing between river recharge and evaporation concentration. Dynamic stability includes the periodic fluctuation amplitude per ten days and per season, used to establish healthy baseline samples. Wellpoint attributes include well depth, aquifer position, and filter location, serving as the basis for three-dimensional hydrological structure reconstruction. The geophysical exploration dataset includes isobathation values, hydraulic gradient direction, low resistivity anomaly zones, and uncertainty distribution, used for isomorphic groundwater depth. Inversion and flow field inference complete sparse well point areas; isobathion depth values ​​include: first-level depth, second-level depth, and third-level depth, used to determine root accessibility; remote sensing ecological index datasets include: NDVI, NPP, seasonal phase transition time, long-term trends, and abrupt change detection, reflecting groundwater support for vegetation and ecological degradation risk; surface water and meteorological datasets include: precipitation, evapotranspiration, temperature and humidity, wind speed, river runoff to multi-year water volume ratio, precipitation and river recharge, evapotranspiration demand, and wind speed and extreme drought, used to determine recharge conditions, seasonal drivers, and evapotranspiration pressure; irrigation and water intake management datasets include: irrigation district layout, planting structure, rotational irrigation quantity and time, water intake, and equipment type, used to determine human activity boundary conditions and achieve scenario prediction.

[0028] The collected multi-source datasets are unified through millisecond-level time synchronization, data calibration, continuity assurance, credibility labeling, and spatial consistency. Based on the sampling frequency of ten days or monthly, time frequency, amplitude, and spatial scale are normalized to build a unified database. NTP is used for millisecond-level unification to ensure complete alignment of different data sources in the time dimension. Zero-point calibration and automatic atmospheric pressure correction are performed regularly on water level and conductivity sensors to ensure consistent measurement dimensions and reliable accuracy. The system is compatible with mainstream protocols for well equipment, GPRS cellular IoT communication, and OGC standardized data services, enabling unified access and secure transmission across vendors. All data in the collected multi-source datasets are organized in the JSON standard format. Edge caching nodes provide breakpoint resume capability, ensuring no data loss during communication interruptions and continuous retransmission after recovery. Simultaneously, each record undergoes missing measurement, drift, and mutation quality detection, automatically generating credibility labels and completing spatial consistency registration and multi-source cross-verification, providing reliable input for groundwater trend analysis, mutation identification, and extraction of ecologically effective water level thresholds.

[0029] This implementation plan achieves comprehensive monitoring of the groundwater system and ecological environment. Millisecond-level time synchronization ensures complete alignment of time dimensions across different data sources, eliminating time discrepancies and guaranteeing data timeliness and reliability. It also ensures the accuracy of collected data, eliminating spatial heterogeneity and dimensional differences, providing reliable input for groundwater trend analysis, abrupt change identification, and extraction of ecologically effective water level thresholds. All processed multi-source datasets are organized in a unified JSON format, ensuring continuous data transmission and integrity. This enhances the accuracy and real-time performance of the groundwater ecological monitoring and early warning system, providing a solid data foundation for ecological risk assessment and water resource management.

[0030] Specifically, the steps for identifying ecologically healthy steady-state time periods and forming historical health samples are as follows: Through rolling analysis within a fixed time window, the time periods in which the region is in an ecologically healthy steady-state are automatically identified. Ecologically healthy steady-state conditions include the following: groundwater levels remain stable and change regularly within the time period; data in the remote sensing ecological index dataset do not show a decrease or deviation exceeding 30%; no human disturbance events such as groundwater extraction or concentrated water withdrawal occur within the time period; the inferred groundwater depth matches the measured water level, and the inferred groundwater depth is obtained from the isodepth values ​​in the geophysical exploration dataset through layered media interpretation and isosurface interpolation methods, forming continuously spatially distributed inferred groundwater depth data; data from all multi-source datasets within the ecologically healthy steady-state time periods are saved as historical health samples, serving as a stable benchmark for ecological threshold extraction and groundwater anomaly diagnosis.

[0031] After identifying and extracting the ecological health steady-state time period, the quality of multi-source data collected within this time period is assessed to ensure the reliability and consistency of historical health samples used for threshold extraction. The missing rate is obtained by comparing the number of valid data collection records for each monitoring well in the wellpoint monitoring dataset within a unified sampling period with the number of records that should be collected in that period. The drift rate is obtained by comparing the water level observations in the wellpoint monitoring dataset with the water levels within the ecological health steady-state time period and performing a stability comparison analysis. The anomaly rate is obtained by comparing the proportion of data in the wellpoint monitoring dataset and the remote sensing ecological indicator dataset that are abnormal or inconsistent with adjacent time windows within a unified sampling period. A wellpoint-level data reliability evaluation model is established based on the missing detection rate, drift rate, and anomaly rate. This model outputs a wellpoint-level reliability value, which quantitatively evaluates the data integrity, observation stability, and anomaly interference of wellpoint monitoring records. This determines the contribution and reliability of different wellpoint data to the analysis results. The complementary value of the missing detection rate multiplied by the missing detection rate weighting coefficient is used as the integrity reliability contribution item; the complementary value of the drift rate multiplied by the drift rate weighting coefficient is used as the stability reliability contribution item; and the complementary value of the anomaly rate multiplied by the anomaly rate weighting coefficient is used as the consistency reliability contribution item. These three items are then summed to obtain the wellpoint-level reliability value. The specific calculation formula for the wellpoint-level reliability value is as follows: ; In the formula, This indicates the missing rate, used to quantify the completeness and continuity of well point monitoring data; It represents the drift rate, used to reflect the long-term stability of the observation equipment in relation to environmental conditions; The anomaly rate reflects the reliability and consistency of the data; The missing test rate weighting coefficient is calculated from the integrity sensitivity analysis of historical healthy samples, with a value range of 0.2–0.5, and is used to strengthen the impact of data integrity on the credibility calculation. The drift rate weighting coefficient is calculated by analyzing the deviation between the monitored water level and the water level during the ecological health steady state period. The value ranges from 0.2 to 0.5 and is used to highlight the contribution of long-term reliability factors to credibility. The anomaly rate weighting coefficient is calculated by the consistency verification of the well point monitoring dataset and the remote sensing ecological index dataset. Its value ranges from 0.2 to 0.5 and is used to characterize the degree of influence of anomaly interference on credibility.

[0032] The system compares the wellpoint-level confidence value with the minimum allowable confidence threshold in real time. When the wellpoint-level confidence value is higher than the minimum allowable confidence threshold, the monitoring data quality is deemed to meet the conditions for anomaly identification and trigger judgment, and anomaly triggering is initiated. If the wellpoint-level confidence value is lower than the minimum allowable confidence threshold, anomaly triggering is not executed immediately. Instead, the wellpoint monitoring data is marked as data to be verified, and supplementary measurement or consistency comparison with adjacent wellpoint data and remote sensing data is triggered to avoid misjudgment or omission due to insufficient data quality.

[0033] This implementation scheme involves real-time synchronous acquisition of multi-source datasets, including well point monitoring, geophysical exploration, remote sensing ecological indicators, surface water, and meteorological data. Millisecond-level time synchronization and normalization are performed to eliminate data heterogeneity and dimensional differences. Quality assessments are conducted using missing rate, drift rate, and anomaly rate, with quality control based on a well point-level reliability calculation model to ensure compliance with anomaly identification and triggering criteria. Through cross-checking and spatial consistency processing of multi-source data, abnormal fluctuations in groundwater and the ecological environment are detected in real time, generating anomaly event identifiers and triggering early warning measures. Finally, the processed data is integrated into stable historical health samples, providing a reliable benchmark for ecological threshold extraction and anomaly diagnosis, supporting dynamic water resource management and ecological protection.

[0034] Specifically, using historical health samples, a three-dimensional aquifer structure is constructed through hierarchical interpolation to simulate the groundwater flow field and identify recharge zones. The specific steps are as follows: Well depth, aquifer level, and filter location are obtained from the wellpoint monitoring dataset. Combined with groundwater depth isosurfaces, hydraulic gradient directions, and low-resistivity anomaly distribution from the geophysical exploration dataset, the three-dimensional aquifer depth structure is obtained through layered interpretation and spatial hierarchical Kriging interpolation. The regional aquifer thickness distribution, depth undulations, and permeability difference zone identification results are output, providing basic constraints for flow field solutions. The well depth, aquifer horizon, and filter pipe location from the well point monitoring dataset form a hard constraint point set. The groundwater depth isosurface from the geophysical exploration dataset is used as the initial value surface, the hydraulic gradient direction as the interpolation trend field, and the low resistivity anomaly zone as a soft constraint for high-value permeability corridors. The hierarchical Kriging interpolation algorithm is used to construct the surface of the aquifer top and bottom interface. Anisotropic spatial constraints along the gradient direction enhance connectivity. Ellipsoidal search neighborhood controls the orientation of fault depressions or paleochannel cutting zones. At the same time, well point-level confidence values ​​are superimposed to adjust the interpolation weights, so that high-confidence well points contribute more to the surface morphology.

[0035] Using the constructed three-dimensional aquifer burial depth structure as the spatial boundary framework, boundary conditions are set according to the hydrogeological unit division and groundwater flow field distribution characteristics: the hydraulic connection of the river section is used as the constant head boundary; the first-level watershed line of the region and the area without hydraulic recharge at the outer edge of the desertification zone are used as the flow-free boundary; the irrigation area and the main precipitation recharge zone are used as the surface source recharge boundary; the surface water and meteorological datasets are integrated with the irrigation and water intake management datasets, and groundwater flow direction simulation is established through the water level burial depth gradient field. This includes reconstructing the surface-scale groundwater flow path, identifying local recharge and discharge areas, analyzing the coupling relationship between groundwater recharge and pumping irrigation, and verifying the consistency between the twin outputs and well point measured change trends; well point-level confidence values ​​are used as dynamic weights to continuously update the three-dimensional groundwater level field, with special annotation of water level rise. The study identifies three groundwater distribution zones: point distribution (ecological restoration areas), drop point distribution (potential risk areas), and zones with rapid changes in burial depth (ecologically sensitive zones). It aims to achieve a dynamic twin representation of groundwater process changes. By integrating farmland irrigation rotation records, groundwater extraction data, and meteorological drivers, a sensitivity attribution analysis is conducted to construct a multi-source driving contribution rate decomposition model, quantifying the causal contribution of groundwater dynamic changes. Through multiple regression, principal component analysis, and impulse response decomposition statistical methods, groundwater volume changes are decomposed into contribution items from different driving mechanisms. These contribution items include differentiated processes such as meteorological depletion-type descent, human activity-induced extraction-type descent, river recharge, lateral infiltration, and recharge in oasis areas versus weak recharge in desert areas, providing objective causal support for the estimation of groundwater safety thresholds and zonal management.

[0036] In this implementation plan, a three-dimensional aquifer structure is constructed using hierarchical interpolation and historical health samples. Groundwater flow fields are simulated and recharge zones are identified by combining well point monitoring and geophysical data. The Kriging interpolation method is employed, combining well point data, depth isosurfaces, hydraulic gradient, and low-resistivity anomaly information to construct a three-dimensional aquifer depth structure, outputting regional aquifer thickness and permeability differences. Boundary conditions are set according to hydrogeological units and groundwater flow field characteristics to simulate groundwater flow paths. Recharge and discharge zones are identified by combining surface water, meteorological data, and irrigation management. The three-dimensional groundwater level field is continuously updated by adjusting weights based on well point confidence, and ecological restoration zones, potential risk zones, and ecologically sensitive zones are marked. Combining farmland irrigation records, water withdrawals, and meteorological data, a multi-source driven contribution rate decomposition model is used to quantitatively analyze the impact of meteorological, human activities, and river recharge factors on groundwater changes, providing a basis for groundwater safety threshold estimation and zonal management.

[0037] Specifically, the steps for obtaining groundwater depth zoning results and classifying different risk levels are as follows: Ground resistivity, ground vibration, gravity anomalies, and geological exploration are used. Ground resistivity indirectly identifies aquifer interfaces through low-resistivity layer characteristics, with moderate reliability. Ground vibration uses abrupt changes in wave velocity to determine the dry-saturated interface, with moderate reliability. Gravity anomalies only macroscopically indicate the aquifer range through density differences, with low resolution and the lowest reliability. Geological exploration directly measures aquifer depth through drilling, with the highest reliability. Based on reliability differences, fusion weights are assigned to obtain groundwater depth data. Spatial structure consistency verification and anomaly identification are performed to infer groundwater depth and obtain iso-depth values ​​for different depth sections. Combining actual water level and well depth data from wellpoint monitoring datasets with iso-depth surface layers from geophysical exploration, groundwater depth zoning is formed. This can display changes in groundwater depth and water level distribution in different areas, reflecting the correlation between groundwater accessibility and the ecological environment. Figure 3 The map showing the distribution of groundwater level monitoring points in 2024 illustrates how these points are located throughout the work area. Each point represents a monitoring well, and well data is a crucial source for obtaining groundwater level information. Figure 4 The map showing the distribution of groundwater level rise and fall in 2024 uses colors to distinguish areas of rising and falling water levels and indicates the magnitude of the changes. Green areas represent rising water levels, while yellow and orange areas represent falling water levels, allowing direct observation of groundwater changes in different regions. Combining well location data and geophysical survey data, spatial interpolation techniques are used to perform three-dimensional reconstruction of the burial depth values, resulting in groundwater depth zoning results for the delineation of ecologically sensitive zones; for example... Figure 5 The 2024 shallow groundwater isohyets and depth map shown reflects the spatial distribution characteristics of groundwater at different depths. The map uses isohyets to represent changes in groundwater levels; denser lines indicate significant changes. Different colors represent the accessibility of groundwater in deeper areas. Shallow groundwater is easily affected by the external environment, while deep groundwater is more difficult for surface ecosystems to utilize directly.

[0038] Using the results of groundwater depth zoning, ecologically sensitive areas are divided into different risk levels. These risk levels include: extremely sensitive zones, moderately sensitive zones, hard risk zones, and stable weakly sensitive zones. Extremely sensitive zones are those with a depth less than or equal to the extremely sensitive threshold, where vegetation roots can directly access groundwater. Moderately sensitive zones are those between the moderate and extremely sensitive thresholds, where vegetation recharge capacity gradually weakens. Hidden risk zones are those between the primary and secondary sensitive thresholds, where groundwater has not triggered an alarm, but the ecological response has weakened. Stable weakly sensitive zones are those above the primary sensitive threshold, where groundwater provides very little support to surface vegetation. This system is used to achieve early risk recognition.

[0039] This implementation plan, by obtaining groundwater depth zoning results, visually demonstrates the changes in groundwater depth and water level distribution in different areas, and reveals the close relationship between groundwater accessibility and the ecological environment. Figure 3 The distribution of water level monitoring points is displayed. Figure 4 Water level rise and fall distribution map and Figure 5 The shallow groundwater isohyets and depth maps clearly identify the dynamic changes in groundwater resources, especially water level fluctuations at different depths, providing a scientific basis for the delineation of ecologically sensitive zones. Ultimately, through detailed analysis of groundwater depth zones, early warning and identification of ecological risks are achieved, providing effective support for ecological environmental protection and water resource management.

[0040] Specifically, combining multi-source datasets and risk levels, the thresholds for ecologically sensitive areas are derived through regression models. The specific steps for early warning response, combining abnormal triggering conditions and wellpoint-level confidence values, are as follows: Key ecological water level thresholds are extracted from wellpoint monitoring data, remote sensing ecological indicators, and groundwater depth zoning results. These ecological water level thresholds include: critical values ​​for root-accessible water depth, inflection points for ecological degradation thresholds, risk red zones, and oasis edge sensitivity values. Based on the root growth patterns of plants and the water requirements of different vegetation types, the water acquisition capacity of the root system is calculated using an ecological model as the critical value for root-accessible water depth. Correlation analysis between groundwater depth and NDVI is performed to obtain the inflection point for ecological degradation thresholds. By combining wellpoint water level data with remote sensing ecological indicator data, the overlapping areas of potential risk areas and hidden risk zones are marked as risk red zones. Statistical analysis of the rise and fall points of wellpoint water level data and remote sensing ecological indicators shown in Tables 1 and 2 yields the oasis edge sensitivity values.

[0041] As shown in Table 1, the project name is "Groundwater Flow Field Survey in Key Areas," the monitoring area is the Ejina Basin in the lower reaches of the Heihe River Plain, and the unified measurement period is the first phase of the high water level survey. The location is on S214 line in Saihantaolai Sumu, Ejina Banner, Inner Mongolia Autonomous Region, with latitude and longitude of 100°15′31.02″, 41°46′19.07″. The groundwater depth measured in Gansu in 2020 was 4.09 km, and the groundwater depth measured in Xining in 2021 was 4.25 km. The groundwater depth increased slightly in 2020 and 2021, indicating small changes in the regional groundwater level. The location is at kilometer C036-4 in Xiaoyou, Mengketu Gacha, Ejina Banner, Inner Mongolia Autonomous Region, with latitude and longitude of... At the monitoring point located at 100°36′28.25″, 41°50′5.32″, the groundwater depth measured in Gansu Province in 2020 was 1.00 mm, while the groundwater depth measured by the Xining center in 2021 was 1.73 mm. The groundwater level at this monitoring point showed significant fluctuations, with the 2021 depth increasing by 0.73 mm compared to 2020. At the monitoring point located in Kenchagan, Ejin Banner, Inner Mongolia Autonomous Region, at 100°38′53.35″, 41°24′35.41″, the groundwater depth measured in Gansu Province in 2020 was 2.01 mm, while the 2021 depth measured by the Xining center was 2.04 mm. The groundwater depth showed minimal fluctuations, with the 2021 depth increasing by 0.03 mm compared to 2020, indicating a stable regional groundwater level. Comparing groundwater depth values ​​at different times helps identify the changing trends of groundwater in the region and its impact on the ecological environment.

[0042] Table 1. Table of Well Point Water Level Increases

[0043] As shown in Table 2, the project name is "Groundwater Flow Field Survey in Key Areas," the monitoring area is the Ejina Basin in the lower reaches of the Heihe River Plain, and the unified measurement period is the first phase of the high water level survey. Project number E2004 is located in Hongxing Second Team, Dongfeng Town, Ejina Banner, Inner Mongolia Autonomous Region, with latitude and longitude of 100°35′7.74″, 41°23′19.52″. The groundwater level depth measured in Gansu in 2020 was 2.51 meters, and the groundwater level depth measured in Xining in 2021 was 2.34 meters, indicating a rise in groundwater level. Project number EQ045 is located near the Heicheng Ruins in Ejina Banner, Inner Mongolia Autonomous Region, with latitude and longitude of 101°6′36″. At coordinates 0.56″ and 41°46′58.64″, the groundwater level depth measured in Gansu Province in 2020 was 9.15, and the groundwater level depth measured in Xining Central Measurement Center in 2021 was 9.12. The groundwater level showed little change and was in a relatively stable state. At coordinates XZ04, located near Saihantaolai Sumu Township, Ejin Banner, Inner Mongolia Autonomous Region, at 100°42′38.14″ and 41°57′59.04″, the groundwater level depth measured in Gansu Province in 2020 was 2.32, and the groundwater level depth measured in Xining Central Measurement Center in 2021 was 1.56. The groundwater level showed a significant decrease, indicating possible over-extraction or other factors affecting groundwater level changes.

[0044] Table 2. Wellpoint Water Level Drop Points

[0045] For different types of vegetation, regression models are used to derive thresholds for ecologically sensitive areas, generating ecological water level thresholds and boundary thresholds. By analyzing the changing trends of NDVI, the potential for ecological restoration is calculated. If the difference between the current NDVI and the historical NDVI is lower than the average boundary threshold of the NDVI mean, it indicates poor ecological resilience in the region, and the water level threshold needs to be lowered to enter a more stringent alert state. Analyzing the changing trends of NDVI can effectively determine the ecological restoration status of the region, thereby adjusting the thresholds and managing potential ecological risks. Ecological water level thresholds include: perennial threshold, low-water threshold, restricted extraction threshold, and water transfer threshold; boundary thresholds include: primary threshold, sensitive threshold, and acute threshold. The primary threshold is the lowest water level that ensures the health and growth of the ecosystem, the water level at which vegetation roots can directly access groundwater. When the water level is above this threshold, the ecosystem is in a healthy state. The sensitive threshold indicates the water level at which the ecosystem begins to decline; when the water level approaches this threshold, monitoring and protection measures need to be taken in advance. When the groundwater level drops below the acute threshold, ecological restoration is almost complete. Impossible; emergency measures must be initiated. When the data in the collected multi-source dataset meets the anomaly triggering conditions and the wellpoint-level confidence value is higher than the minimum confidence threshold, the system automatically triggers anomaly identification and reporting. Anomaly triggering conditions include, but are not limited to, the following types: single threshold trigger, combined logic trigger, trend deviation trigger, and distribution comparison trigger. Single threshold trigger refers to a drop in groundwater level or a shallow burial depth exceeding the ecological sensitivity threshold. Combined logic trigger refers to a drop in water level accompanied by a continuous decline in remote sensing ecological indicator data. Trend deviation trigger refers to a drop in water level or a fluctuation range deviating from the statistical range of historical healthy samples. Distribution comparison trigger refers to a difference between the wellpoint water level and the geophysical inferred burial depth. Anomaly warning information is pushed out, and the warning response phase begins. Anomaly event tags and triggering reasons are recorded to form a traceability basis. The digital twin computing module and threshold verification mechanism are activated to achieve source tracing analysis and ecological risk assessment of the anomaly state. Through the joint judgment of confidence and anomaly detection results, false alarms and missed alarms caused by low-quality data can be effectively avoided, enhancing the accuracy and reliability of groundwater ecological monitoring and early warning.

[0046] This implementation plan constructs a comprehensive sensing system for groundwater and the ecological environment by collecting and processing multi-source data in real time, ensuring data consistency in both space and time. Combining wellpoint water levels, groundwater depth, and remote sensing ecological indicators, key ecological water level thresholds are extracted using regression models and NDVI analysis. Through real-time updates using digital twin technology and considering the reliability of wellpoint-level data, water level thresholds are dynamically adjusted to promptly identify potential ecological risks and trigger early warning responses. Multivariate regression analysis is used to optimize water resource allocation strategies, ensuring ecological sustainability while avoiding false alarms and missed alarms, thus improving the accuracy and reliability of water resource management.

[0047] Specifically, the specific steps for delineating water resource safety zones and implementing early warning and response measures by combining multi-source datasets are as follows: Different levels of water resource safety zones are delineated using groundwater level changes and groundwater depth data. These different levels reflect the degree of impact of groundwater resources on the ecosystem. The water resource safety zone levels include: safe zones, warning zones, and risk zones. A safe zone is defined as a water level above the primary threshold, a healthy ecosystem, NDVI changes within the statistical fluctuation range of the past few months, and no decline in vegetation growth. A warning zone is defined as a water level close to the sensitivity threshold, a continuous decline in NDVI, water level fluctuations, and a decline in vegetation growth. A risk zone is defined as a water level drop exceeding the sensitivity threshold, a decrease in NDVI, an inability for vegetation to recover, and a water level below the acute threshold.

[0048] When entering the early warning response phase, different early warning measures are implemented based on different triggering conditions. Early warning responses include: yellow zone response, orange zone response, and red zone response. Yellow zone response refers to increasing the frequency of wellhead inspections when the water resource safety zone is in the warning zone, ensuring the safety of equipment at each monitoring point and the accuracy of data recording, initiating more frequent water level monitoring, increasing monitoring intensity, monitoring changes in NDVI and remote sensing ecological indicators, assessing the health status of vegetation, and evaluating hydrological and meteorological conditions based on surface water and meteorological data to ensure timely detection of early signs of ecological changes. Orange zone response refers to implementing rotational irrigation in the affected area when the water level exceeds the sensitivity threshold and the water resource safety zone is in the warning zone. Dispatch involves detailed recording of water usage in affected areas to ensure transparency in water resource allocation and prevent water waste. Red zone response refers to the activation of extraction restriction measures when water levels fall below the critical threshold and the water resource safety zone is in a risk zone. This reduces groundwater extraction, prioritizes water resources for the ecological environment and human life, initiates river ecological water replenishment programs, allocates water from rivers and lakes to replenish groundwater, ensures continuous water supply, and mandates the implementation of ecological water replenishment programs, including irrigation system regulation and river ecological water replenishment. This ensures effective utilization of water resources during ecological restoration, strengthens water source protection in ecological areas, strictly controls pollution source emissions, and guarantees the purity of ecological water replenishment.

[0049] The ecological water replenishment plan includes ecological protection and restoration strategies and water balance optimization schemes. Regarding ecological protection and restoration strategies, for areas with excessive groundwater extraction and ecological degradation, and in areas where groundwater levels have dropped to sensitive thresholds, a combination of river ecological water replenishment and artificial recharge will be implemented. This will involve replenishing groundwater through river diversion, lake regulation, and flood irrigation / infiltration. For severely degraded oasis fringe areas, rotational irrigation and zoned water replenishment will be implemented, prioritizing the restoration of root zone water levels during the dry season to prevent further ecological degradation. For areas with significant ecological degradation, ecological replanting and vegetation type reconstruction projects will be implemented, prioritizing drought-resistant, deep-rooted, and ecologically sound native plants to restore surface cover and improve regional evapotranspiration stability. Water quality and pollution source control in ecological red zones will be strengthened, and ecological purification zones and monitoring buffer zones will be established to prevent recharged water from polluting the groundwater system.

[0050] Regarding water balance optimization strategies, multi-source data fusion is employed, integrating groundwater level monitoring, surface runoff, meteorological precipitation, irrigation water intake, and evapotranspiration data to establish a basin-wide water budget model. This model dynamically tracks the replenishment, discharge, and consumption processes within the region. Based on monitoring results, the balance relationships of precipitation replenishment, irrigation infiltration, lateral infiltration replenishment, evaporation, extraction, and outflow are analyzed in real time. Water imbalance zones are automatically identified, and a dynamic threshold correction mechanism adjusts the water intake and replenishment ratios. Based on the water level response and vegetation water demand index of ecologically sensitive areas, differentiated management is implemented by zone. Groundwater flow direction and distribution are visualized and predicted. Combining historical samples and real-time monitoring data, hydrological parameters and flow allocation strategies are continuously revised, ultimately achieving long-term dynamic stability of regional water balance by prioritizing ecological water use, optimizing agricultural water use, and ensuring regional water balance.

[0051] After completing the corresponding early warning and response measures, detailed labeling and recording are required to establish a traceability mechanism. This ensures that when an anomaly occurs, the cause, process, response measures, and final effect of the event can be traced, providing basic data for future risk management and early warning optimization. Labeling content includes: triggering cause, time and location, early warning response measures, and affected area. The triggering cause describes the specific conditions that triggered the abnormal event, such as a drop in groundwater level, changes in ecological health indicators, or external environmental impacts. The time and location record the time and specific location of the event for comparison and traceability. The early warning response measures record the early warning measures taken and their implementation status, including wellhead inspections and irrigation scheduling. The system includes: limited extraction measures; designated areas to indicate the specific regions where anomalies occur, and descriptions of the affected ecosystems and irrigation areas within those areas; a traceability mechanism including event effect analysis, time-series comparison, and risk assessment and improvement; event effect analysis records the effects of each early warning response, such as NDVI trends and water level recovery, to help analyze the actual effectiveness of response measures; time-series comparison compares data before and after the anomaly to examine changes in the ecosystem and assess the effectiveness of early warning measures; and risk assessment and improvement refers to the system's periodic generation of reports based on anomaly event tags and response data, analyzing anomaly types, response efficiency, and areas for optimization in water resource management.

[0052] This implementation plan delineates different levels of water resource safety zones by combining groundwater level changes and depth data, specifically dividing them into safe zones, warning zones, and risk zones. These safety zone levels reflect the degree of impact of groundwater resources on the ecosystem and provide a basis for early warning and response measures. During the early warning and response phase, based on different triggering conditions, yellow, orange, and red zone response measures are adopted, respectively, through increasing inspection frequency, implementing rotational irrigation scheduling, and initiating limited extraction measures, to ensure the protection of the ecological environment and human water resource needs. After the response measures are implemented, a traceability mechanism is established through tagging records to conduct detailed analysis and evaluation of abnormal events, ensuring the effectiveness of early warning measures and providing data support for future risk management and optimization. Tag content includes the triggering cause, response measures, and affected areas, forming a complete event traceability system to help continuously improve water resource management and early warning response efficiency.

[0053] Specifically, the steps for establishing a three-dimensional hydrogeological model and dynamically correcting the water level threshold are as follows: A comprehensive simulation of the flow, infiltration, recharge, and discharge processes of groundwater is conducted using a multidimensional dataset. Through spatial interpolation and dynamic simulation, a three-dimensional hydrogeological model is established to accurately reflect the multidimensional changes in the groundwater system, helping to assess the dynamic evolution of groundwater resources and their impact on the ecological environment. By simulating the flow path, recharge area, and discharge area of ​​groundwater, potential water resource risk areas are identified, and abnormal water level changes are detected in a timely manner, supporting intelligent water resource and ecological environment management. Based on real-time monitoring data and the simulation results of the three-dimensional hydrogeological model, the water level threshold is reviewed and corrected.

[0054] The following terms are used: a baseline water level threshold as the baseline threshold term; the product of the water quantity deficit anomaly field coefficient and the water quantity deficit anomaly field as the water quantity deviation contribution term; the product of the ecological response gap field coefficient and the ecological response gap field as the ecological gap correction term; the product of the structural sensitivity field coefficient and the structural sensitivity field as the structural vulnerability modulation term; the product of the model-observation bias coefficient and the model-observation bias as the model correction compensation term; and the product of the certainty penalty coefficient and the certainty penalty term as the uncertainty suppression term. The corrected dynamic threshold is obtained by summing these six terms: baseline threshold term, water quantity deviation contribution term, ecological gap correction term, structural vulnerability modulation term, model correction compensation term, and uncertainty suppression term. Dynamic threshold correction allows for timely adjustment of the threshold to maintain its adaptability to the ecological environment. The specific calculation formula for the corrected dynamic threshold is as follows: ; In the formula, This indicates the corrected dynamic threshold. Indicates the current time; Indicates spatial location; The reference water level threshold can be a primary threshold, a sensitive threshold, or an emergency threshold, representing the reference water level in the current groundwater system. The coefficient representing the abnormal field of water balance is obtained by modeling and statistical analysis of precipitation, evapotranspiration and water intake data. Its value ranges from 0.2 to 0.5 and is used to balance the impact of water balance on water level threshold adjustment. The ecological response gap field coefficient is calculated by examining the correlation between vegetation health and water level, and obtained through regression analysis. Its value ranges from 0.2 to 0.5, and it is used to correct for the impact of ecological factors on water level threshold adjustment. The structural sensitivity field coefficient is obtained by spatial interpolation using geophysical exploration data, and its value ranges from 0.2 to 0.5. It is used to measure the impact of structural changes on threshold adjustment. The model-observation bias coefficient is derived by comparing the deviation between the simulation results of the three-dimensional hydrogeological model and the well point monitoring data using statistical analysis methods. Its value ranges from 0.2 to 0.5 and is used to adjust the threshold according to the difference between the model and the actual observation data. This represents the certainty penalty coefficient, which is obtained by performing uncertainty analysis on a multidimensional dataset. Its value ranges from 0.2 to 0.5 and is used to correct the threshold of high uncertainty regions, thereby increasing conservatism. It represents an abnormal water balance field, which is calculated by combining meteorological and hydrological data to analyze the balance and reflect the fluctuations in groundwater level caused by changes in hydrological processes. This represents the ecological response gap field, obtained through correlation analysis between remote sensing data and water level changes, reflecting the relationship between groundwater level changes and the ecosystem; It represents the structural sensitivity field, which reflects the changes in the physical structure of the groundwater layer and the local instability of groundwater flow through sensitivity analysis. The model-observation bias is represented by the error value calculated by comparing the output of the three-dimensional hydrogeological model with the actual observation data, which measures the difference between the simulated groundwater level and the actual observation data. This represents a certainty penalty term, calculated by error analysis and confidence interval calculation of the threshold band width. It is used to adjust the width of the threshold band and increase the conservatism of the high uncertainty region.

[0055] Based on the revised dynamic threshold, water resource management is optimized in real time, and water level thresholds are updated regularly. Zonal management is implemented in risk areas, and water resource allocation strategies are adjusted to ensure that resources support ecological restoration and human needs to the greatest extent. For risk areas, extraction restriction and ecological water replenishment measures are promptly initiated to reduce groundwater extraction and prioritize the supply of water for ecological environment and farmland irrigation. Water level thresholds are updated on a rolling basis at ten-day or monthly intervals to ensure that management strategies are always consistent with the actual state of groundwater resources and ecological environment, thereby achieving water resource and ecological environment management. When the change in water level threshold is less than the allowable change range over several consecutive periods, and the difference between the measured data and the revised dynamic threshold is within the difference range threshold, the threshold is considered to be stable, and routine management measures are adopted.

[0056] This implementation plan simulates the flow, infiltration, recharge, and discharge processes of groundwater using a multidimensional dataset, and establishes a three-dimensional hydrogeological model using spatial interpolation. Based on real-time monitoring data and model simulation results, a revised dynamic water level threshold is calculated. A rolling update cycle (ten-day or monthly) and a threshold stability assessment mechanism ensure the timeliness and accuracy of threshold updates, while also guaranteeing consistency between management strategies and the actual conditions of groundwater resources and the ecological environment. This approach solves the problems of fixed and rigid water level thresholds in traditional management, which cannot dynamically respond to hydrological changes and ecological needs, and the disconnect between model simulations and actual control measures. It provides a data and technological foundation for intelligent management of groundwater resources and the ecological environment, improves the ecological adaptability and management accuracy of water level control, and enhances the scientific effectiveness and traceability of water resource allocation and ecological replenishment measures.

[0057] Specifically, the steps for conducting a comprehensive assessment based on the revised dynamic thresholds, generating periodic reports, and writing back the governance results to form a closed loop are as follows: A comprehensive assessment mechanism is constructed based on the revised dynamic thresholds to periodically evaluate the effectiveness of water resource use and ecological restoration; this comprehensively reflects the current state of water resource and ecological management, identifies potential risk areas, and promptly initiates corresponding management measures. The threshold trigger delay value multiplied by the response weight coefficient is used as the timeliness term; the comprehensive judgment value multiplied by the effect weight coefficient is used as the effectiveness term; the hydrological sequence value multiplied by the stability weight coefficient is used as the steady-state term; and the correction ratio multiplied by the learning weight coefficient is used as the enhancement term. The sum of the timeliness, effectiveness, steady-state, and enhancement terms yields the comprehensive assessment value; this helps to optimize water resource management strategies in real time. The specific calculation formula for the comprehensive assessment value is as follows: ; In the formula, This represents the threshold trigger delay value, which is obtained by comparing the delay difference between the water level threshold triggering the early warning and the actual response measures taken, thus measuring the delay from threshold triggering to actual response. The comprehensive judgment value is obtained by using regression analysis and correlation analysis statistical methods to measure the actual impact of water level changes on vegetation growth through correlation analysis of water level changes and vegetation health. The hydrological sequence value is obtained by comparing the multiplication deviation between the prediction results of the three-dimensional hydrogeological model and the actual observation data, and the magnitude of the error is calculated to measure the stability of the system. The correction ratio is calculated by using a convergence analysis method to compare the deviations between the prediction results of the three-dimensional hydrogeological model and the actual observation data, reflecting the system's learning and self-optimization capabilities. The response weighting coefficient is obtained through statistical analysis and sensitivity analysis of historical health samples. It is used to characterize the relative importance of responsiveness in the comprehensive evaluation, and its value ranges from 0.2 to 0.5. The effect weighting coefficient is obtained by quantifying the correlation analysis between water level changes and vegetation health. It is used to enhance the degree of influence of ecological effectiveness in the comprehensive evaluation, and its value ranges from 0.2 to 0.5. The stability weighting coefficient is obtained by calculating the ratio of false alarms to missed alarms and performing parameter fitting. It is used to emphasize the contribution of early warning stability to system evaluation, and its value ranges from 0.2 to 0.5. The learning weight coefficient is obtained through real-time monitoring data and dynamic threshold evaluation analysis. It is used to reflect the weight of the system's intelligent evolution capability in the comprehensive evaluation, and its value ranges from 0.2 to 0.5.

[0058] The system compares the comprehensive assessment value with the decision threshold in real time. When the comprehensive assessment value exceeds the decision threshold, emergency response measures are taken to control water resource use in a timely manner. Emergency response measures include initiating emergency water resource management, restricting groundwater extraction; initiating river ecological water replenishment plans to increase water supply; increasing the frequency of on-site inspections; adjusting water level thresholds in a timely manner; strengthening water source protection in ecological areas and strictly controlling pollution source emissions; and initiating ecological restoration plans for ecologically degraded areas. When the comprehensive assessment value is less than or equal to the decision threshold, it indicates that the system is in a relatively stable state, and routine management measures are taken. Routine management measures include maintaining the existing water resource management strategy to ensure that the water level is maintained within a reasonable range; and conducting routine wellhead inspections and water level monitoring to ensure the accuracy and real-time nature of the data.

[0059] The comprehensive evaluation results are shown in Table 3. The water level change of monitoring well W010 is -0.92, indicating a drop in water level. The water quality score is 73.9, the ecological health index is 0.48, the threshold trigger delay value is 12.3, the comprehensive identification value is 0.69, the hydrological sequence value is 0.59, the correction ratio is 0.85, the response weight coefficient is 0.20, the effect weight coefficient is 0.36, the stability weight coefficient is 0.21, the learning weight coefficient is 0.22, the comprehensive evaluation value is 3.06, the risk level is 2, the safe zone status is safe, the condition requires intervention, and the management measures are emergency response measures. The water level change of monitoring well W011 is 0.25, the water quality score is 61.2, the ecological health index is 0.74, the threshold trigger delay value is 2.4, the comprehensive identification value is 0.87, the hydrological sequence value is 0.16, the correction ratio is 0.4, the response weight coefficient is 0.20, the effect weight coefficient is 0.36, the stability weight coefficient is 0.21, the learning weight coefficient is 0.22, the comprehensive evaluation value is 0.92, the risk level is 2, the safe zone status is safe zone, the status is that intervention is required, and the management measures are emergency response measures. The water level change of monitoring well W012 is -0.64, the water quality score is 70.2, the ecological health index is 0.11, the threshold trigger delay value is 1.5, the comprehensive identification value is 0.17, the hydrological sequence value is 0.41, the correction ratio is 0.25, the response weight coefficient is 0.20, the effect weight coefficient is 0.36, the stability weight coefficient is 0.21, the learning weight coefficient is 0.22, the comprehensive evaluation value is 0.5, the risk level is 2, the safety zone status is safe, the status is normal, and the management measures are routine management measures.

[0060] Table 3. Comprehensive Evaluation Results of Dynamic Thresholds

[0061] By integrating and visualizing the status of the monitoring well network, ecological health indicators, and safety zones, a global view is provided, enabling real-time monitoring of the latest dynamics of groundwater resources and the ecological environment. Weekly, monthly, and quarterly reports for individual wells and zones are automatically generated, including water level change trends, ecological restoration status, and risk level changes. The automated reporting system reduces manual analysis work, improving the efficiency and accuracy of decision-making. The implementation results of water extraction restriction and water conveyance management measures are automatically written back, forming a complete feedback loop. Through closed-loop management, water resource allocation and ecological restoration strategies are automatically adjusted based on real-time monitoring data and assessment results, ensuring the effective utilization of resources and the continuous improvement of the ecological environment.

[0062] In this implementation plan, a comprehensive evaluation mechanism is constructed by combining the revised dynamic threshold, and then the comprehensive evaluation value is compared with the decision threshold in real time: when the evaluation value exceeds the threshold, emergency measures such as limiting groundwater extraction and ecological water replenishment are initiated; when it is below the threshold, the existing strategy is maintained and routine inspections and management are carried out. At the same time, a visual evaluation table integrating the status of the monitoring well network and ecological indicators is output, a report is automatically generated, and the results of limiting extraction and water conveyance management are written back to form a feedback loop. Through closed-loop management and dynamic adjustment of strategies, the problems of single evaluation dimensions and delayed response in traditional management are solved, providing support for the refined management of groundwater resources and the ecological environment, improving the timeliness of measures and the scientific nature of decision-making, and ensuring the effective utilization of resources and continuous ecological improvement.

[0063] Specifically, such as Figure 2The diagram shown illustrates the structure of an ecological threshold extraction and monitoring early warning system based on multi-source data provided in this application embodiment. The second aspect of this invention provides an ecological threshold extraction and monitoring early warning system based on multi-source data, applied to the aforementioned ecological threshold extraction and monitoring early warning method based on multi-source data. It includes a groundwater multi-source data acquisition and management module for acquiring multi-source datasets. Through NTP millisecond-level time synchronization technology, it ensures complete alignment of each data source in the time dimension, eliminating time differences between different devices. Furthermore, normalization processing technology is employed to standardize and unify the acquired multi-source data, eliminating data heterogeneity and dimensional differences, thereby providing a solid data foundation for data analysis. Through data quality inspection and anomaly identification, a unified database is established, and based on this, ecological health steady-state time periods are identified to form historical health samples, serving as a stable benchmark for ecological threshold extraction and groundwater anomaly diagnosis. A 3D hydrogeological digital twin module is used to construct a 3D aquifer structure using historical health samples and a hierarchical interpolation method. This simulates the groundwater flow field and identifies recharge zones, resulting in groundwater depth zoning. Different risk levels are defined, and key areas in groundwater flow are identified, providing decision support for the rational utilization of groundwater resources and ecological protection. The final output of groundwater depth zoning provides foundational data for ecological water level threshold extraction and risk assessment. The ecological effective water level threshold extraction module combines multi-source datasets with risk levels to derive ecologically sensitive area thresholds through regression models. These thresholds include the root-reachable water depth critical value, the ecological degradation threshold inflection point, risk red zones, and oasis edge sensitivity values. The root-reachable water depth critical value is calculated using an ecological model combining plant root growth patterns and water demand; the ecological degradation threshold is derived through correlation analysis between groundwater depth and NDVI; and the oasis edge sensitivity value is based on statistical analysis of wellpoint water level data and the rise and fall points of remote sensing ecological indicators. Early warning responses are generated by combining anomaly triggering conditions and wellpoint-level reliability values, ensuring the system can promptly identify potential ecological risks and trigger corresponding protection measures. The dynamic water resource security zone hierarchical early warning module is used to delineate water resource security zones by combining multi-source datasets, implement early warning and response, establish a three-dimensional hydrogeological model, dynamically correct water level thresholds, adjust the safety boundaries of the area according to real-time hydrological changes and ecological needs, and generate visualized hierarchical early warning reports to provide decision-makers with clear action guidelines. The closed-loop management and evaluation module is used to conduct comprehensive evaluations based on the corrected dynamic thresholds, generate periodic reports, record in detail the changes in water resources and the ecological environment, and provide feedback on the governance results, writing the governance results back to form a closed loop.

[0064] In this implementation plan, the groundwater multi-source data acquisition and management module eliminates data heterogeneity and dimensional differences through millisecond-level time synchronization and normalization processing. After quality testing and database construction, it identifies ecologically healthy steady-state time periods to form historical healthy samples. The three-dimensional hydrogeological digital twin module uses these samples to construct a three-dimensional aquifer structure and simulate the groundwater flow field using a layered interpolation method, dividing groundwater burial depth zones and risk levels. The ecologically effective water level threshold extraction module combines multi-source data and risk levels to derive the critical value of the root system's reachable water depth through a regression model method, and combines it with abnormal condition early warning response. The dynamic water resource safety zone hierarchical early warning module delineates safety zones, dynamically corrects water level thresholds, and generates hierarchical early warning reports. The closed-loop management and evaluation module combines dynamic threshold comprehensive evaluation, generates periodic reports, and writes back the management results to form a closed loop. The system solves the problems of asynchronous monitoring data, inaccurate threshold extraction, and delayed early warning response in traditional systems, providing full-process technical support for groundwater resource protection and ecological risk prevention and control, and improving the scientific nature and timeliness of management decisions.

[0065] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0066] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for ecological threshold extraction and monitoring early warning based on multi-source data, characterized in that, Includes the following steps: S1: Collect multi-source datasets, perform NTP millisecond-level synchronization and normalization processing, and build a unified database to identify ecological health steady-state time periods to form historical health samples; S2 uses historical health samples to construct a three-dimensional aquifer structure through hierarchical interpolation, simulates the groundwater flow field and identifies the recharge zone, obtains groundwater depth zoning results, and classifies different risk levels. S3 combines multi-source datasets and risk levels to derive thresholds for ecologically sensitive areas through a regression model, and then uses anomaly triggering conditions and well-point level confidence values ​​to provide early warning responses. S4, combined with multi-source datasets, delineates water resource safety zones, implements early warning and response, establishes a three-dimensional hydrogeological model, and dynamically corrects water level thresholds; S5, combined with the revised dynamic threshold, performs a comprehensive assessment, generates a periodic report, and writes back the governance results to form a closed loop.

2. The method for ecological threshold extraction and monitoring early warning based on multi-source data according to claim 1, characterized in that: The specific steps for collecting multi-source datasets and performing NTP millisecond-level synchronization and normalization processing to build a unified database are as follows: The collaborative collection of multi-source datasets was carried out based on the comprehensive natural resource observation system in the lower reaches of the Heihe River. The multi-source datasets include: well point monitoring datasets, geophysical exploration datasets, remote sensing ecological index datasets, surface water and meteorological datasets, and irrigation and water intake management datasets. The multi-source datasets were unified through millisecond-level time synchronization, data calibration, continuity assurance, credibility labeling, and spatial consistency. The time frequency, amplitude, and spatial scale were normalized according to the sampling frequency of ten days or monthly, and a unified database was constructed.

3. The method for ecological threshold extraction and monitoring early warning based on multi-source data according to claim 1, characterized in that: The specific steps for identifying historical health samples during ecological health steady-state periods are as follows: By using rolling analysis within a fixed time window, the system automatically identifies time periods when the region is in a state of ecological health. Data from all multi-source datasets within these ecologically healthy time periods are saved as historical health samples. The missing rate is obtained by comparing the number of valid records collected from each monitoring well in the wellpoint monitoring dataset within a unified sampling period with the number of records that should be collected in that period. The drift rate is obtained by comparing the water level observations in the wellpoint monitoring dataset with the water levels within the ecologically healthy time periods and performing a stability comparison analysis. The anomaly rate is obtained by comparing the proportion of data in the wellpoint monitoring dataset and the remote sensing ecological indicator dataset that are abnormal or inconsistent with adjacent time windows within a unified sampling period. The complementary value of the missing rate multiplied by the missing rate weighting coefficient is used as the integrity credibility contribution; the complementary value of the drift rate multiplied by the drift rate weighting coefficient is used as the stability credibility contribution; and the complementary value of the anomaly rate multiplied by the anomaly rate weighting coefficient is used as the consistency credibility contribution. The integrity credibility contribution, stability credibility contribution, and consistency credibility contribution are summed to obtain the wellpoint-level credibility value. The system compares the wellpoint-level confidence value with the minimum allowable confidence threshold in real time. When the wellpoint-level confidence value is higher than the minimum allowable confidence threshold, the monitoring data quality is deemed to meet the conditions for anomaly identification and trigger judgment, and anomaly triggering is initiated. If the wellpoint-level confidence value is lower than the minimum allowable confidence threshold, anomaly triggering is not executed immediately. Instead, the wellpoint monitoring data is marked as data to be verified, and supplementary measurement or consistency comparison with adjacent wellpoint data and remote sensing data is triggered to avoid misjudgment or omission due to insufficient data quality.

4. The method for ecological threshold extraction and monitoring early warning based on multi-source data according to claim 1, characterized in that: The specific steps for constructing a three-dimensional aquifer structure using historical health samples and a hierarchical interpolation method to simulate the groundwater flow field and identify recharge zones are as follows: The well depth, aquifer position, and filter location of the well point monitoring dataset are obtained. Combined with the groundwater depth isosurface, hydraulic gradient direction, and low resistivity anomaly distribution in the geophysical exploration dataset, the three-dimensional structure of the aquifer depth is obtained through layered interpretation and spatial stratification Kriging interpolation method. The output results include the distribution of aquifer thickness, depth undulation, and permeability difference zones in the region. Using the constructed three-dimensional structure of aquifer burial depth as the spatial boundary framework, boundary conditions were set according to the hydrogeological unit division and groundwater flow field distribution characteristics: the hydraulic connection of the river section was used as the constant head boundary; the first-level watershed line of the region and the area without hydraulic recharge at the outer edge of the desertification zone were used as the flow-free boundary; surface water and meteorological datasets and irrigation and water intake management datasets were integrated, and groundwater flow direction simulation was established through the water level burial depth gradient field; by integrating farmland irrigation rotation records, groundwater intake data and meteorological drivers, sensitivity attribution analysis was conducted to construct a multi-source driving contribution rate decomposition model; and groundwater volume changes were decomposed into contribution items of different driving mechanisms through multiple regression, principal component analysis and impulse response decomposition statistical methods.

5. The method for ecological threshold extraction and monitoring early warning based on multi-source data according to claim 1, characterized in that: The specific steps for obtaining the groundwater depth zoning results and classifying different risk levels are as follows: By using methods such as underground resistivity, ground vibration, gravity anomaly, and geological exploration, groundwater depth data is obtained. Spatial structure consistency verification and anomaly identification are then performed to predict groundwater depth and obtain isostatic depth values ​​for different depth sections. Combined with actual water level and well depth data from wellpoint monitoring datasets and compared with isostatic depth surface layers obtained from geophysical exploration, groundwater depth zoning is formed. Spatial interpolation technology is used to perform three-dimensional reconstruction of the depth values ​​to obtain groundwater depth zoning results. Using the groundwater depth zoning results, ecologically sensitive areas are divided into different risk levels.

6. The method for ecological threshold extraction and monitoring early warning based on multi-source data according to claim 1, characterized in that: The specific steps for combining multi-source datasets and risk levels to derive thresholds for ecologically sensitive areas through regression models, and then combining anomaly triggering conditions and well-point level confidence values ​​to conduct early warning responses are as follows: Key ecological water level thresholds are extracted from well point monitoring data, remote sensing ecological indicators, groundwater depth and other zoning results; for different types of vegetation, ecologically sensitive area thresholds are obtained through regression models, and ecological water level thresholds and boundary thresholds are generated; when the data in the collected multi-source dataset meets the anomaly triggering conditions and the well point level confidence value is higher than the minimum confidence threshold, the system automatically triggers anomaly identification and reporting actions. Push out abnormal warning information and enter the warning response stage; record the abnormal event label and triggering reason, and start the digital twin computing module and threshold verification mechanism.

7. The method for ecological threshold extraction and monitoring early warning based on multi-source data according to claim 1, characterized in that: The specific steps for delineating water resource security zones and implementing early warning and response measures by combining multi-source datasets are as follows: Different levels of water resource security zones are delineated based on groundwater level changes and groundwater depth data. Water resource safety zones are categorized into three levels: safe zones, warning zones, and risk zones. When entering the early warning response phase, different early warning measures are implemented based on different triggering conditions. Early warning responses include yellow zone response, orange zone response, and red zone response. After completing the corresponding early warning measures, detailed labeling and recording are required, and a traceability mechanism must be established.

8. The method for ecological threshold extraction and monitoring early warning based on multi-source data according to claim 1, characterized in that: The specific steps for establishing a three-dimensional hydrogeological model and dynamically correcting the water level threshold are as follows: By comprehensively simulating the flow, infiltration, recharge, and discharge processes of groundwater using a multidimensional dataset, and establishing a three-dimensional hydrogeological model through spatial interpolation and dynamic simulation, the water level threshold is reviewed and corrected based on real-time monitoring data and the simulation results of the three-dimensional hydrogeological model. The following terms are used as a framework: a baseline water level threshold is used as the baseline threshold term; the product of the water quantity deficit anomaly field coefficient and the water quantity deficit anomaly field is used as the water quantity deviation contribution term; the product of the ecological response gap field coefficient and the ecological response gap field is used as the ecological gap correction term; the product of the structural sensitivity field coefficient and the structural sensitivity field is used as the structural vulnerability modulation term; the product of the model-observation bias coefficient and the model-observation bias is used as the model correction compensation term; and the product of the certainty penalty coefficient and the certainty penalty term is used as the uncertainty suppression term. The six terms—the baseline threshold term, the water quantity deviation contribution term, the ecological gap correction term, the structural vulnerability modulation term, the model correction compensation term, and the uncertainty suppression term—are summed to obtain the corrected dynamic threshold. Based on the corrected dynamic threshold, water resource management is optimized in real time, and the water level threshold is updated periodically.

9. The method for ecological threshold extraction and monitoring early warning based on multi-source data according to claim 1, characterized in that: The specific steps for combining the corrected dynamic threshold, conducting a comprehensive evaluation, generating a periodic report, and writing back the governance results to form a closed loop are as follows: By combining the corrected dynamic threshold, a comprehensive evaluation mechanism is constructed to periodically assess the effects of water resource use and ecological restoration. The threshold trigger delay value is multiplied by the response weight coefficient as the timeliness term, the comprehensive judgment value is multiplied by the effect weight coefficient as the effectiveness term, the hydrological sequence value is multiplied by the stability weight coefficient as the steady-state term, and the correction ratio is multiplied by the learning weight coefficient as the enhancement term. The comprehensive evaluation value is obtained by summing the four terms: timeliness term, effectiveness term, steady-state term, and enhancement term. The system compares the comprehensive evaluation value with the decision threshold in real time. When the comprehensive evaluation value is greater than the decision threshold, emergency response measures are taken to control water resource use in a timely manner. When the comprehensive evaluation value is less than or equal to the decision threshold, it indicates that the system is in a relatively stable state, and routine management measures are taken. By integrating and visualizing the status of the monitoring well network, ecological health indicators, and safety zones, weekly, monthly, and quarterly reports for individual wells and zones are automatically generated. The implementation results of water extraction restriction and water conveyance management measures are automatically written back, forming a complete feedback loop.

10. An ecological threshold extraction and monitoring early warning system based on multi-source data, employing the ecological threshold extraction and monitoring early warning method based on multi-source data as described in any one of claims 1-9, characterized in that, include: The groundwater multi-source data acquisition and management module is used to collect multi-source datasets, perform NTP millisecond-level synchronization and normalization processing, build a unified database, and identify ecological health steady-state time periods to form historical health samples. The three-dimensional hydrogeological digital twin module is used to construct a three-dimensional aquifer structure using historical health samples and hierarchical interpolation, simulate the groundwater flow field and identify recharge areas, obtain groundwater depth zoning results, and classify different risk levels. The ecological effective water level threshold extraction module is used to combine multi-source datasets and risk levels to derive the threshold of ecologically sensitive areas through a regression model, and to conduct early warning response by combining abnormal triggering conditions and well point-level confidence values. The dynamic water resource security zone hierarchical early warning module is used to delineate water resource security zones by combining multi-source datasets, implement early warning and response, establish a three-dimensional hydrogeological model, and dynamically correct water level thresholds; The closed-loop management and evaluation module is used to conduct a comprehensive assessment by combining the revised dynamic thresholds, generate periodic reports, and write back the governance results to form a closed loop.

Images