A method and system for determining an ecological water level threshold of a plateau lake based on multi-stressor identification

Abstract

The present application provides a kind of plateau lake ecological water level threshold determination method and system based on multiple stressor identification, it is related to ecological protection and water resources management technical field.The method comprises: collecting multi-source data preprocessing and constructing standardized data set;Weighted calculation lake comprehensive ecological health index;Screening key stressor and dimensionless;Gradient boosting decision tree is used to build multiple stress-ecological response coupling model, quantifies factor independent contribution and interaction;Using Pettitt mutation test to determine the health mutation critical value, inversion minimum ecological water level red line;Simulate multiple water level scenarios, draw response curve, determine suitable ecological water level interval in combination with water balance and water exchange period verification;Verify threshold rationality and establish dynamic updating mechanism.The existing lake ecological water level threshold determination method cannot systematically identify and quantify the synergistic effect of multiple stressors, resulting in the problem that the threshold determined cannot effectively correlate water level protection and ecological restoration.

Description

Technical Field

[0001] This invention relates to the field of ecological protection and water resource management technology, and more specifically, to a method and system for determining the ecological water level threshold of plateau lakes based on the identification of multiple stress factors. Background Technology

[0002] Plateau lakes are the core carriers of regional water resource reserves and the maintenance of ecosystem services. Their ecological water level thresholds are key control parameters for ensuring that the lake ecosystem does not suffer severe degradation and for achieving sustainable water resource utilization. Ecological water level thresholds typically include two levels: the minimum ecological water level and the suitable ecological water level. The former is the baseline water level for maintaining the basic ecological functions of the lake and preventing ecosystem collapse, while the latter is the target water level for supporting the restoration of aquatic vegetation, inhibiting eutrophication, and promoting positive ecosystem succession. Scientifically determining these thresholds is of significant practical importance for coordinating the contradiction between watershed economic development and lake ecological protection, and for guiding the scheduling of water diversion projects from other watersheds.

[0003] Typical plateau lakes, exemplified by Dianchi Lake, face extremely complex and multi-stressed environments in determining their ecological water level thresholds. These include: plateau lakes are mostly in closed or semi-closed states, with long water exchange cycles, poor self-purification capacity, and high sensitivity to pollution loads entering the lakes; rapid urbanization in the basin has led to drastic changes in land use patterns, resulting in a complex superposition of multiple point and non-point source pollution sources, such as industrial wastewater, urban sewage, agricultural non-point sources, and initial urban rainwater runoff, forming a continuous external pollution stress; long-term accumulated polluted lakebed sediments release nutrients under specific hydrodynamic conditions, constituting an endogenous pollution stress that is difficult to eliminate; the basin has a prominent contradiction between water supply and demand, with a large amount of water used for production and domestic purposes crowding out ecological water use, coupled with the fact that evaporation from the lake surface in plateau areas far exceeds precipitation, resulting in a severe shortage of natural water inflow and a water shortage stress; and water diversion projects implemented from other basins to alleviate the water crisis, while supplementing clean water sources to the lakes, have also profoundly altered the natural hydrological rhythms and hydrodynamic conditions of the lakes, triggering an eco-hydrological situation change stress.

[0004] Currently, the mainstream methods for determining the ecological water level threshold of lakes mainly fall into three categories: The first method is the water balance method, which calculates the dynamic balance of factors such as inflow, outflow, precipitation, and evaporation to estimate the minimum amount of water needed to maintain the target water level. This method is sensitive to parameters such as groundwater exchange and is difficult to directly reflect the ecosystem's response to water level. The second method is the minimum water level method, which sets a single control water level based on historical scheduling experience or regulatory requirements. This method is simple to operate and easy to connect with management objectives, but the water level determined can often only meet the minimum survival needs to prevent the lake from drying up, and cannot reflect the requirements of ecological restoration for water level stability and duration. Third is the curve correlation method, which finds the threshold by establishing response curves between water level and single ecological indicators such as chlorophyll a concentration and dominant algal species density. This method can reflect certain ecological response characteristics when there is sufficient data, but it is difficult to fully characterize the comprehensive problem of ecosystem structural degradation under multiple stresses by relying on only a few indicators.

[0005] The aforementioned methods share a common flaw: they all fail to systematically identify and quantify the synergistic effects of multiple stress factors on lake ecosystems. In real-world scenarios, eutrophication stress, hydrological situational change stress, and water scarcity stress do not act independently on lake ecosystems but are coupled and mutually reinforcing. Low water levels not only directly compress aquatic habitats but also amplify the ecological effects of pollution stress by reducing environmental capacity. While diluting nutrients, water diversion from other basins may exacerbate endogenous release from sediments due to hydraulic disturbances, and temperature differences may alter the phenological rhythms of aquatic organisms. Ignoring the interactive and cumulative effects of these multiple stress factors makes it difficult to establish an effective causal relationship between "water quantity assurance" and "ecological restoration" for the ecological water level thresholds determined by existing methods. This often leads to the dilemma in practice where water levels meet standards but the ecosystem does not improve, failing to provide precise scheduling targets for the restoration of aquatic ecosystems in plateau lakes.

[0006] Therefore, there is an urgent need to establish a method that can systematically identify multiple stress factors, quantify their synergistic impact mechanisms, and determine hierarchical ecological water level thresholds accordingly, so as to provide reliable technical support for the ecological protection of plateau lakes and the scientific allocation of water resources. Summary of the Invention

[0007] The purpose of this invention is to provide a method and system for determining the ecological water level threshold of plateau lakes based on the identification of multiple stress factors. This aims to solve the problem that existing methods for determining the ecological water level threshold of lakes cannot systematically identify and quantify the synergistic effects of multiple stress factors, resulting in the inability of the determined threshold to effectively link water level protection and ecological restoration.

[0008] This invention is achieved through the following technical solution: A method for determining the ecological water level threshold of plateau lakes based on the identification of multiple stress factors includes the following steps: Multi-source long-term time series data of plateau lake basins were collected, and the multi-source long-term time series data were subjected to standardized preprocessing such as missing value imputation, outlier removal and time scale unification to construct a standardized dataset; wherein, the multi-source long-term time series data includes hydrological and meteorological data, water quality data, water ecology data, basin socio-economic data and water conservancy project scheduling data; Phytoplankton diversity index, submerged plant coverage, benthic animal richness index, and comprehensive trophic status index were selected as core ecological indicators. The weight of each core ecological indicator was determined by the entropy weight method, and the comprehensive ecological health index of the lake was obtained by weighted calculation, which is used to characterize the overall state of the lake ecosystem. Candidate stress factors were extracted from the standardized dataset. Pearson correlation analysis combined with principal component analysis was used to screen out key stress factors that are related to the state of the lake ecosystem. Dimensionless processing was then performed on each key stress factor. Using the key stress factors as input variables and the comprehensive ecological health index of the lake as output variables, a multi-stress-ecological response coupling model is constructed using the gradient boosting decision tree algorithm, and the model parameters are optimized through cross-validation. Based on the trained multi-stress-ecological response coupling model, the independent contribution of each key stress factor to the ecological health of the lake and the interaction strength among each key stress factor are quantified for the quantitative identification of the synergistic influence mechanism of multiple stress factors. The Pettt mutation test was used to identify the critical value for irreversible mutations in the comprehensive ecological health index of lakes. The corresponding water level range for the critical value was obtained through inversion using a multi-stress-ecological response coupling model. Combined with preset requirements, the minimum ecological water level red line was determined. Different water level scenarios with gradient intervals were set, and the comprehensive ecological health index of lakes under each water level scenario was simulated using the multi-stress-ecological response coupling model. Water level-comprehensive ecological health index response curves were plotted, and the stable range in which the comprehensive ecological health index of lakes reaches the preset ecological restoration target value was identified. The appropriate ecological water level range was determined by combining the water balance method and the water exchange cycle method. The minimum ecological water level red line and the appropriate ecological water level range were used as ecological water level thresholds. The rationality of the ecological water level threshold was verified using historical independent monitoring data, and the differences in the state of the lake ecosystem within and outside the ecological water level threshold range were compared. An annual dynamic update mechanism was established to recalibrate the multi-stress-ecological response coupling model and update the ecological water level threshold based on the newly added monitoring data of the year.

[0009] Optionally, the step of extracting candidate stress factors from the standardized dataset, using Pearson correlation analysis combined with principal component analysis to screen out key stress factors related to the lake ecosystem state, and performing dimensionless processing on each key stress factor specifically includes: Calculate the Pearson correlation coefficient between each candidate stress factor and the lake's comprehensive ecological health index, and screen out candidate stress factors whose absolute value of the correlation coefficient is greater than the first threshold as initial stress factors; Principal component analysis was performed on the initial screening stress factors to extract principal components with eigenvalues ​​greater than the second threshold. The initial screening stress factors with the largest absolute loading value among the principal components were selected and incorporated into the key stress factor set. The key stress factors were dimensionless by using the deviation standardization method.

[0010] Optionally, the key stress factors include eutrophication stress factors, hydrological situation stress factors, water scarcity stress factors, endogenous pollution stress factors, and human activity stress factors; wherein: From the water quality data and the aquatic ecosystem data, total nitrogen concentration, total phosphorus concentration, chlorophyll a concentration, transparency and permanganate index were selected to calculate the comprehensive nutrient status index as the eutrophication stress factor. Based on the hydrological and meteorological data and the water conservancy project scheduling data, the annual water level variation coefficient of the lake, the outflow variation coefficient caused by water diversion from other basins, and the deviation of the water level from the multi-year average natural water level are calculated, and the hydrological situation stress factor is obtained by weighted summation. Based on the hydrological and meteorological data and the water conservancy project scheduling data, the ratio of the water replenishment required to maintain the target ecological water level of the lake to the natural runoff of the watershed is calculated using the water balance equation, and is used as a water shortage stress factor. Using lake sediment monitoring data, the total nitrogen release rate and total phosphorus release rate of the sediment were calculated, normalized and superimposed with the average lake depth to obtain the endogenous pollution stress factor. The total urban domestic sewage discharge, total industrial wastewater discharge, agricultural non-point source nitrogen and phosphorus emission intensity, and the proportion of impermeable area in the lakeside zone were extracted from the socio-economic data of the basin. The first principal component was extracted as a human activity stress factor through principal component analysis.

[0011] Optionally, the specific process of constructing the multi-stress-ecological response coupling model is as follows: The key stress factors after dimensionless processing are used as input variables, and the comprehensive ecological health index of the lake is used as the output variable to construct a sample dataset. The gradient boosting decision tree algorithm is adopted, and the number of decision trees, learning rate and maximum depth parameter range are set. K-fold cross-validation is performed on the sample dataset. The optimal hyperparameter combination is searched with the minimum root mean square error as the criterion to obtain the trained multi-stress-ecological response coupled model. Based on the trained multi-stress-ecological response coupled model, the average contribution of each key stress factor to the reduction of the loss function in the model prediction is calculated one by one, and the independent contribution value of each key stress factor to the ecological health of the lake is obtained. Using the partial dependency function, the ratio of the sum of squares of the differences between the joint partial dependency and the individual marginal partial dependency of any two key stress factors to the total variance is calculated to obtain the interaction strength between the two factors. By traversing all pairwise factor combinations, an interaction strength matrix is ​​formed to complete the quantitative identification of the synergistic influence mechanism of multiple stress factors.

[0012] Optionally, determining the minimum ecological water level red line specifically includes: The historical lake comprehensive ecological health index series is input into the Pettt mutation test, and the corresponding statistics and their significance levels are calculated. When the significance level is less than the third threshold, it is determined that there is an irreversible mutation point in the series, and the lake comprehensive ecological health index corresponding to the irreversible mutation point is used as the critical value. Each of the key stress factors is fixed to its multi-year average value. The multi-stress-ecological response coupled model is trained with water level as the only variable. The predicted value of the lake comprehensive ecological health index corresponding to different water level values ​​is generated. The water level value that causes the predicted value of the lake comprehensive ecological health index to drop to the critical value is determined by linear interpolation. The water level determined by linear interpolation is compared with the minimum control water level set by the lake protection regulations, and the higher of the two water levels is taken as the minimum ecological water level red line.

[0013] Optionally, determining the suitable ecological water level range specifically includes: Using the minimum ecological water level red line as the lower limit and the highest historical operating water level of the lake as the upper limit, a gradient water level scenario set is generated at preset equal intervals; Each water level value in the gradient water level scenario set is taken as a fixed value, while the other key stress factors are kept at the baseline year level. The results are input into the trained multi-stress-ecological response coupling model to obtain the comprehensive ecological health index of the lake corresponding to each water level scenario. A response curve was plotted with water level as the abscissa and lake comprehensive ecological health index as the ordinate. After smoothing the response curve by moving average, the local slope of each point was calculated. Water level intervals with local slope absolute values ​​less than the fourth threshold and lake comprehensive ecological health index continuously greater than the fifth threshold were extracted as candidate suitable water level intervals. For each candidate suitable water level range, the total ecological water replenishment required to maintain the water level of that range throughout the year is calculated using the water balance method, and the corresponding lake water exchange cycle is calculated using the water exchange cycle method. Ranges with ecological water replenishment less than the maximum adjustable water volume of the watershed water diversion project and water exchange cycle less than 1.5 years are selected, and the range with the highest average comprehensive ecological health index of the lake is selected as the suitable ecological water level range.

[0014] Optionally, the establishment of an annual dynamic update mechanism, which recalibrates the multi-stress-ecological response coupling model and updates the ecological water level threshold based on the newly added monitoring data of the year, specifically involves: At the end of each year, newly added hydrological, water quality, water ecology, and water conservancy project scheduling monitoring data are acquired and expanded into the standardized dataset after standardized preprocessing. Based on the existing multi-stress-ecological response coupled model, the model is incrementally trained using the expanded dataset, the tree structure and weights of the gradient boosting decision tree are updated, and the model is recalibrated. A subset of data is extracted from the standardized dataset using a preset sliding time window. The entire process of mutation detection of the lake's comprehensive ecological health index, determination of the minimum ecological water level red line, and determination of the suitable ecological water level range is re-executed for this subset of data to generate the updated ecological water level threshold for the year. The convergence criterion is set as follows: if the variation of the minimum ecological water level red line is less than 0.05 meters for three consecutive years, and the variation of the upper and lower limits of the suitable ecological water level range is less than 0.1 meters, then the current ecological water level threshold is locked and the annual update is stopped; otherwise, the update for the next year will continue.

[0015] Based on the same inventive concept, this invention also provides a system for determining the ecological water level threshold of plateau lakes based on multi-stress factor identification, used to implement the aforementioned method for determining the ecological water level threshold of plateau lakes based on multi-stress factor identification, including: The data acquisition module is used to collect hydrological and meteorological data, water quality data, aquatic ecological data, watershed socio-economic data, and water conservancy project scheduling data of plateau lake basins. It also performs missing value imputation, outlier removal, and time scale unification processing to output standardized datasets. The health index generation module, connected to the data acquisition module, is used to extract phytoplankton diversity index, submerged plant coverage, benthic animal richness index and comprehensive trophic status index from the standardized dataset, determine the weights using the entropy weight method and calculate the weighted average to generate the lake's comprehensive ecological health index. The stress factor screening module is connected to the data acquisition module and the health index generation module, respectively. It is used to extract candidate stress factors from the standardized dataset, screen out key stress factors that are significantly related to the lake's comprehensive ecological health index through Pearson correlation analysis and principal component analysis, and perform dimensionless processing on each key stress factor. The model building and collaborative quantification module is connected to the stress factor screening module and the health index generation module, respectively. It is used to construct a multi-stress-ecological response coupled model by using the dimensionless key stress factors as input and the lake comprehensive ecological health index as output, and using gradient boosting decision tree. Based on the model, it calculates the independent contribution value of each key stress factor and the interaction strength between each pair of factors to form an interaction strength matrix, thereby completing the quantitative identification of the collaborative influence mechanism of multiple stress factors. The water level threshold calculation module, connected to the model construction and collaborative quantification module, is used to identify the irreversible mutation critical value of the lake's comprehensive ecological health index using the Pettitt mutation test method, determine the minimum ecological water level red line through inversion of the multi-stress-ecological response coupling model, set a gradient water level scenario set and drive the multi-stress-ecological response coupling model to generate response curves, and determine the appropriate ecological water level range by combining the water balance method and the water exchange cycle method for verification. The verification and dynamic update module is connected to the water level threshold calculation module, the data acquisition module, and the model construction and collaborative quantization module, respectively. It is used to verify the rationality of the ecological water level threshold using historical independent monitoring data, and to expand the standardized dataset and incrementally train the multi-stress-ecological response coupled model based on the annually added monitoring data. Before meeting the convergence judgment criteria, it updates the minimum ecological water level red line and suitable ecological water level range annually.

[0016] Based on the same inventive concept, the present invention also provides an electronic device, including a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to perform the above-described method for determining the ecological water level threshold of plateau lakes based on the identification of multiple stress factors.

[0017] Based on the same inventive concept, the present invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described method for determining the ecological water level threshold of plateau lakes based on the identification of multiple stress factors.

[0018] The technical solution of the present invention has at least the following advantages and beneficial effects: By constructing a multi-stress-ecological response coupling model based on gradient boosting decision trees, we can quantitatively identify the independent contributions and interaction strengths of multiple stress factors such as eutrophication, hydrological changes, and water shortage in plateau lakes. This model overcomes the fundamental deficiency of traditional methods that neglect the coupling and reinforcement effects of stress factors, establishes a causal relationship between "water quantity guarantee" and "ecological restoration" at the mechanistic level, and solves the technical bottleneck of existing methods where water levels meet standards but the ecosystem continues to degrade.

[0019] By integrating core ecological indicators from multiple dimensions, such as phytoplankton diversity, submerged plant coverage, benthic animal richness, and comprehensive trophic status index, a comprehensive ecological health index is constructed using the entropy weight method. This overcomes the one-sidedness of relying on single indicators such as chlorophyll a, which are difficult to reflect the degradation of ecosystem structure under multiple stresses, and provides a more scientific and comprehensive ecological benchmark for determining water level thresholds.

[0020] The Pettit mutation test is used to identify the critical value of irreversible degradation of the ecosystem. Combined with model inversion, the minimum ecological water level red line is determined to fundamentally prevent the collapse of the ecosystem. Through multi-scenario water level simulation and water level-ecological health response curves, and combined with the water balance method and water exchange cycle method for verification, a suitable ecological water level range is determined, forming a stratified threshold that takes into account both bottom line protection and active restoration, providing precise scheduling targets for lake ecological restoration.

[0021] The gradient boosting decision tree algorithm can accurately characterize the nonlinear relationship and complex interaction effects between key stress factors and ecological responses. The threshold determination accuracy is better than that of traditional empirical statistical methods. The model has an annual dynamic update mechanism and can continuously self-calibrate based on newly added monitoring data. This enables the ecological water level threshold to dynamically adapt to new conditions such as changes in the underlying surface of the watershed and changes in the water transfer situation from other watersheds, supporting the adaptive management of ecological scheduling of plateau lakes. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating the method for determining the ecological water level threshold of plateau lakes based on the identification of multiple stress factors, according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of the plateau lake ecological water level threshold determination system based on multi-stress factor identification, according to an embodiment of the present invention. Detailed Implementation

[0023] The following is a detailed description of the embodiments, in conjunction with the accompanying drawings.

[0024] Reference Figure 1 A method for determining the ecological water level threshold of plateau lakes based on the identification of multiple stress factors includes the following steps: Step 1: Collect multi-source long-term time series data of the plateau lake basin, and perform standardized preprocessing on the multi-source long-term time series data, including missing value imputation, outlier removal, and time scale unification, to construct a standardized dataset; wherein, the multi-source long-term time series data includes hydrological and meteorological data, water quality data, aquatic ecological data, basin socio-economic data, and water conservancy project scheduling data.

[0025] In some embodiments, long-term time-series data can be collected from multiple sources, including watershed hydrological stations, meteorological stations, automatic water quality monitoring stations and manual sampling sections, remote sensing inversion platforms, and water conservancy project scheduling and operation logs. Hydrological and meteorological data include lake level, inflow and outflow, precipitation, evaporation, temperature, and wind speed; water quality data includes total nitrogen, total phosphorus, permanganate index, chlorophyll a concentration, and transparency; aquatic ecological data includes phytoplankton density and diversity, submerged plant distribution and coverage, and benthic animal species and abundance; watershed socioeconomic data includes urban and industrial wastewater discharge, agricultural non-point source nitrogen and phosphorus emission intensity, and the proportion of impervious area in the lakeshore; and water conservancy project scheduling data includes water diversion from other watersheds, dam discharge processes, and ecological water replenishment records.

[0026] The collected raw data were aligned using a unified time series index. Based on the continuous length and missing pattern of each indicator's data, linear interpolation or cubic spline interpolation was used for short-term missing data, while multiple interpolation models were established based on auxiliary variables highly correlated with the target variable (such as using the relationship between rainfall and water level to interpolate missing water levels) to generate a complete series. Outlier removal adopted a dual discrimination process combining physical thresholds and statistical tests. Values ​​that exceeded the instrument's range or did not conform to physical logic were directly marked as outliers. The median absolute deviation method and box plot method were used to identify statistical outliers. The identified outliers were replaced with the climatological average value or the effective values ​​before and after that period.

[0027] Multi-source data undergoes unified time-scale processing. Monitoring data recorded at different frequencies such as hourly, daily, and ten-day periods are aggregated to a monthly scale through methods such as summation, averaging, or taking the instantaneous value at the end of the month, ensuring that all variables have the same time step and start and end times. Unit conversions are performed for each variable (e.g., water level is standardized to "m", flow rate to "m³ / s", and water quality index concentration to "mg / L"), and systematic biases caused by changes in monitoring methods are corrected. Finally, a continuous, regular, and time-frequency standardized dataset is formed, providing a consistent data foundation for subsequent stress factor screening and model construction.

[0028] Step 2: Select phytoplankton diversity index, submerged plant coverage, benthic animal richness index, and comprehensive trophic status index as core ecological indicators. Use the entropy weight method to determine the weight of each core ecological indicator, and calculate the comprehensive ecological health index of the lake by weighting, which is used to characterize the overall state of the lake ecosystem.

[0029] In some embodiments, the phytoplankton diversity index can be calculated using the Shannon-Wiener diversity index, which reflects both species richness and evenness of the community. A higher value indicates a more stable community structure and a healthier primary productivity state, as shown in the following formula:

[0030] in, This is the phytoplankton diversity index; This represents the total number of phytoplankton species identified in the sample. For the first Individual density (or cell density) of a species. This represents the total individual density of all phytoplankton in the same sample.

[0031] Submerged plant cover can be determined by remote sensing interpretation or field quadrat surveys, as a percentage of the total lake area covered by submerged vegetation (up to 100%), as shown in the following formula:

[0032] in, Submerged plant coverage; This represents the actual distribution area of ​​submerged plants. This represents the total area of ​​the lake. Submerged plants are a key group for maintaining clear water stability, and their coverage is a core indicator for measuring the degree of restoration of grass-based ecosystems. Benthic animal richness index can be used to measure the species abundance of a benthic animal community. This index is sensitive to habitat heterogeneity and water quality changes, and can indicate the long-term quality of the substrate environment, as shown in the following formula:

[0033] in, The index of benthic animal richness; This represents the total number of taxa of benthic macroinvertebrates in the sample. This represents the total number of benthic animals in the same sample.

[0034] A modified Carlson composite nutrient status index method can be used, with chlorophyll a (Chl-a) as the benchmark, integrating total phosphorus (TP), total nitrogen (TN), transparency (SD), and permanganate index (COD). Mn The calculation is based on a weighted average of five parameters. The single-factor nutrient status index for each parameter is calculated as follows:

[0035] in, For the first Single-factor trophic status index corresponding to each water quality parameter; and All are the first The empirical constants (coefficients) of the water quality parameters are determined through statistical regression of a large amount of lake survey data; For the first The measured concentration (or value) of each water quality parameter needs to be substituted into units that match the coefficient; then the comprehensive trophic state index is shown in the following formula:

[0036] in, The comprehensive nutritional status index; For the first Weighting coefficients for each water quality parameter. Empirical constants for each parameter. , and weight Refer to the Chinese standards and specifications for evaluating eutrophication of lakes (e.g., chlorophyll a: Total phosphorus: Total nitrogen: ;transparency: Permanganate index: ). The higher the value, the more severe the eutrophication and the worse the ecological health.

[0037] The entropy weight method determines the weights based on the dispersion of each indicator's data. The greater the variation in the indicator data, the more information it carries, and the higher its weight should be. This avoids the arbitrariness of subjective weight assignment. The specific calculation process is as follows: It has A matrix consisting of 1 time series sample (monthly or yearly) and 4 core ecological indicators. ,in The sample number. Corresponding in sequence , , , To eliminate the influence of dimensions and orders of magnitude and to unify the direction, the range transformation method is used to process positive and negative indices separately. , , It is a positive indicator (the larger the better). It is a negative indicator (the smaller the better).

[0038] The standardized positive index is shown in the following formula:

[0039] Negative Liability Index (TLI) Standardization:

[0040] in, For the first The sample, the first The values ​​of the core ecological indicators after standardization are all in the range of [0,1], and the direction is consistent: the larger the value, the healthier the ecosystem. For the first The sample, the first The original observed values ​​of the core ecological indicators; Indicates the first Key ecological indicators, in all samples arrive In this case, the original value of this core ecological indicator is taken. The minimum value; Indicates the first Key ecological indicators, in all samples arrive In this case, the original value of this core ecological indicator is taken. The maximum value.

[0041] Calculate the first The first item under the indicator The proportion of each sample value As shown in the following formula:

[0042] Calculate the first Entropy value of the item index As shown in the following formula:

[0043] like Then define Entropy satisfy .

[0044] Calculate the first Coefficient of difference of the items (Information entropy redundancy), as shown in the following formula:

[0045] The larger the difference coefficient, the more effective information the indicator carries.

[0046] Determine the first Weight of each indicator As shown in the following formula:

[0047] Finally, the entropy weight method weight vector is obtained. , , , and The weights corresponding to the phytoplankton diversity index, submerged plant coverage, benthic animal richness index, and comprehensive trophic status index are respectively satisfied. .

[0048] The standardized core ecological indicator values ​​are then weighted and summed to obtain the first value. Comprehensive Ecological Health Index of Lakes in Each Sample As shown in the following formula:

[0049] The exponential sequence This fully reflects the dynamic changes in the ecosystem state of the plateau lake under long-term multi-stress conditions, and will be used as the output variable of the subsequent multi-stress-ecological response model for mutation testing and water level threshold inversion.

[0050] Step 3: Extract candidate stress factors from the standardized dataset, use Pearson correlation analysis combined with principal component analysis to screen out key stress factors that are related to the state of the lake ecosystem, and perform dimensionless processing on each key stress factor.

[0051] In some embodiments, the extraction of candidate stress factors from a standardized dataset, the screening of key stress factors related to the lake ecosystem state using Pearson correlation analysis combined with principal component analysis, and the dimensionless processing of each key stress factor specifically include: Calculate the Pearson correlation coefficient between each candidate stress factor and the lake's comprehensive ecological health index, and screen candidate stress factors with an absolute correlation coefficient greater than a first threshold as initial stress factors. Perform principal component analysis on the initial stress factors, extract principal components with eigenvalues ​​greater than a second threshold, and select the initial stress factors with the largest absolute loadings among the principal components to be included in the key stress factor set. Use deviation standardization to perform dimensionless processing on the key stress factors. The first threshold can be set to 0.3–0.5, and the second threshold can be set to 1.0.

[0052] In some embodiments, the key stress factors include eutrophication stress factors, hydrological situation stress factors, water scarcity stress factors, endogenous pollution stress factors, and human activity stress factors; wherein: From the water quality data and the aquatic ecosystem data, total nitrogen concentration, total phosphorus concentration, chlorophyll a concentration, transparency, and permanganate index are selected to calculate the comprehensive trophic status index as an eutrophication stress factor, referring to the calculation of the comprehensive trophic status index in step two.

[0053] Based on the aforementioned hydrological and meteorological data and water conservancy project scheduling data, the annual water level variation coefficient, the outflow variation coefficient caused by water diversion from other basins, and the deviation of the water level from the multi-year average natural water level are calculated. The hydrological stress factor is obtained by weighted summation, as shown in the following formula:

[0054] in, This is a hydrological stress factor; the larger the value, the more drastic the change in hydrological rhythm. This represents the average daily water level over a statistical period (such as a year). This represents the standard deviation of daily water levels within the statistical period. This is the annual water level variation coefficient, reflecting the degree of water level fluctuation. This represents the average daily lake flow during the statistical period. The standard deviation of the daily lake flow during the statistical period is given, and its variation is mainly caused by the water discharge scheduling of the water diversion project. The coefficient of variation of outflow from the lake caused by water diversion from other basins; The average water level for the statistical year; It is the multi-year average natural water level, obtained through long-term historical data or natural runoff reconstruction calculations. This refers to the range of natural water level changes over many years. This represents the standardized deviation of the water level from its natural state. , and The weight coefficients of the corresponding terms satisfy the following conditions: It can be objectively determined based on the degree of variation of the interannual series of each sub-indicator using the entropy weight method.

[0055] Based on the aforementioned hydrological and meteorological data and the aforementioned water conservancy project scheduling data, the ratio of the water replenishment required to maintain the target ecological water level of the lake to the natural runoff of the watershed is calculated using the water balance equation, and used as a water shortage stress factor, as shown in the following formula:

[0056] in, As a stressor of water scarcity, the higher the ratio, the more scarce the natural water resources are and the more severe the encroachment on the ecosystem. The total natural runoff of the basin within the statistical period (e.g., year); The minimum ecological water replenishment required to maintain the target ecological water level is calculated using the water balance equation, as shown below:

[0057] in, This represents the total evaporation from the lake surface during the statistical period. This represents the total precipitation over the lake during the statistical period. The area of ​​the lake; This refers to the total amount of natural runoff flowing into the lake during the statistical period. The total outflow of water from the lake within the statistical period after meeting the basic ecological and water needs of the downstream areas; The increase in lake storage required to maintain the target water level; if the target water level has already been reached that year, this item is 0. When natural water supply meets demand, stress factors... Take 0.

[0058] Using lake sediment monitoring data, the total nitrogen and total phosphorus release rates from the sediments were calculated, normalized by the average lake depth, and then superimposed to obtain the endogenous pollution stress factor, as shown in the following formula:

[0059] in, These are endogenous pollution stress factors, and are dimensionless after standardization. The average water depth of the lake during the statistical period; The average release rate of total nitrogen in sediment was determined by in-situ sediment culture experiments or static release experiments using column core samples. The average release rate of total phosphorus in the sediment; This is the ecotoxicity equivalent conversion factor between total nitrogen and total phosphorus, which unifies the environmental impacts of both. For simplification, we can take [value missing]. (Based on the approximate trade-off between the nitrogen-to-phosphorus ratio of algae growth (N:P≈7:1) and their respective contributions to eutrophication).

[0060] From the aforementioned watershed socioeconomic data, we extract the total urban domestic sewage discharge, total industrial wastewater discharge, agricultural non-point source nitrogen and phosphorus emission intensity, and the proportion of impermeable area in the lakeside zone. Principal component analysis is then used to extract the first principal component as a human activity stress factor, as shown in the following formula:

[0061] in, As a stressor of human activity; , , and The values ​​are, in order, the total discharge of urban domestic sewage, the total discharge of industrial wastewater, the intensity of nitrogen and phosphorus emissions from agricultural non-point sources, and the proportion of impermeable area in the lakeside zone, after Z-score standardization (mean 0, standard deviation 1) to eliminate the influence of dimensions. , , and These are the loading coefficients of the first principal component on each standardized variable. , , and The value of is determined by the actual long-term socio-economic monitoring data of the target plateau lake basin. The value is based on the mathematical criterion in principal component analysis (PCA) that maximizes the variation information of the original four variables when extracting the first principal component from the standardized variables. This is an existing technology and will not be elaborated here.

[0062] Step 4: Using the key stress factors as input variables and the lake's comprehensive ecological health index as output variables, construct a multi-stress-ecological response coupling model using the gradient boosting decision tree algorithm, and optimize the model parameters through cross-validation; based on the trained multi-stress-ecological response coupling model, quantify the independent contribution of each key stress factor to the lake's ecological health and the intensity of the interaction between each key stress factor, for the quantitative identification of the synergistic influence mechanism of multiple stress factors.

[0063] In some embodiments, the specific process of constructing the multi-stress-ecological response coupling model is as follows: The key stress factors after dimensionless processing are used as input variables, and the comprehensive ecological health index of the lake is used as the output variable to construct a sample dataset. A gradient boosting decision tree algorithm is employed, with the number of decision trees, learning rate, and maximum depth parameters set within a given range. K-fold cross-validation is performed on the sample dataset, and the optimal hyperparameter combination is searched using the minimum root mean square error as the criterion, resulting in a trained multi-stress-ecological response coupled model. Furthermore, the range of values ​​for the number of decision trees can be... The learning rate can take any value within the range of... The maximum tree depth can be within the range of values. In K-fold cross-validation, the value of K is 10.

[0064] Based on the trained multi-stress-ecological response coupled model, the average contribution of each key stress factor to the reduction of the loss function in the model prediction is calculated one by one, and the independent contribution value of each key stress factor to the ecological health of the lake is obtained. Using the partial dependency function, the ratio of the sum of squares of the differences between the joint partial dependency and the individual marginal partial dependency of any two key stress factors to the total variance is calculated to obtain the interaction strength between the two factors. By traversing all pairwise factor combinations, an interaction strength matrix is ​​formed to complete the quantitative identification of the synergistic influence mechanism of multiple stress factors.

[0065] Step 5: Use the Pettitt mutation test to identify the critical value of irreversible mutation in the lake's comprehensive ecological health index. Obtain the corresponding water level range for the critical value through inversion using a multi-stress-ecological response coupling model. Combined with preset requirements, determine the minimum ecological water level red line. Set different water level scenarios with gradient intervals, input the multi-stress-ecological response coupling model to simulate the lake's comprehensive ecological health index under each water level scenario, plot the water level-comprehensive ecological health index response curve, identify the stable range in the curve where the lake's comprehensive ecological health index reaches the preset ecological restoration target value, and verify using the water balance method and water exchange cycle method to determine the suitable ecological water level range. Use the minimum ecological water level red line and the suitable ecological water level range as the ecological water level threshold.

[0066] In some embodiments, determining the minimum ecological water level red line specifically includes: The historical lake comprehensive ecological health index series is input into the Pettt mutation test, and the corresponding statistic and its significance level are calculated. When the significance level is less than the third threshold, it is determined that there is an irreversible mutation point in the series. The lake comprehensive ecological health index corresponding to the irreversible mutation point is used as the critical value. The preferred value of the third threshold is 0.05.

[0067] Each of the key stress factors is fixed to its multi-year average value. The multi-stress-ecological response coupled model is trained with water level as the only variable. The predicted value of the lake comprehensive ecological health index corresponding to different water level values ​​is generated. The water level value that causes the predicted value of the lake comprehensive ecological health index to drop to the critical value is determined by linear interpolation. The water level determined by linear interpolation is compared with the minimum control water level set by the lake protection regulations, and the higher of the two water levels is taken as the minimum ecological water level red line.

[0068] In some embodiments, determining the suitable ecological water level range specifically includes: Using the minimum ecological water level red line as the lower limit and the highest historical operating water level of the lake as the upper limit, a gradient water level scenario set is generated at preset equal intervals; Each water level value in the gradient water level scenario set is taken as a fixed value, while the other key stress factors are kept at the baseline year level. The results are input into the trained multi-stress-ecological response coupling model to obtain the comprehensive ecological health index of the lake corresponding to each water level scenario. A response curve is plotted with water level as the x-axis and the lake's comprehensive ecological health index as the y-axis. After smoothing the response curve by moving average, the local slope at each point is calculated. Water level intervals where the absolute value of the local slope is less than the fourth threshold and the lake's comprehensive ecological health index is continuously greater than the fifth threshold are extracted as candidate suitable water level intervals. Since the slope magnitudes of response curves vary among different lakes (e.g., some curves have slopes between 0.1 and 0.5, while others have slopes between 0.01 and 0.05), the fourth threshold can be set as a low quantile of the slope sequence or an empirical value close to 0 by calculating the local slopes at all points of the entire response curve. In practice, the fifth threshold can be based on the lake's comprehensive ecological health index value of historical health status or directly set as a target value on a standardized scale (e.g., 0.6, 0.7, or 0.8), representing that the ecosystem is at a "good" or "healthy" level.

[0069] For each candidate suitable water level range, the total ecological water replenishment required to maintain the water level of that range throughout the year is calculated using the water balance method, and the corresponding lake water exchange cycle is calculated using the water exchange cycle method. Ranges with ecological water replenishment less than the maximum adjustable water volume of the watershed water diversion project and water exchange cycle less than 1.5 years are selected, and the range with the highest average comprehensive ecological health index of the lake is selected as the suitable ecological water level range.

[0070] Step 6: Verify the rationality of the ecological water level threshold using historical independent monitoring data, and compare the differences in lake ecosystem status between the ecological water level threshold range and outside the range; establish an annual dynamic update mechanism, and recalibrate the multi-stress-ecological response coupling model and update the ecological water level threshold based on the newly added monitoring data of the year.

[0071] In some embodiments, the establishment of an annual dynamic update mechanism, which recalibrates the multi-stress-ecological response coupling model and updates the ecological water level threshold based on newly added monitoring data for the year, specifically involves: At the end of each year, newly added hydrological, water quality, water ecology, and water conservancy project scheduling monitoring data are acquired and expanded into the standardized dataset after standardized preprocessing. Based on the existing multi-stress-ecological response coupled model, the model is incrementally trained using the expanded dataset, the tree structure and weights of the gradient boosting decision tree are updated, and the model is recalibrated. A subset of data is extracted from the standardized dataset using a preset sliding time window. The entire process of mutation detection of the lake's comprehensive ecological health index, determination of the minimum ecological water level red line, and determination of the suitable ecological water level range is re-executed for this subset of data to generate the updated ecological water level threshold for the year. The convergence criterion is set as follows: if the variation of the minimum ecological water level red line is less than 0.05 meters for three consecutive years, and the variation of the upper and lower limits of the suitable ecological water level range is less than 0.1 meters, then the current ecological water level threshold is locked and the annual update is stopped; otherwise, the update for the next year will continue.

[0072] Based on the same inventive concept, and corresponding to any of the above embodiments, refer to... Figure 2 This invention provides a system for determining the ecological water level threshold of plateau lakes based on the identification of multiple stress factors, used to implement the aforementioned method for determining the ecological water level threshold of plateau lakes based on the identification of multiple stress factors, comprising: The data acquisition module is used to collect hydrological and meteorological data, water quality data, aquatic ecological data, watershed socio-economic data, and water conservancy project scheduling data of plateau lake basins. It also performs missing value imputation, outlier removal, and time scale unification processing to output standardized datasets. The health index generation module, connected to the data acquisition module, is used to extract phytoplankton diversity index, submerged plant coverage, benthic animal richness index and comprehensive trophic status index from the standardized dataset, determine the weights using the entropy weight method and calculate the weighted average to generate the lake's comprehensive ecological health index. The stress factor screening module is connected to the data acquisition module and the health index generation module, respectively. It is used to extract candidate stress factors from the standardized dataset, screen out key stress factors that are significantly related to the lake's comprehensive ecological health index through Pearson correlation analysis and principal component analysis, and perform dimensionless processing on each key stress factor. The model building and collaborative quantification module is connected to the stress factor screening module and the health index generation module, respectively. It is used to construct a multi-stress-ecological response coupled model by using the dimensionless key stress factors as input and the lake comprehensive ecological health index as output, and using gradient boosting decision tree. Based on the model, it calculates the independent contribution value of each key stress factor and the interaction strength between each pair of factors to form an interaction strength matrix, thereby completing the quantitative identification of the collaborative influence mechanism of multiple stress factors. The water level threshold calculation module, connected to the model construction and collaborative quantification module, is used to identify the irreversible mutation critical value of the lake's comprehensive ecological health index using the Pettitt mutation test method, determine the minimum ecological water level red line through inversion of the multi-stress-ecological response coupling model, set a gradient water level scenario set and drive the multi-stress-ecological response coupling model to generate response curves, and determine the appropriate ecological water level range by combining the water balance method and the water exchange cycle method for verification. The verification and dynamic update module is connected to the water level threshold calculation module, the data acquisition module, and the model construction and collaborative quantization module, respectively. It is used to verify the rationality of the ecological water level threshold using historical independent monitoring data, and to expand the standardized dataset and incrementally train the multi-stress-ecological response coupled model based on the annually added monitoring data. Before meeting the convergence judgment criteria, it updates the minimum ecological water level red line and suitable ecological water level range annually.

[0073] Based on the same inventive concept, corresponding to any of the above embodiments, the present invention provides an electronic device, including a memory and a processor. The memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute the method for determining the ecological water level threshold of plateau lakes based on the identification of multiple stress factors in the embodiment.

[0074] Alternatively, the aforementioned electronic device may be a server.

[0075] In addition, this embodiment also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the method for determining the ecological water level threshold of plateau lakes based on the identification of multiple stress factors in this embodiment.

[0076] It is understood that the processor in the embodiments of the present invention may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. The general-purpose processor may be a microprocessor or any conventional processor.

[0077] The method steps in the embodiments of the present invention can be implemented in hardware or by a processor executing software instructions. The software instructions can consist of corresponding software modules, which can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, portable hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can reside in an ASIC.

[0078] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of the present invention is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a storage medium or transmitted through a storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive (SSD)).

Claims

1. A method for determining the ecological water level threshold of plateau lakes based on the identification of multiple stress factors, characterized in that, Includes the following steps: Multi-source long-term time series data of plateau lake basins were collected, and the multi-source long-term time series data were subjected to standardized preprocessing such as missing value imputation, outlier removal and time scale unification to construct a standardized dataset; wherein, the multi-source long-term time series data includes hydrological and meteorological data, water quality data, water ecology data, basin socio-economic data and water conservancy project scheduling data; Phytoplankton diversity index, submerged plant coverage, benthic animal richness index, and comprehensive trophic status index were selected as core ecological indicators. The weight of each core ecological indicator was determined by the entropy weight method, and the comprehensive ecological health index of the lake was obtained by weighted calculation, which is used to characterize the overall state of the lake ecosystem. Candidate stress factors were extracted from the standardized dataset. Pearson correlation analysis combined with principal component analysis was used to screen out key stress factors that are related to the state of the lake ecosystem. Dimensionless processing was then performed on each key stress factor. Using the key stress factors as input variables and the comprehensive ecological health index of the lake as output variables, a multi-stress-ecological response coupling model is constructed using the gradient boosting decision tree algorithm, and the model parameters are optimized through cross-validation. Based on the trained multi-stress-ecological response coupling model, the independent contribution of each key stress factor to the ecological health of the lake and the interaction strength among each key stress factor are quantified for the quantitative identification of the synergistic influence mechanism of multiple stress factors. The Pettt mutation test was used to identify the critical value for irreversible mutations in the comprehensive ecological health index of lakes. The corresponding water level range for the critical value was obtained through inversion using a multi-stress-ecological response coupling model. Combined with preset requirements, the minimum ecological water level red line was determined. Different water level scenarios with gradient intervals were set, and the comprehensive ecological health index of lakes under each water level scenario was simulated using the multi-stress-ecological response coupling model. Water level-comprehensive ecological health index response curves were plotted, and the stable range in which the comprehensive ecological health index of lakes reaches the preset ecological restoration target value was identified. The appropriate ecological water level range was determined by combining the water balance method and the water exchange cycle method. The minimum ecological water level red line and the appropriate ecological water level range were used as ecological water level thresholds. The rationality of the ecological water level threshold was verified using historical independent monitoring data, and the differences in the state of the lake ecosystem within and outside the ecological water level threshold range were compared. An annual dynamic update mechanism was established to recalibrate the multi-stress-ecological response coupling model and update the ecological water level threshold based on the newly added monitoring data of the year.

2. The method for determining the ecological water level threshold of plateau lakes based on multi-stress factor identification as described in claim 1, characterized in that, The process involves extracting candidate stress factors from a standardized dataset, using Pearson correlation analysis combined with principal component analysis to screen for key stress factors related to the lake ecosystem state, and then performing dimensionless processing on each key stress factor. Specifically, this includes: Calculate the Pearson correlation coefficient between each candidate stress factor and the lake's comprehensive ecological health index, and screen out candidate stress factors whose absolute value of the correlation coefficient is greater than the first threshold as initial stress factors; Principal component analysis was performed on the initial screening stress factors to extract principal components with eigenvalues ​​greater than the second threshold. The initial screening stress factors with the largest absolute loading value among the principal components were selected and incorporated into the key stress factor set. The key stress factors were dimensionless by using the deviation standardization method.

3. The method for determining the ecological water level threshold of plateau lakes based on multi-stress factor identification as described in claim 1, characterized in that, The key stress factors include eutrophication stress factors, hydrological situation stress factors, water scarcity stress factors, endogenous pollution stress factors, and human activity stress factors; among which: From the water quality data and the aquatic ecosystem data, total nitrogen concentration, total phosphorus concentration, chlorophyll a concentration, transparency and permanganate index were selected to calculate the comprehensive nutrient status index as the eutrophication stress factor. Based on the hydrological and meteorological data and the water conservancy project scheduling data, the annual water level variation coefficient of the lake, the outflow variation coefficient caused by water diversion from other basins, and the deviation of the water level from the multi-year average natural water level are calculated, and the hydrological situation stress factor is obtained by weighted summation. Based on the hydrological and meteorological data and the water conservancy project scheduling data, the ratio of the water replenishment required to maintain the target ecological water level of the lake to the natural runoff of the watershed is calculated using the water balance equation, and is used as a water shortage stress factor. Using lake sediment monitoring data, the total nitrogen release rate and total phosphorus release rate of the sediment were calculated, normalized and superimposed with the average lake depth to obtain the endogenous pollution stress factor. The total urban domestic sewage discharge, total industrial wastewater discharge, agricultural non-point source nitrogen and phosphorus emission intensity, and the proportion of impermeable area in the lakeside zone were extracted from the socio-economic data of the basin. The first principal component was extracted as a human activity stress factor through principal component analysis.

4. The method for determining the ecological water level threshold of plateau lakes based on multi-stress factor identification as described in claim 1, characterized in that, The specific process for constructing the multi-stress-ecological response coupled model is as follows: The key stress factors after dimensionless processing are used as input variables, and the comprehensive ecological health index of the lake is used as the output variable to construct a sample dataset. The gradient boosting decision tree algorithm is adopted, and the number of decision trees, learning rate and maximum depth parameter range are set. K-fold cross-validation is performed on the sample dataset. The optimal hyperparameter combination is searched with the minimum root mean square error as the criterion to obtain the trained multi-stress-ecological response coupled model. Based on the trained multi-stress-ecological response coupled model, the average contribution of each key stress factor to the reduction of the loss function in the model prediction is calculated one by one, and the independent contribution value of each key stress factor to the ecological health of the lake is obtained. Using the partial dependency function, the ratio of the sum of squares of the differences between the joint partial dependency and the individual marginal partial dependency of any two key stress factors to the total variance is calculated to obtain the strength of their interaction. By traversing all pairwise factor combinations, an interaction strength matrix is ​​formed, enabling the quantitative identification of the synergistic influence mechanism of multiple stress factors.

5. The method for determining the ecological water level threshold of plateau lakes based on multi-stress factor identification as described in claim 1, characterized in that, The determination of the minimum ecological water level red line specifically includes: The historical lake comprehensive ecological health index series is input into the Pettt mutation test, and the corresponding statistics and their significance levels are calculated. When the significance level is less than the third threshold, it is determined that there is an irreversible mutation point in the series, and the lake comprehensive ecological health index corresponding to the irreversible mutation point is used as the critical value. Each of the key stress factors is fixed to its multi-year average value. The multi-stress-ecological response coupled model is trained with water level as the only variable. The predicted value of the lake comprehensive ecological health index corresponding to different water level values ​​is generated. The water level value that causes the predicted value of the lake comprehensive ecological health index to drop to the critical value is determined by linear interpolation. The water level determined by linear interpolation is compared with the minimum control water level set by the lake protection regulations, and the higher of the two water levels is taken as the minimum ecological water level red line.

6. The method for determining the ecological water level threshold of plateau lakes based on multi-stress factor identification as described in claim 1, characterized in that, The determination of the suitable ecological water level range specifically includes: Using the minimum ecological water level red line as the lower limit and the highest historical operating water level of the lake as the upper limit, a gradient water level scenario set is generated at preset equal intervals; Each water level value in the gradient water level scenario set is taken as a fixed value, while the other key stress factors are kept at the baseline year level. The results are input into the trained multi-stress-ecological response coupling model to obtain the comprehensive ecological health index of the lake corresponding to each water level scenario. A response curve was plotted with water level as the abscissa and lake comprehensive ecological health index as the ordinate. After smoothing the response curve by moving average, the local slope of each point was calculated. Water level intervals with local slope absolute values ​​less than the fourth threshold and lake comprehensive ecological health index continuously greater than the fifth threshold were extracted as candidate suitable water level intervals. For each candidate suitable water level range, the total ecological water replenishment required to maintain the water level of that range throughout the year is calculated using the water balance method, and the corresponding lake water exchange cycle is calculated using the water exchange cycle method. Ranges with ecological water replenishment less than the maximum adjustable water volume of the watershed water transfer project and water exchange cycle less than 1.5 years are selected, and the range with the highest average comprehensive ecological health index of the lake is selected as the suitable ecological water level range.

7. The method for determining the ecological water level threshold of plateau lakes based on multi-stress factor identification as described in claim 1, characterized in that, The establishment of an annual dynamic update mechanism, which recalibrates the multi-stress-ecological response coupling model and updates the ecological water level threshold based on newly added monitoring data for the year, specifically involves: At the end of each year, newly added hydrological, water quality, water ecology, and water conservancy project scheduling monitoring data are acquired and expanded into the standardized dataset after standardized preprocessing. Based on the existing multi-stress-ecological response coupled model, the model is incrementally trained using the expanded dataset, the tree structure and weights of the gradient boosting decision tree are updated, and the model is recalibrated. A subset of data is extracted from the standardized dataset using a preset sliding time window. The entire process of mutation detection of the lake's comprehensive ecological health index, determination of the minimum ecological water level red line, and determination of the suitable ecological water level range is re-executed for this subset of data to generate the updated ecological water level threshold for the year. The convergence criterion is set as follows: if the variation of the minimum ecological water level red line is less than 0.05 meters for three consecutive years, and the variation of the upper and lower limits of the suitable ecological water level range is less than 0.1 meters, then the current ecological water level threshold is locked and the annual update is stopped; otherwise, the update for the next year will continue.

8. A system for determining the ecological water level threshold of plateau lakes based on multi-stress factor identification, used to implement the method for determining the ecological water level threshold of plateau lakes based on multi-stress factor identification as described in any one of claims 1-7, characterized in that, include: The data acquisition module is used to collect hydrological and meteorological data, water quality data, aquatic ecological data, watershed socio-economic data, and water conservancy project scheduling data of plateau lake basins. It also performs missing value imputation, outlier removal, and time scale unification processing to output standardized datasets. The health index generation module, connected to the data acquisition module, is used to extract phytoplankton diversity index, submerged plant coverage, benthic animal richness index and comprehensive trophic status index from the standardized dataset, determine the weights using the entropy weight method and calculate the weighted average to generate the lake's comprehensive ecological health index. The stress factor screening module is connected to the data acquisition module and the health index generation module, respectively. It is used to extract candidate stress factors from the standardized dataset, screen out key stress factors that are significantly related to the lake's comprehensive ecological health index through Pearson correlation analysis and principal component analysis, and perform dimensionless processing on each key stress factor. The model building and collaborative quantification module is connected to the stress factor screening module and the health index generation module, respectively. It is used to construct a multi-stress-ecological response coupled model by using the dimensionless key stress factors as input and the lake comprehensive ecological health index as output, and using gradient boosting decision tree. Based on the model, it calculates the independent contribution value of each key stress factor and the interaction strength between each pair of factors to form an interaction strength matrix, thereby completing the quantitative identification of the collaborative influence mechanism of multiple stress factors. The water level threshold calculation module, connected to the model construction and collaborative quantification module, is used to identify the irreversible mutation critical value of the lake's comprehensive ecological health index using the Pettitt mutation test method, determine the minimum ecological water level red line through inversion of the multi-stress-ecological response coupling model, set a gradient water level scenario set and drive the multi-stress-ecological response coupling model to generate response curves, and determine the appropriate ecological water level range by combining the water balance method and the water exchange cycle method for verification. The verification and dynamic update module is connected to the water level threshold calculation module, the data acquisition module, and the model construction and collaborative quantization module, respectively. It is used to verify the rationality of the ecological water level threshold using historical independent monitoring data, and to expand the standardized dataset and incrementally train the multi-stress-ecological response coupled model based on the annually added monitoring data. Before meeting the convergence judgment criteria, it updates the minimum ecological water level red line and suitable ecological water level range annually.

9. An electronic device, characterized in that, The device includes a memory and a processor, wherein the memory stores a computer program and the processor runs the computer program to enable the electronic device to perform the method for determining the ecological water level threshold of plateau lakes based on the identification of multiple stress factors as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for determining the ecological water level threshold of plateau lakes based on the identification of multiple stress factors as described in any one of claims 1-7.

Images