Method, recording medium and system for adjusting ecological flow of river channel

Abstract

The invention belongs to the technical field of hydrology and water resource management, and discloses a method for adjusting the ecological flow of a riverway, which comprises the following steps: collecting the time sequence data of the exposed area of a sandbar in the riverway and the section runoff volume of a fixed measuring point; determining two intermediate values by using a jump detection algorithm, and dividing the wind and sand activity intensity into three sections; function models are established respectively, and according to the adaptability of different models, a particle swarm algorithm or a Bayesian algorithm is selected to optimize model parameters; and continuously sampling the time sequence data by adopting a self-adaptive dynamic sliding window algorithm, and inputting the time sequence data into a model for calculation to obtain a reference value of the minimum ecological flow. According to the method, technologies in the fields of hydrology, meteorology, landform and artificial intelligence are fused, a linkage model of'sand bare area-runoff-sand probability 'is established, the problem of cross-space-time decoupling is solved, and the method is suitable for the collaborative optimization control process of ecological flow and sand suppression. The invention also provides a non-transitory readable recording medium storing the program of the method and a system comprising the medium, and the program can be called through a processing circuit to execute the method.

Description

Technical Field

[0001] The present invention belongs to the technical field of hydrological water resources management, and discloses a method, a recording medium and a system for regulating the ecological flow of a river course. Background Art

[0002] Dust sources, strong winds and unstable atmospheric circulation are three key factors for generating dust. Among them, the wind condition is particularly crucial for sand and dust activities, and there are significant differences in wind speed, wind direction and probability of sand - raising wind in different regions. Understanding these wind condition characteristics and variation laws is extremely important for accurately grasping the sand and dust probability and formulating targeted sand and dust prevention and control measures. However, current monitoring and prediction models for sand and dust probability often can only be targeted at specific regions and seasons, lacking universality; even in the same region and at the same time period, the impacts brought by different sand and dust intensities are difficult to fit into a function curve, there are change - point detection problems, and it cannot meet the growing demand for sand and dust prevention and control.

[0003] Ecological flow of water is indispensable for maintaining the structure and function of water ecological systems such as rivers and lakes. It can not only ensure the stability of river channel morphology, prevent river drying - up, enhance the self - purification ability of rivers, recharge groundwater, but also maintain the water volume and water level required by aquatic organisms in the dry season, which is of great significance for maintaining the water ecological balance.

[0004] However, due to the limitations of natural endowment conditions, unreasonable development and utilization, and global climate change, etc., the contradictions among domestic, production and ecological water uses in some basin regions are prominent, and it is difficult to guarantee the ecological flow of rivers and lakes, which has triggered a series of severe problems such as river drying - up, lake shrinkage, damaged biodiversity, and decline of ecological service functions. For example, the attenuation of water resources in the Yellow River Basin and the increase in economic and social water use have led to the imbalance between water supply and demand, the reduction of river runoff, the weakening of water power, and it is difficult to guarantee the basic sediment - carrying water required for maintaining the functions of sediment - laden rivers. How to scientifically and reasonably determine the ecological flow target and effectively guarantee the ecological flow has become a key problem to be solved urgently in the current fields of water resources management and water ecological protection.

[0005] At present, the correlative research between sand and dust probability and ecological flow of water is relatively less, and the two are usually studied as independent fields. However, in fact, there is a close connection between them. Sand and dust activities will change the nature of the underlying surface, affect surface runoff and soil moisture evaporation, and then affect the distribution and circulation of water resources, indirectly acting on the ecological flow of water. On the contrary, the change of ecological flow of water will also affect the vegetation growth status, and vegetation, as an important influencing factor of sand and dust activities, its change will in turn affect the sand and dust probability. For example, in arid and semi - arid regions, the reduction of river water volume leads to the degradation of vegetation along the river banks, which intensifies land desertification, and sandbars without vegetation appear in the river channel, resulting in an increase in sand and dust probability.

[0006] Emergencies can also affect the probability of sandstorms and the water ecological flow. For example, building a dam or a water diversion project on a river course, or planting trees on the ground surface to prevent wind and sand. This makes the interaction between the probability of sandstorms and the water ecological flow not a simple functional relationship, and the factors involved are intricate. If directly fitting the relationship between the probability of sandstorms and the water ecological flow through artificial intelligence, a simple model will miss key features, resulting in underfitting and unable to truly reflect the corresponding relationship between the two. If the model is complex, it will cause overfitting and lack of generalization ability. Just like whether chemotherapy has a positive or negative effect on the treatment of cancer, artificial intelligence cannot fit the direct functional relationship between the two. Due to the lack of effective comprehensive analysis methods and technical means, it restricts the comprehensive understanding and effective protection of the overall situation of the regional ecological environment. Summary of the Invention

[0007] The inventive concept of this application lies in finding intermediate parameters that have direct connections with both the water ecological flow and the probability of sandstorms, establishing models in segments to reveal the indirect relationship between the water ecological flow and the probability of sandstorms, and ultimately improving the impact of sandstorm activities on humans based on the allocation of water resources. This requires overcoming many obstacles in the cross-domain docking process. The present invention provides a method for regulating the ecological flow of a river course, including the following steps:

[0008] S1. Collect time series data of the exposed area of sandbars in the river course, the runoff at fixed measuring points, and the probability of sandstorm activities;

[0009] S2. Establish a function model between the exposed area of the sandbar and the runoff at the fixed measuring point at the same moment, and use the particle swarm optimization algorithm to optimize the parameters of the function model;

[0010] S3. Establish a sandstorm activity intensity model, determine two intermediate values of the sandstorm activity intensity through the jump detection algorithm, and divide the sandstorm activity intensity into three intervals: mild, moderate, and severe by the two intermediate values; among them, the jump detection algorithm is synthesized based on the trimmed linear time exact algorithm and the change point detection algorithm, and is used for the non-linear structure identification of the time series of sandstorm activity intensity values;

[0011] S4. Establish a sandstorm response model that describes the relationship between the probability of sandstorm activities and the exposed area of the sandbar within the same time period for the three intervals respectively. For the continuous sampling of time series data involved in the function model, use the adaptive dynamic sliding window algorithm, and the window length is dynamically adjusted according to the sandstorm activity intensity, with a range of 2 - 4 months and an initial sliding step of 10 days;

[0012] S5. Determine the critical sandbar exposed area according to the set threshold of the probability of sand and wind activities in the sand and wind response model, substitute the critical exposed area into the function model in step S2, and calculate the corresponding runoff of the fixed measurement point section as the minimum ecological flow; adjust the sliding step according to the change range of the minimum ecological flow within the set time period;

[0013] S6. Manually adjust the runoff of the fixed measurement point section of the river channel to be above the minimum ecological flow, and update the model data for optimization in real time.

[0014] Preferably, the sand and wind activity intensity model is: set the sand and wind activity intensity SWAD = α·FD + β·BD + γ·DS + δ·TSD; where FD is the frequency of floating dust, BD is the frequency of blowing sand, DS is the frequency of sandstorms, TSD is the duration of dust storms, and α, β, γ, δ are seasonal weight coefficients, which are determined by the principal component analysis method in the same season; set the intermediate value separating the mild and moderate intervals as SWAD_L, and the intermediate value separating the moderate and severe intervals as SWAD_H; then the three sand and wind response models:

[0015] In the mild interval, the sand and wind response model corresponding to the state of SWAD≤SWAD_L:

[0016] P = k1·A + d1;

[0017] In the moderate interval, the sand and wind response model corresponding to the state of SWAD_L<SWAD≤SWAD_H:

[0018] P = k2·A² + k3·A + d2;

[0019] In the severe interval, the sand and wind response model corresponding to the state of SWAD>SWAD_H:

[0020] P = k4·e λA + d3;

[0021] Among them, λ is the exponential growth rate coefficient, representing the rate of sand and wind outbreaks, k1, k2, k3, k4 are proportionality coefficients, d1, d2, d3 are constant terms, which are optimized by the Bayesian algorithm, P is the probability of sand and wind activities, A is the exposed area of the sandbar, and belongs to the observed data.

[0022] Preferably, the function model between the exposed area of the sandbar in step S2 and the runoff of the fixed measurement point section at the same moment is:

[0023] When the exposed area of the sandbar , it is determined that the sandbar is completely submerged, and it is default that the ecological flow does not need to be adjusted;

[0024] Set the seasonal exposed sandbar area threshold , when When the fixed measurement point cross-sectional runoff

[0025] If A > A thres the fixed measurement point cross-sectional runoff

[0026] Wherein:

[0027] A sand : The total area of exposed sandbars, determined by the maximum historical exposed area extracted from remote sensing images;

[0028] ζ: Submergence threshold coefficient, default 0.02, optimization range 0.01 - 0.05;

[0029] Q base : Base ecological flow, taking the 90th percentile of the historical minimum monthly average flow;

[0030] κ: Flow compensation coefficient, determined by linear regression;

[0031] a is the maximum exposed area coefficient, representing the theoretical maximum exposed area in the zero-flow state;

[0032] b is the attenuation coefficient, representing the reduction rate of the exposed area when the flow increases;

[0033] c is the shape parameter, determining the non-linearity degree of the function curve;

[0034] μ is the average reference exposed area;

[0035] ω is the seasonal variation coefficient, characterizing the fluctuation amplitude of the exposed area relative to the reference area;

[0036] φ is the phase parameter, characterizing the time shift of the seasonal variation;

[0037] t is the observed time point, with the serial number sorted from 1 to 365 within a year as the t value;

[0038] The parameters a, b, c, μ, ω, φ are optimized by the particle swarm algorithm, and its hyperparameters are:

[0039] Goodness of fit R² ≥ 0.90; The root mean square error of the sandbar exposed area A is less than 10% of the average sandbar exposed area.

[0040] Preferably, the steps of collecting the time series data of the sandbar exposed area in the river channel include:

[0041] The spatial resolution of the selected remote sensing camera should be ≤ 30×30 m 2, the time coverage is at least 3 hydrological years including wet, normal, and dry years. When the cloud cover rate of the image ≤ 5%, collect remote sensing images of the river channel area; use ENVI software to perform radiometric calibration and atmospheric correction on the remote sensing images, combine the normalized difference vegetation index NDVI with the supervised classification of the depth residual convolutional neural network to extract the vector boundaries of the exposed sandbars in each period, and calculate the exposed area A of the sandbars; NDVI = (NIR - R) / (NIR + R), where NIR is the reflectance of the near-infrared band and R is the reflectance of the red light band; the identification standard for the exposed sandbars is NDVI < 0.2 and the topographic elevation is between the lowest water level and the flood level of the river channel; verify whether the extraction accuracy of the exposed sandbar area meets the requirements through on-site measurement.

[0042] Preferably, the depth residual convolutional neural network is based on the ResNet-50 architecture, includes 5 residual blocks, and a total of 50 convolutional layers; a channel attention module is added in the residual block to improve the sensitivity to the characteristics of the sandbars; a feature pyramid structure is adopted to combine feature maps of different scales to improve the detection ability for sandbars of different sizes; a boundary refinement network is added to accurately extract the boundaries of the sandbars; a combination of weighted cross-entropy loss and boundary-aware loss is used to improve the accuracy of sandbar boundary recognition; among them, the neural network training parameters include the learning rate: the initial value is 0.001, which is adjusted by the cosine annealing strategy; an adaptive moment estimation optimizer is used, the exponential decay rate of the first moment β1 = 0.9, and the exponential decay rate of the second moment β2 = 0.999; the batch size is 16, and the initial number of training epochs is set to 200 epochs. If the validation accuracy does not improve after 10 epochs, the training is completed in advance.

[0043] Preferably, when the cloud cover rate > 5% but < 20%: use cloud detection and cloud removal algorithm (Fmask algorithm) for processing, and perform temporal interpolation to supplement the area under the cloud; when there are data missing in the local area: use the spatio-temporal fusion algorithm to fill in the data in combination with the data at adjacent time points; when there is completely missing data in a specific period: use the Markov chain Monte Carlo method based on historical data of the same period to generate simulated data.

[0044] Preferably, the window length in months is: WL = 2 + 2×(SWAD - SWAD_min) / (SWAD_max - SWAD_min); where SWAD_min is the minimum SWAD value of historical observations; SWAD_max is the maximum SWAD value of historical observations; when the change in the minimum ecological flow calculated twice consecutively exceeds 20%, the step size is reduced to 5 days; when the change in the minimum ecological flow calculated four times consecutively is less than 5%, the step size is increased to 15 days; when the change trends of the minimum ecological flow Q_eco calculated twice consecutively are opposite and the amplitude exceeds 15% in both cases, a change point detection is triggered to determine whether it is a real change in hydro-meteorological conditions or an abnormal fluctuation. When it is determined to be a real change, an artificial verification process is started; when it is determined to be an abnormal fluctuation, a smoothing processing technique is used to reduce the influence of the fluctuation.

[0045] Preferably, it further includes a model dynamic update mechanism: evaluate the improvement of sand and dust activities after the implementation of ecological flow regulation every quarter; when the SWAD value of the sand and dust activity intensity decreases by more than 15% for two consecutive quarters of monitoring, automatically adjust the SWAD_L and SWAD_H values and the sand and dust response model parameters using the latest monitoring data; through the verification set test, when the prediction accuracy improves by more than 5%, update the model parameters, otherwise retain the original model, incorporate the new data into the training set, and wait for the next update; when the cumulative new observation data volume exceeds 25% of the original training set or the environmental conditions change significantly, comprehensively update all models.

[0046] Another aspect of the present invention is to provide a non-transitory readable recording medium for storing one or more programs containing a plurality of instructions, which when executed, will cause the processing circuit to execute the above method for regulating the ecological flow of a river channel.

[0047] Another aspect of the present invention is to provide a system for regulating the ecological flow of a river channel, including a processing circuit and a memory electrically coupled thereto. The memory is configured to store at least one program, the program contains a plurality of instructions, and the processing circuit runs the program to be able to execute the above method for regulating the ecological flow of a river channel.

[0048] Compared with the prior art, the method, recording medium, and system for regulating the ecological flow of a river channel provided by the present invention have the following beneficial effects:

[0049] A seasonal sand and dust activity characteristic value (SWAD) evaluation system is proposed, which breaks through the limitations of a single index and realizes the differential quantification and comparison of the intensity of sand and dust activities. By integrating multi-dimensional indicators such as floating dust, blowing sand, and sandstorms, a unified standard is established, enabling the horizontal comparison of sand and dust risks in different regions and periods.

[0050] Multi-threshold response model: A method for segmental modeling of the intensity of sand and wind activities is proposed, which solves the problem that the traditional single model cannot accurately capture the non-linear relationship between sand and wind activities of different intensities and the exposed area of sandbars. This method automatically adapts to the characteristics of different sand and wind intensity intervals through a mathematical model, especially significantly improving the prediction accuracy in the critical area of sand and wind outbreaks.

[0051] A jump detection algorithm synthesized based on the Pruned Exact Linear Time (PELT) algorithm and the change point detection algorithm is used for the non-linear structure identification of the time series of sand and wind activity intensity values; the automatic identification of the sand and wind activity threshold is realized, getting rid of the deficiencies of the traditional method relying on expert experience and subjective judgment. This technology can accurately capture the change points of sand and wind activity patterns and provide a scientific basis for differential management.

[0052] A dynamic time window algorithm driven by sand and wind intensity is proposed, which solves the limitation that the fixed window cannot adapt to the seasonal changes and emergencies of sand and wind activities. This method can automatically adjust the calculation time window and step size according to the intensity of sand and wind activities, improving the timeliness and accuracy of ecological flow calculation.

[0053] River channel - sand and wind response composite modeling: It breaks through the traditional simple power function relationship between flow and area, introduces a seasonal correction term, and realizes the dynamic response modeling of the exposed area of sandbars to the change of river channel flow. This composite function can simultaneously consider the hydrological process and seasonal periodic influence, improving the area prediction accuracy.

[0054] Aiming at the problem of sandbar area extraction, the deep residual convolutional neural network is improved. Through innovations such as the attention mechanism and boundary enhancement, problems such as fuzzy sandbar boundaries, variable scales, and complex backgrounds are solved, greatly improving the accuracy of sandbar area calculation.

[0055] By integrating technologies in the fields of hydrology, meteorology, geomorphology, and artificial intelligence, a closed-loop linkage system of "exposed area - runoff - sand and wind probability" is established, which solves the problem of decoupling of each state parameter across time and space and realizes the collaborative optimization of ecological flow and sand and wind suppression. Brief Description of the Drawings

[0056] Figure 1 It is the architecture diagram of the "sandbar exposed area - runoff - sand and wind probability" model of the present invention;

[0057] Figure 2 It is the data flow diagram of the model of the present invention;

[0058] Figure 3 It is the non-linear response model of sandbar exposed area - runoff in the embodiment of the present invention;

[0059] Figure 4 It is the relationship between the dust frequency and the exposed area in the section of the southern wide valley of the present invention. Specific Embodiments

[0060] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be described below with reference to the accompanying drawings in the embodiments of the present invention. The described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts fall within the protection scope of the present invention.

[0061] As Figures 1-4 shown, an embodiment of a method for regulating the ecological flow of a river provided by the present invention is as follows:

[0062] 1. Interpretation of multi-temporal remote sensing data

[0063] (1) Acquisition and preprocessing of remote sensing data: The spatial resolution of the high-resolution remote sensing image should be ≤ 30 × 30 m 2 , and the time coverage should be at least 3 complete hydrological years (including wet, normal, and dry years), and the cloud cover rate of the image < 5%;

[0064] Radiometric calibration and atmospheric correction of the remote sensing image are performed using ENVI software. The improved deep residual convolutional neural network (DR-CNN) is used to extract the exposed area of the sandbar, and the classification accuracy needs to be verified through a confusion matrix (Kappa coefficient ≥ 0.80);

[0065] Structure of the deep residual convolutional neural network (DR-CNN): Based on the ResNet-50 architecture, it contains 5 residual blocks and a total of 50 convolutional layers;

[0066] Introduction of an attention mechanism: A channel attention module (Squeeze-and-Excitation module) is added to the residual block to enhance the sensitivity to sandbar features;

[0067] Multi-scale feature fusion: A feature pyramid structure (FPN) is adopted to combine feature maps of different scales to improve the detection ability for sandbars of different sizes;

[0068] Boundary enhancement module: A boundary refinement network (BRN) is added to accurately extract the sandbar boundary;

[0069] Loss function optimization: A combination of weighted cross-entropy loss and boundary-aware loss is used to improve the accuracy of sandbar boundary recognition.

[0070] Network training parameters: The initial value of the learning rate is 0.001, and the cosine annealing strategy is adopted for adjustment; The adaptive moment estimation (Adam) optimizer is used, with β1 = 0.9 and β2 = 0.999; Batch size: 16; Number of training epochs: 200 epochs, and the early stopping threshold is "no improvement in validation accuracy for 10 epochs".

[0071] When the cloud cover rate > 5% but < 20%: Use cloud detection and cloud removal algorithms (Fmask algorithm) for processing, and perform temporal interpolation and supplementation on the area under the cloud; When there are missing data in a local area: Use spatio-temporal fusion algorithms (ESTARFM) to fill in the data by combining data from adjacent time points; When there is completely missing data during a specific period: Use the Markov Chain Monte Carlo (MCMC) method based on historical data of the same period to generate simulated data.

[0072] (2) Method for extracting the exposed area of sandbars:

[0073] Adopt the method combining the Normalized Difference Vegetation Index (NDVI) and supervised classification to extract the vector boundaries of the exposed sandbars in each period, and calculate the area value A_t;

[0074] NDVI calculation formula: NDVI = (NIR - R) / (NIR + R), where NIR is the reflectance of the near-infrared band and R is the reflectance of the red light band;

[0075] Identification criteria for exposed sandbars: NDVI < 0.2 and the topographic elevation is between the lowest water level and the flood level of the river channel;

[0076] Accuracy requirements for extracting the exposed area of sandbars: Relative error < 5%, verified by on-site verification points.

[0077] 2. Modeling the relationship between the exposed area of sandbars and runoff

[0078] (1) Synchronously collect the river channel cross-section flow data Q_t at time point t, with the requirement that the time resolution ≤ 10 days and the flow measurement accuracy ≤ ±3%;

[0079] (2) Establish a non-linear response model between the exposed area A of the sandbar and the river channel cross-section flow Q:

[0080] A = f(Q) = a·e^(-b·Q^c) + ω·sin(2π·t / 365 + φ);

[0081] Since the calculation of the minimum ecological flow requires deriving the flow from the exposed area, an inverse relationship conversion is needed:

[0082] (a)When the exposed area of the sandbar , it is determined that the sandbar is completely submerged, and it is default that no ecological flow adjustment is required;

[0083] (b)Set the threshold of the seasonally exposed sandbar area , when , the runoff of the fixed measurement point cross-section ;

[0084] (c) If A > A thres , fix the runoff of the measuring point section

[0085] Where: A is the exposed area of the sandbar (km²); Q is the cross-sectional flow of the river channel (m³ / s); A sand : total area of the sandbar (determined by the maximum historical exposed area extracted from remote sensing images); Q base : basic ecological flow (m³ / s), taking the 90th percentile of the historical minimum monthly average flow; ζ: submergence threshold coefficient (dimensionless, default 0.02, optimization range 0.01 - 0.05); κ: flow compensation coefficient (m³ / s / m²), determined by linear regression; a is the maximum exposed area coefficient, representing the theoretical maximum exposed area in the zero-flow state; b is the attenuation coefficient, controlling the reduction rate of the exposed area when the flow increases; c is the shape parameter, determining the nonlinear degree of the curve; ω is the seasonal amplitude area, representing the amplitude of the seasonal change of the exposed area; φ is the phase parameter, controlling the time shift of the seasonal change; t is the observed time point, using the serial number sorted from 1 to 365 within the year as the t value; the parameters a, b, c, φ, ω, φ are fitted by the particle swarm optimization (PSO) algorithm.

[0086] Fitting requirements: coefficient of determination R² ≥ 0.90; root mean square error RMSE should be less than 10% of the average exposed area; the width of the 95% confidence interval of each parameter should not exceed 30% of the parameter estimate value; Particle Swarm Optimization algorithm parameter settings: population size: 50; number of iterations: 1000; inertia weight: 0.7; individual learning factor: 1.5; global learning factor: 1.5; termination condition: the improvement amplitude of the optimal solution is < 0.1% for 50 consecutive iterations or reaches the maximum number of iterations.

[0087] 3. Determination of the threshold of wind-sand activity intensity

[0088] (1) Introduce the seasonal wind-sand activity intensity value (SWAD) to characterize the system:

[0089] SWAD = α·FD + β·BD + γ·DS + δ·TSD;

[0090] Where: FD is the floating dust frequency (d), referring to the number of days when the PM10 concentration > 150 μg / m³ and the duration ≥ 2 hours within a unit time;

[0091] BD is the blowing sand frequency (d), referring to the number of days when the horizontal visibility < 5000 m and the duration ≥ 2 hours within a unit time;

[0092] DS is the frequency of sand and dust storms (d), which refers to the number of days with a horizontal visibility < 1000 m and a duration ≥ 30 minutes within a unit time;

[0093] TSD is the duration of sand and dust storms (h), which refers to the cumulative duration in hours of all sand and dust storm events within a unit time;

[0094] α, β, γ, and δ are weight coefficients, which are determined by principal component analysis (PCA).

[0095] Specific method for determining weights:

[0096] 1) Standardize the data of each index: Z = (X - μ) / σ, where X is the original index value, μ is the average value, and σ is the standard deviation;

[0097] 2) Calculate the index correlation matrix R;

[0098] 3) Perform eigenvalue decomposition on R to solve the eigenvalues λ and eigenvectors V;

[0099] 4) Calculate the contribution rate of the principal components. Based on the loading coefficients of the first principal component, the weight coefficients are obtained after normalization, that is, α + β + γ + δ = 1;

[0100] The initial weights are determined by referring to the value ranges of the weight coefficients in different seasons according to the regional characteristics and the importance of historical sand and wind events. The value ranges of the weight coefficients in different seasons are as follows:

[0101] Winter: α ∈ [0.10, 0.20], β ∈ [0.20, 0.30], γ ∈ [0.30, 0.40], δ ∈ [0.20, 0.30];

[0102] Spring: α ∈ [0.15, 0.25], β ∈ [0.25, 0.35], γ ∈ [0.25, 0.35], δ ∈ [0.15, 0.25];

[0103] Summer: α ∈ [0.25, 0.35], β ∈ [0.25, 0.35], γ ∈ [0.15, 0.25], δ ∈ [0.15, 0.25];

[0104] Autumn: α ∈ [0.20, 0.30], β ∈ [0.20, 0.30], γ ∈ [0.20, 0.30], δ ∈ [0.20, 0.30];

[0105] (2) Construct a multi-level sand and wind response model:

[0106] Based on the SWAD value, the intensity of sand and wind activities is divided into three intervals, and different forms of fitting functions are used for each interval:

[0107] ① Mild sand - wind activity interval (SWAD ≤ SWAD_L):

[0108] Use the linear equation: P = k1·A + d1;

[0109] This function is applicable to the low - intensity stage where sand - wind activity has a linear positive correlation with the exposed area. Among them:

[0110] P is the probability of sand - wind activity (a value between 0 and 1);

[0111] A is the exposed area of the sandbar (km²);

[0112] k1 is the linear slope coefficient, representing the increase amplitude of sand - wind probability caused by the increase in exposed area per unit;

[0113] d1 is the constant term, representing the basic sand - wind activity probability;

[0114] Recommended range of parameters: k1 ∈ [0.001, 0.005], d1 ∈ [0.05, 0.15];

[0115] The lower limit of the parameter represents the sand - wind insensitive area, and the upper limit represents the highly sensitive area.

[0116] ② Moderate sand - wind activity interval (SWAD_L < SWAD ≤ SWAD_H):

[0117] Use the quadratic polynomial equation: P = k2·A² + k3·A + d2;

[0118] This function can capture the non - linear characteristic that the sand - wind activity probability increases rapidly with the increase in the exposed area. Among them:

[0119] k2 is the quadratic term coefficient, representing the intensity of the non - linear acceleration effect;

[0120] k3 is the linear term coefficient;

[0121] d2 is the constant term;

[0122] Recommended range of parameters: k2 ∈ [0.00005, 0.0002], k3 ∈ [-0.01, 0.01], d2 ∈ [0.1, 0.3];

[0123] k2 is always positive to ensure the acceleration effect; k3 can be positive or negative, determined according to the characteristics of specific river sections.

[0124] ③ Severe sand - wind activity interval (SWAD > SWAD_H):

[0125] Use the exponential function equation: P = k4·e^(λA)+d3

[0126] This function is applicable to the situation where the wind and sand activity explodes exponentially after the exposed area exceeds the critical threshold, where:

[0127] λ is the exponential growth rate coefficient, which represents the rate of sandstorm outbreak;

[0128] k4 is the proportionality coefficient;

[0129] d3 is a constant term;

[0130] Recommended parameter ranges: k4∈[0.01, 0.1], λ∈[0.005, 0.02], d3∈[0.2, 0.4].

[0131] The parameter value is related to factors such as regional sand sensitivity, surface material composition, and vegetation coverage. Model parameter determination method:

[0132] Data partitioning: historical data is divided into three intervals according to the SWAD threshold;

[0133] Independent fitting of each interval: Use nonlinear least squares method to fit the model parameters of each interval;

[0134] Parameter optimization: Bayesian optimization method is used to minimize the prediction error;

[0135] Model validation: Use cross-validation to evaluate the predictive ability of the model in each interval;

[0136] Parameter adjustment: fine-tune parameters based on verification results and check the smoothness of interval transition;

[0137] (3) Determination and conversion method of the threshold of the wind and sand response model:

[0138] ① Threshold determination method:

[0139] The two key thresholds SWAD_L and SWAD_H are determined in a data-driven manner:

[0140] a. Data preparation: Collect observation data on wind and sand activities for at least three complete hydrological years, including FD, BD, DS, and TSD; divide the data set by season (spring, summer, autumn, and winter), ensuring that the sample size for each season is ≥12 periods.

[0141] b. Multimodal distribution detection:

[0142] Perform kernel density estimation (KDE) on the SWAD data series of each season;

[0143] Analyze the KDE curve morphology and identify multi-peak features;

[0144] Use Hilbert-Huang transform (HHT) to highlight nonlinear features in SWAD data;

[0145] c. Piecewise clustering analysis:

[0146] Use the Expectation-Maximization (EM) algorithm for Gaussian Mixture Model (GMM) fitting;

[0147] Use the Bayesian Information Criterion (BIC) to determine the optimal number of clusters, generally in the range of 3 - 5;

[0148] When the decrease in BIC value is < 5%, select a smaller number of clusters;

[0149] When the difference in BIC for different numbers of clusters is < 10, select the clustering result with a clearer physical meaning;

[0150] Calculate the statistical characteristics (mean, standard deviation, quantiles) of each cluster;

[0151] d. Jump point detection:

[0152] Apply the Pruned Exact Linear Time (PELT) and change point detection algorithms for jump detection, obtaining SWAD_L defined as the SWAD value of the first significant jump point, SWAD_H defined as the SWAD value of the second significant jump point, and SWAD_H > SWAD_L

[0153] e. Threshold verification:

[0154] Use historical sand and dust disaster records to verify the rationality of the threshold;

[0155] Calculate the change rate of sand and dust activity probability at each threshold, with the requirement that the change rate ≥ 40%;

[0156] If the verification fails, adjust to the corresponding quantile value according to the sand and dust activity probability distribution:

[0157] Adjust SWAD_L to approximately the 75% quantile; adjust SWAD_H to approximately the 90% quantile.

[0158] ② Implementation of the Jump Detection Algorithm:

[0159] Algorithm principle: Based on the modified Pruned Exact Linear Time (PELT) and enhanced change point detection method, used for identifying the non-linear structure of the SWAD time series.

[0160] Mathematical expression:

[0161] Cost function: \(C(SWAD_{1:n})=\sum_{i = 1}^{m + 1}[L(SWAD_{\tau_{i-1}+1:\tau_i})+\eta]\);

[0162] where \(L\) is the negative log-likelihood function, \(\eta\) is the complexity penalty term, \(\tau\) is the change point position, and \(m\) is the number of change points;

[0163] Optimal segmentation search: ;

[0164] Pruning strategy: If for any \(k\lt t\lt s\), \(F(t)+C(SWAD_{t + 1:s})\geq F(s)\), then \(t\) can be pruned from the candidate change points after \(k\);

[0165] Program example, algorithm pseudocode:

[0166] function jump_detection_algorithm(SWAD_series, penalty = 0.05, min_segment_length = 5, significance_level = 0.01):

[0167] """

[0168] Implement the jump detection algorithm for SWAD sequences

[0169] Parameters:

[0170] SWAD_series: SWAD time series data

[0171] penalty: Complexity penalty parameter, default 0.05

[0172] min_segment_length: Minimum segment length, default 5

[0173] significance_level: Significance level, default 0.01

[0174] Returns:

[0175] jump_points: List of detected jump points and their SWAD values

[0176] """

[0177] n = len(SWAD_series)

[0178] # Cost matrix initialization

[0179] cost = zeros(n + 1)

[0180] cost[0] = -penalty # Boundary condition

[0181] # Optimal segmentation record

[0182] cp = {} # Save the optimal change point at each position

[0183] cp[0] = []

[0184] # Feasible set (pruning set)

[0185] R = {0} # Start searching from position 0

[0186] # Main loop

[0187] for s in range(1, n+1):

[0188] # Calculate the cost function

[0189] candidates = {}

[0190] for t in R:

[0191] if s - t < min_segment_length:

[0192] continue

[0193] # Calculate the negative log-likelihood of the current segment

[0194] segment_data = SWAD_series[t:s]

[0195] segment_cost = neg_log_likelihood(segment_data)

[0196] candidates[t] = cost[t] + segment_cost + penalty

[0197] # Find the optimal previous change point

[0198] t_star = min(candidates, key=lambda t: candidates[t])

[0199] cost[s] = candidates[t_star]

[0200] cp[s] = cp[t_star] + [t_star]

[0201] # Pruning the search space

[0202] R_new = {s} # Always include the latest position

[0203] for t in R:

[0204] if cost[t] + min_cost_possible(SWAD_series[t:]) < cost[s]:

[0205] R_new.add(t)

[0206] R = R_new

[0207] # Extracting jump points

[0208] jump_points = cp[n]

[0209] # Significance test

[0210] validated_jumps = []

[0211] for jump in jump_points:

[0212] # Calculating the mean of SWAD before and after the jump point

[0213] pre_jump_mean=mean(SWAD_series[max(0,jump-min_segment_length):jump])

[0214] post_jump_mean=mean(SWAD_series[jump:min(n, jump+min_segment_length)])

[0215] # Performing the t-test

[0216] t_stat, p_value = ttest_ind(

[0217] SWAD_series[max(0, jump-min_segment_length):jump],

[0218] SWAD_series[jump:min(n,jump+min_segment_length)] )

[0219] # Calculating the percentage of change amplitude

[0220] change_percent = abs((post_jump_mean - pre_jump_mean) / pre_jump_mean) * 100

[0221] # Only retain significant and large - magnitude jump points

[0222] if p_value < significance_level and change_percent >= 40:

[0223] validated_jumps.append((jump, change_percent, p_value))

[0224] # Sort according to the change magnitude

[0225] validated_jumps.sort(key = lambda x: x[1], reverse = True)

[0226] # Return the jump point positions and their SWAD values

[0227] return [(jump, SWAD_series[jump]) for jump, _, _ in validated_jumps]

[0228] function neg_log_likelihood(data):

[0229] """

[0230] Calculate the negative log - likelihood of a data segment

[0231] Parameters:

[0232] data: Data sequence

[0233] Returns:

[0234] nll: Negative log - likelihood value

[0235] """

[0236] # Assume the data follows a normal distribution

[0237] mu = mean(data)

[0238] sigma = std(data)

[0239] n = len(data)

[0240] # Calculate negative log-likelihood

[0241] # L = -n / 2 * log(2π) - n / 2 * log(σ²) - 1 / (2σ²) * Σ( - μ)²

[0242] nll = n / 2 * log(2*π) + n / 2 * log(sigma**2) + sum((data - mu)**2) / (2 * sigma**2)

[0243] return nll

[0244] function min_cost_possible(data):

[0245] """

[0246] Calculate the minimum possible cost for a data segment

[0247] Parameters:

[0248] data: data sequence

[0249] Returns:

[0250] min_cost: minimum possible cost value

[0251] """

[0252] # The minimum possible cost assumes the entire sequence as one segment

[0253] return neg_log_likelihood(data)

[0254] ```

[0255] Algorithm parameter settings:

[0256] Complexity penalty parameter (η): Winter: 0.035, Spring: 0.040, Summer: 0.030, Autumn: 0.035;

[0257] Minimum segment length: 5 data points (about 50 days) to ensure the detected change points have temporal persistence;

[0258] Significance level: p < 0.01, requiring a significant difference in SWAD values before and after the change point;

[0259] Required change magnitude: ≥40% to ensure the detected change points have practical physical significance;

[0260] Post - processing method:

[0261] Temporal smoothing: LOESS (Locally Weighted Scatterplot Smoothing) is used to reduce noise interference;

[0262] Mutation point clustering: Mutation points with a distance less than 30 days are merged;

[0263] Seasonal adjustment: The detection threshold is dynamically adjusted by ±10% according to seasonal characteristics;

[0264] ③ Model conversion smoothing processing:

[0265] To avoid jumps during model conversion between adjacent intervals, a weighted transition zone design is adopted:

[0266] a. Set a transition interval of ±0.5 SWAD units on both sides of each segmentation point (SWAD_L, SWAD_H);

[0267] b. In the transition zone, the weighted fusion method is used to calculate the sand - dust probability:

[0268] Taking SWAD_L as an example, when SWAD ∈ [SWAD_L - 0.5, SWAD_L + 0.5]:

[0269] P = w1·P_mild + w2·P_moderate;

[0270] where the weights w1 = (SWAD_L + 0.5 - SWAD) / 1.0 and w2 = 1 - w1;

[0271] c. Smooth transition ensures the continuity of the first - order derivative of the model at the segmentation point, avoiding unreasonable jumps in the prediction results;

[0272] (4) Sand - dust response model parameter optimization method:

[0273] ① Parameter initialization:

[0274] The initial values of the model parameters for each interval are obtained through piece - wise linear regression of historical data, requiring the initial fitting R² ≥ 0.75;

[0275] ② Parameter optimization algorithm:

[0276] The Bayesian optimization framework is used to determine the optimal parameter set for each model:

[0277] a. Objective function: Minimize the log - loss between the predicted sand - dust probability and the actual observation:

[0278] L(θ) = -∑[y_i·log(P_i) + (1 - y_i)·log(1 - P_i)]

[0279] where $\theta$ represents the set of model parameters, $y_i$ is the actually observed sand and dust event (0 or 1), and $P_i$ is the model prediction probability

[0280] b. Bayesian optimization process:

[0281] Set the prior distribution of parameters: Based on the physical meaning, set a reasonable prior distribution for each parameter;

[0282] Use Gaussian process regression to construct a probabilistic surrogate model of the objective function;

[0283] Determine the next set of parameters to be evaluated by maximizing the expected improvement (EI) acquisition function;

[0284] Iteratively optimize until convergence (the improvement amplitude in 5 consecutive iterations < 1% or reach the maximum number of iterations);

[0285] c. Cross-validation: Use 10-fold cross-validation to evaluate the parameter robustness, and require the standard deviation of the performance on each fold of the validation set < 10%;

[0286] ③ Seasonal parameter adjustment:

[0287] The model parameters are corrected according to the season, and the season adjustment coefficient $S_{adj}$ is introduced:

[0288] a. Winter (December - February): Keep the original parameters unchanged, $S_{adj}=1.0$;

[0289] b. Spring (March - May): The period of intensified sand and dust activities, the non-linear coefficient is increased, $k2_{spring}=k2\cdot1.15$, $\lambda_{spring}=\lambda\cdot1.20$, and other parameters are multiplied by $S_{adj}=1.05$;

[0290] c. Summer (June - August): The period of weakened sand and dust activities, each parameter is multiplied by $S_{adj}=0.85$;

[0291] d. Autumn (September - November): The transitional period of sand and dust activities, each parameter is multiplied by $S_{adj}=0.95$.

[0292] ④ Model evaluation index system:

[0293] a. Precision: The proportion of correctly predicted sand and dust events in all predicted events, requiring $\geq0.80$;

[0294] b. Recall: The proportion of correctly predicted sand and dust events in all actual events, requiring $\geq0.75$;

[0295] c. F1 value: The harmonic mean of precision and recall, requiring $\geq0.75$;

[0296] d. AUC value: Area Under the ROC Curve, required to be ≥ 0.85;

[0297] e. Brier Skill Score (BSS): Prediction skill relative to the climate average, required to be ≥ 0.40.

[0298] (5) The Gradient Boosting Decision Tree (GBDT) model is used to integrate seasonal factors, meteorological conditions, and the exposed area of sandbars to predict the probability of sand and dust activities. The AUC value of the model prediction accuracy needs to be ≥ 0.85.

[0299] (6) Application example of the three-stage sand and dust response model:

[0300] Taking the middle reaches of the Yarlung Zangbo River in Shannan as an example, combined with the measured data from 2018 to 2020, the model construction and application process are as follows:

[0301] ① SWAD calculation and threshold determination:

[0302] According to the sand and dust event data recorded by the local meteorological station, calculate the SWAD sequence and determine it through the jump detection algorithm:

[0303] SWAD_L = 8.5 (light and moderate demarcation point);

[0304] SWAD_H = 13.2 (moderate and severe demarcation point);

[0305] ② Optimization results of model parameters in each interval:

[0306] Light interval (SWAD ≤ 8.5):

[0307] k1 = 0.0025, d1 = 0.08, goodness of fit R² = 0.86;

[0308] Moderate interval (8.5 < SWAD ≤ 13.2):

[0309] k2 = 0.00012, k3 = 0.0015, d2 = 0.15, goodness of fit R² = 0.92;

[0310] Severe interval (SWAD > 13.2):

[0311] k4 = 0.035, λ = 0.011, d3 = 0.28, goodness of fit R² = 0.94;

[0312] ③ Model prediction verification:

[0313] During the high-incidence period of sand and dust in spring 2020 (March - May):

[0314] Actual observed incidence of sandstorm events: 56.7%;

[0315] Probability of sandstorm predicted by the three-stage model: 52.3%, prediction error 7.8%;

[0316] Prediction by the single linear model: 42.1%, prediction error 25.7%;

[0317] Further evaluation metrics:

[0318] Precision: 0.83; Recall: 0.79; F1-score: 0.81; AUC: 0.88; BSS: 0.45;

[0319] The verification results show that the three-stage model can accurately capture the non-linear outbreak characteristics of the high sandstorm period, significantly outperforming the traditional single model.

[0320] 4. Dynamic calculation and regulation of the minimum ecological flow

[0321] (1) Determination of the critical bare area: Based on the sandstorm response model, determine the critical bare area A_critical acceptable for sandstorm risk; In the mild sandstorm interval, take the maximum A value with P ≤ 0.3 as A_critical_light; In the moderate sandstorm interval, take the maximum A value with P ≤ 0.5 as A_critical_medium; In the severe sandstorm interval, take the maximum A value with P ≤ 0.7 as A_critical_heavy; Select the corresponding critical area threshold according to the SWAD value.

[0322] (2) Calculation of the minimum ecological flow:

[0323] Substitute A_critical into the inverse A-Q relationship model in step 2 to solve the minimum ecological flow Q_eco that meets the sandstorm suppression requirement. The required flow is dynamically adjusted according to the season and sandstorm intensity, forming an adaptive minimum ecological flow curve.

[0324] (3) Adaptive dynamic sliding window algorithm:

[0325] ① Principle of window length design:

[0326] Dynamically adjust the sliding window length according to the sandstorm activity intensity, with a range of 2 - 4 months and a step size of 10 days.

[0327] Calculation formula for the window length (WL):

[0328] WL = 2 + 2 × (SWAD - SWAD_min) / (SWAD_max - SWAD_min) (months);

[0329] Where:

[0330] SWAD is the current seasonal wind-sand activity characteristic value;

[0331] SWAD_min is the minimum SWAD value observed historically;

[0332] SWAD_max is the maximum SWAD value observed historically;

[0333] When SWAD ≤ SWAD_L, WL takes the lower limit value of 2 months;

[0334] When SWAD ≥ SWAD_H, WL takes the upper limit value of 4 months;

[0335] When SWAD_L < SWAD < SWAD_H, WL is calculated by linear interpolation;

[0336] ② Window sliding and update strategy:

[0337] a. The basic step size is 10 days, and it slides forward by one step each time;

[0338] b. Calculate the average SWAD value within the new window and adjust the window length accordingly;

[0339] c. When the change in the minimum ecological flow between two consecutive calculations exceeds 20%, reduce the step size to 5 days to improve the time resolution;

[0340] d. When the change in the minimum ecological flow in four consecutive calculations is less than 5%, increase the step size to 15 days to reduce the calculation amount;

[0341] ③ Change point detection trigger mechanism:

[0342] a. When the change trends of Q_eco are opposite in two consecutive times and the amplitudes both exceed 15%, trigger the change point detection;

[0343] b. Judge whether it is a real change in hydrometeorological conditions or an abnormal fluctuation;

[0344] c. When it is determined to be a real change, start the manual verification process;

[0345] d. When it is determined to be an abnormal fluctuation, use the smoothing processing technology to reduce the influence of the fluctuation;

[0346] (4) Manually adjust the runoff of the fixed measurement point section of the river channel to be above the minimum ecological flow, and update the model data for optimization in real time. Continuously optimize the minimum ecological flow calculation model by monitoring the changes in sand and wind activities after real-time application:

[0347] ① Establish an application effect evaluation system to evaluate the improvement of sand and wind activities after the implementation of ecological flow regulation every quarter:

[0348] a. Define the improvement index: SWAD improvement rate = (SWAD_before - SWAD_after) / SWAD_before × 100%;

[0349] b. Sandbar area control rate = (A_before - A_after) / A_before × 100%;

[0350] c. Sand and wind event probability reduction rate = (P_before - P_after) / P_before × 100%;

[0351] ② Trigger the update condition:

[0352] a. When the SWAD value is monitored to decrease by more than 15% for two consecutive quarters, trigger the automatic update process of model parameters;

[0353] b. When the cumulative new observation data volume exceeds 25% of the original training set, conduct a comprehensive update;

[0354] c. When significant changes occur in environmental conditions (such as the completion of large-scale engineering construction in the basin), force an update;

[0355] ③ Bayesian update framework:

[0356] Use the latest monitoring data to correct the parameters of the non-linear response model:

[0357] a. Bayesian formula: P(θ|D_new,D_old) ∝ P(D_new|θ)·P(θ|D_old)

[0358] Where:

[0359] θ represents the model parameter set (a, b, c, ω, φ, etc.);

[0360] D_new is the newly obtained monitoring data;

[0361] D_old is the historical training data;

[0362] P(θ|D_old) is the prior distribution of the parameters;

[0363] P(D_new|θ) is the likelihood function;

[0364] P(θ|D_new,D_old) is the posterior distribution of the updated parameters;

[0365] b. Implementation method: The Markov Chain Monte Carlo (MCMC) method is used for posterior distribution sampling;

[0366] The Metropolis-Hastings algorithm is used to construct a Markov chain;

[0367] The first 20% of the samples are discarded as the burn-in period;

[0368] The last 80% of the samples are retained to calculate the posterior mean and 95% confidence interval of the parameters;

[0369] c. Parameter update constraints:

[0370] The change rate of any parameter in a single update does not exceed ±20%;

[0371] The updated parameter combination must satisfy the physical meaning constraints;

[0372] The goodness of fit R² of the updated model shall not be lower than 95% of the original model;

[0373] ④ Dynamic adjustment of the SWAD threshold:

[0374] a. The SWAD_L and SWAD_H thresholds are dynamically adjusted with the improvement of sand and dust activities;

[0375] b. A similar Bayesian framework is used to update the posterior distribution of the threshold;

[0376] c. The threshold change rate is limited to ±10% / year to ensure system stability;

[0377] ⑤ Update verification mechanism:

[0378] a. An independent verification dataset (the latest 20% of the data) is constructed;

[0379] b. The evaluation metrics include RMSE, MAE, R², and prediction bias;

[0380] c. The new model is only officially applied when the prediction accuracy improvement exceeds 5%;

[0381] d. Otherwise, the original model is retained, and the new data is incorporated into the training set and waits for the next update.

[0382] (5) Scope of application and limitations

[0383] River reach types applicable: Seasonal sandstorm activities in low- and mid-latitude (15°-45°) regions affect river sections;

[0384] The riverbed is mainly sandy, with significant seasonal sandbar exposure;

[0385] Wide and shallow rivers with a river width / water depth ratio of >20;

[0386] River sections with annual flow variation >300%;

[0387] Data requirements:

[0388] Remote sensing data: spatial resolution ≤ 30m, temporal resolution ≤ 1 month, continuous observation ≥ 3 years;

[0389] Flow data: time resolution ≤ 10 days, measurement error ≤ ± 3%;

[0390] Meteorological data: at least three indicators including wind speed, visibility, and PM10 concentration;

[0391] Application Restrictions:

[0392] It is not applicable to high-altitude river sections where the ice and snow cover period is more than 2 months;

[0393] It is not applicable to river sections without obvious seasonal sand bar exposure features;

[0394] It is not applicable to areas where there is no significant correlation between wind and sand activities and the exposed area of the river channel;

[0395] The river sections that are strongly regulated by water conservancy projects need to add human factor correction modules;

[0396] The model applies boundary conditions:

[0397] The SWAD index system is applicable to areas where the annual change in probability of wind and sand activity is ≥ 200%;

[0398] The AQ relationship model is applicable to river sections with sandbar area response curve R²≥0.75;

[0399] When encountering extreme climate events (once-in-a-century rainstorms, severe sandstorms), expert intervention is required.

[0400] Compiling the above method steps into a program and then storing it on a hard disk or other non-transitory storage medium constitutes an embodiment of "a non-transitory readable recording medium" of the present invention; and electrically connecting the storage medium to a computer processor and being able to regulate the ecological flow of a river through data processing constitutes an embodiment of "a system for regulating the ecological flow of a river" of the present invention.

[0401] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0402] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0403] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0404] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0405] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for regulating the ecological flow of a river channel, characterized in that, It includes the following steps: S1. Collect the time series data of the exposed area of sandbars in the river channel, the runoff of the fixed measuring point section, and the probability of sand and wind activities; S2. Establish a function model between the exposed area of the sandbar and the runoff of the fixed measuring point section at the same moment, and use the particle swarm algorithm to optimize the parameters of the function model; S3. Establish a sand and wind activity intensity model, determine the intermediate value of the two sand and wind activity intensities through the jump detection algorithm, and divide the sand and wind activity intensity into three intervals: mild, moderate, and severe by the two intermediate values; among them, the jump detection algorithm is synthesized based on the trimmed linear time exact algorithm and the change point detection algorithm, and is used for the non-linear structure identification of the time series of sand and wind activity intensity values; S4. Establish a sand and wind response model for describing the relationship between the probability of sand and wind activities and the exposed area of the sandbar in the same time period for the three intervals respectively. For the continuous sampling of the time series data involved in the function model, use the adaptive dynamic sliding window algorithm, and the window length is dynamically adjusted according to the sand and wind activity intensity, with a range of 2-4 months and an initial sliding step of 10 days; S5. Determine the critical exposed area of the sandbar according to the set probability threshold of sand and wind activities in the sand and wind response model, substitute the critical exposed area into the function model in step S2, and solve the corresponding runoff of the fixed measuring point section as the minimum ecological flow; adjust the sliding step according to the change range of the minimum ecological flow within the set time period; S6. Manually adjust the runoff of the fixed measuring point section of the river channel to be above the minimum ecological flow, and update the model data for optimization in real time.

2. The method for regulating the ecological flow of a river channel according to claim 1, characterized in that The sand and wind activity intensity model is: set the sand and wind activity intensity SWAD = α·FD + β·BD + γ·DS + δ·TSD; where, FD is the floating dust frequency, BD is the sand blowing frequency, DS is the sandstorm frequency, TSD is the dust storm duration, and α, β, γ, δ are seasonal weight coefficients, which are determined by the principal component analysis method in the same season; set the intermediate value separating the mild and moderate intervals as SWAD_L, and the intermediate value separating the moderate and severe intervals as SWAD_H; then the three sand and wind response models: In the mild interval, the sand and wind response model corresponding to the state of SWAD≤SWAD_L: P = k1·A + d1; In the moderate interval, the sand and wind response model corresponding to the state of SWAD_L<SWAD≤SWAD_H: P = k2·A² + k3·A + d2; In the severe interval, the sand and wind response model corresponding to the state of SWAD>SWAD_H: P = k4·e λA + d3; where, λ is the exponential growth rate coefficient, representing the rate of sand and wind outbreaks, k1, k2, k3, k4 are proportionality coefficients, d1, d2, d3 are constant terms, which are optimized by the Bayesian algorithm, P is the probability of sand and wind activities, A is the exposed area of the sandbar, and belongs to the observed data.

3. The method for regulating the ecological flow of a river channel according to claim 2, characterized in that The function model between the exposed area of the sandbar and the runoff of the fixed measuring point section in step S2 is: When the exposed area of the sandbar is such that it is determined that the sandbar is completely submerged, it is defaulted that there is no need to adjust the ecological flow rate; Set the area threshold of seasonal exposed sandbars , when , the runoff of the fixed measuring point section ; If A > A thres , fix the runoff of the measuring point section ; Where: A sand : The total area of exposed sandbars is determined by the maximum historical exposed area extracted from remote sensing images; ζ: Submergence threshold coefficient, default 0.02, optimization range 0.01-0.05; Q base : Base ecological flow, taking the 90th percentile of the historical minimum monthly average flow; κ: Flow compensation coefficient, determined by linear regression; a is the maximum exposed area coefficient, which indicates the theoretical maximum exposed area under zero flow condition; b is the attenuation coefficient, which represents the rate of reduction of the exposed area when the flow rate increases; c is the shape parameter, which determines the nonlinearity of the function curve; μ is the average baseline exposed area; ω is the seasonal variation coefficient, which represents the fluctuation range of the exposed area relative to the base area; φ is the phase parameter, which characterizes the time offset of seasonal changes; t is the time point of observation, and the serial number in the year from 1 to 365 is used as the t value; Parameters a, b, c, μ, ω, and φ are optimized by particle swarm optimization, and their hyperparameters are: The goodness of fit R²≥0.90; the root mean square error of the sandbar exposed area A is less than 10% of the average sandbar exposed area.

4. A method for regulating the ecological flow of a river channel according to claim 3, characterized in that, The steps to collect time series data of exposed sandbar area in the river include: The selected spatial resolution of remote sensing images should be ≤ 30×30 m 2 , with a time coverage of at least 3 hydrological years including wet, normal, and dry years. When the cloud cover rate of the images is ≤ 5%, collect remote sensing images of the river channel area; The ENVI software was used to perform radiometric calibration and atmospheric correction on the remote sensing images. The normalized difference vegetation index NDVI was combined with the supervised classification of the deep residual convolutional neural network to extract the vector boundaries of the exposed sandbars in each period and calculate the exposed area A of the sandbars. NDVI = (NIR - R) / (NIR + R), where NIR is the reflectivity of the near-infrared band and R is the reflectivity of the red light band. The identification standard for exposed sandbars is NDVI < 0.2 and the terrain elevation is between the lowest water level and the flood level of the river. Field measurements were performed to verify whether the accuracy of the exposed area extraction of the sandbars met the requirements.

5. A method for regulating the ecological flow of a river channel according to claim 4, characterized in that, The deep residual convolutional neural network is based on the ResNet-50 architecture, which contains 5 residual blocks and a total of 50 convolutional layers. A channel attention module is added to the residual block to improve the sensitivity to sandbar features. A feature pyramid structure is adopted, combined with feature maps of different scales, to improve the detection ability of sandbars of different sizes; a boundary refinement network is added to accurately extract sandbar boundaries; a combination of weighted cross entropy loss and boundary-aware loss is used to improve the accuracy of sandbar boundary recognition; the neural network training parameters include learning rate: the initial value is 0.001, which is adjusted by the cosine annealing strategy; an adaptive moment estimation optimizer is used, the exponential decay rate of the first-order moment is β1=0.9, and the exponential decay rate of the second-order moment is β2=0.999; the batch size is 16, and the number of training rounds is initially set to 200. If there is no improvement in the verification accuracy after 10 rounds, the training will be completed in advance.

6. The method for regulating the ecological flow of a river channel according to claim 3, characterized in that, When the cloud coverage is >5% but <20%, cloud detection and cloud removal algorithms are used, and time series interpolation is performed on the area under the cloud. When data is missing in a local area, a space-time fusion algorithm is used to fill the data by combining data from adjacent time points. When data is completely missing for a specific period of time, a Markov Chain Monte Carlo method based on historical data for the same period is used to generate simulated data.

7. A method for regulating the ecological flow of a river channel according to claim 3, characterized in that, The window length in months is: WL = 2 + 2×(SWAD - SWAD_min) / (SWAD_max - SWAD_min); where SWAD_min is the minimum SWAD value of historical observations; SWAD_max is the maximum SWAD value of historical observations; when the change in the minimum ecological flow for two consecutive calculations exceeds 20%, the step size is reduced to 5 days; when the change in the minimum ecological flow for four consecutive calculations is less than 5%, the step size is increased to 15 days; when the change trends of the minimum ecological flow Q_eco calculated twice consecutively are opposite and the amplitudes both exceed 15%, a change point detection is triggered to determine whether it is a real change in hydro-meteorological conditions or an abnormal fluctuation. When it is determined to be a real change, an artificial verification process is started; when it is determined to be an abnormal fluctuation, a smoothing processing technique is used to reduce the impact of the fluctuation.

8. A method for regulating the ecological flow of a river channel according to claim 3, characterized in that It also includes a model dynamic update mechanism: The improvement of sandstorm activities after the implementation of ecological flow regulation is evaluated quarterly; when the SWAD value of the sandstorm activity intensity decreases by more than 15% for two consecutive quarters, the SWAD_L and SWAD_H values and the sandstorm response model parameters are automatically adjusted dynamically using the latest monitoring data; through the verification set test, when the prediction accuracy improves by more than 5%, the model parameters are updated, otherwise the original model is retained, the new data is incorporated into the training set, and waiting for the next update; when the cumulative amount of newly added observation data exceeds 25% of the original training set or the environmental conditions change significantly, all models are comprehensively updated.

9. A non-transitory readable recording medium for storing one or more programs including a plurality of instructions, characterized in that, When the instruction is executed, it will cause the processing circuit to execute a method for regulating the ecological flow of a river channel according to any one of claims 1-8.

10. A system for regulating the ecological flow of a river channel, comprising a processing circuit and a memory electrically coupled thereto, characterized in that, The memory is configured to store at least one program, the program includes a plurality of instructions, and when the processing circuit runs the program, it can execute a method for regulating the ecological flow of a river channel according to any one of claims 1-8.

Images