Staged design flood estimation method under influence of climate change and reservoir regulation and storage

Abstract

The invention discloses a staged design flood estimation method under the influence of climate change and reservoir regulation and storage. The method comprises the following steps: acquiring meteorological and hydrological data, human water consumption data and global climate mode output data of a reservoir; a reduced natural reservoir runoff series is obtained based on a water balance method, a GR4J-9 hydrological model is calibrated to achieve natural reservoir runoff simulation, and a long-short-term memory model is constructed to simulate the influence of human activities such as reservoir regulation and storage on runoff; based on a global climate mode set and a quantile deviation correction method, driving a GR4J-9 hydrological model and a long-short term memory model to obtain a reservoir runoff series under climate change and performing flood season staging; establishing a staging design flood calculation model of a future scene based on a Copula function, and deducing the most probable staging design flood; and constructing an emergence constraint model based on the dew point temperature so as to obtain a corrected stage design flood result. According to the method, the comprehensive influence of climate change and reservoir regulation and storage is considered, and a reference basis is provided for deducing stage design flood.

Description

Technical Field

[0001] The present invention relates to the technical field of design flood estimation, and in particular to a method for estimating design floods by stages under the influence of climate change and reservoir storage. Background Art

[0002] The staged design flood is an important basis for determining the staged flood limit water level. It is essential to optimize the design flood to reflect the evolution of floods during the flood season. The staged design flood must not only meet flood control standards but also reflect the seasonal patterns of floods. my country's current staged design flood method assumes that the frequency of staged design floods is equal to the inverse of the flood control standard, which does not meet the requirements of the flood control standard. Domestic and foreign scholars have made some attempts to explore staged design flood calculation techniques that simultaneously meet these two conditions. These techniques are mainly categorized into two types: the full probability formula method and the joint distribution method.

[0003] The total probability formula method assumes that the annual maximum floods for each phase are independent of each other. Using the total probability formula, a relationship is established between the annual maximum flood and the annual maximum flood for each phase, proposing a method for estimating the phased design flood. However, this method generates very small samples during non-main flood seasons, ultimately resulting in unreliable phased frequency analysis results. Furthermore, this method differs from the current sampling method for phased maximum values ​​in my country, making it difficult to directly apply. The key technology behind the latter lies in the construction of a joint distribution based on a Copula function. However, the joint distribution method yields countless combinations of phased design flood results, posing a critical challenge in selecting scientifically appropriate design values ​​on contour lines or surfaces. The aforementioned literature often artificially assumes the same flood frequency for each phase, but the rationale for this assumption warrants further investigation.

[0004] Another prominent issue is that, due to climate change and upstream reservoir regulation, floodwaters entering reservoirs are dynamically changing, posing a challenge in estimating phased design floods that reflect objective physical laws. If reservoir operation and management continue to rely on construction-period design floods while ignoring the regulation and storage effects of upstream cascade reservoirs, this will inevitably lead to low operating water levels during flood season and a need to improve comprehensive utilization efficiency. Furthermore, since the late Second Industrial Revolution, with the continuous development of human society, atmospheric carbon dioxide concentrations have also increased significantly, and global warming has become an indisputable fact. As the two most important components of a changing environment, climate change and human activities (primarily reservoir operation and management) have garnered widespread attention both domestically and internationally for their impacts on hydrological processes. However, most current research focuses on the impact of either climate change or reservoir operation and management on hydrological processes, while few studies consider the combined impact of both on phased design floods. Furthermore, no literature has effectively controlled the uncertainty in estimating phased design floods.

[0005] Therefore, it is necessary to design a phased design flood estimation method under the influence of climate change and reservoir regulation to overcome the above problems. Summary of the Invention

[0006] To avoid the above problems, a method for estimating staged design floods under the influence of climate change and reservoir storage is provided. This method not only considers the combined impact of climate change and reservoir storage, but also considers the correlation between flood characteristics in different stages, providing a reference basis for estimating staged design floods.

[0007] The present invention provides a method for estimating staged design floods under the influence of climate change and reservoir regulation, comprising the following steps:

[0008] S1. Collect daily runoff data for reservoirs, including engineering characteristics and operational data, and extract meteorological and hydrological reanalysis data, water diversion data, industrial and agricultural water use data, and global climate model output data for the reservoir basin.

[0009] S2. Based on the water balance method, the restored natural inflow runoff series is obtained and the GR4J-9 hydrological model is calibrated to simulate the natural inflow runoff.

[0010] S3. Based on natural inflow runoff, actual inflow runoff, and meteorological and hydrological reanalysis data, a long short-term memory (LSTM) model was constructed to simulate the impact of human activities on runoff and its parameters were calibrated.

[0011] S4. Based on a global climate model ensemble and quantile bias correction methods, a meteorological simulation series under climate change is obtained. This is then used to drive the GR4J-9 hydrological model and the long short-term memory (LSTM) model to obtain a series of daily inflows under climate change and stratify the flood season.

[0012] S5. Based on the daily runoff series obtained in step S4, a Copula function is used to establish a staged design flood calculation model for future scenarios to derive the most likely staged design flood;

[0013] S6. Based on the most likely staged design flood value derived in step S5 and the global climate model output data, an emergence constraint model based on dew point temperature is constructed to obtain a corrected staged design flood result.

[0014] Preferably, step S1 specifically includes:

[0015] 1.1 Obtain the reservoir's daily runoff series from the reservoir management agency, collect data on the reservoir's engineering characteristics and operation, and extract basic data on water diversion and industrial and agricultural water use in the basin where the reservoir is located;

[0016] 1.2 For each grid point in the reservoir basin, daily meteorological and hydrological reanalysis data were collected from the fifth-generation atmospheric reanalysis dataset ERA5-Land of the European Centre for Medium-Range Weather Forecasts for the period 1950–2022. Specifically, these data included: daily maximum and minimum temperatures, daily mean temperatures, air pressure, dew point temperature, snowfall, precipitation, runoff depth, shortwave radiation, and longwave radiation.

[0017] 1.3 Obtain the daily maximum temperature, daily minimum temperature, daily mean temperature, daily mean relative humidity, daily precipitation, daily snowfall, daily runoff depth, daily mean specific humidity, daily shortwave radiation intensity, and daily longwave radiation for each global climate model over the historical period 1950–2014;

[0018] 1.4 Obtain daily temperature, relative humidity, precipitation, snowfall, runoff depth, specific humidity, shortwave radiation intensity, and longwave radiation for multiple global climate models for the future period 2015–2100 under the shared socioeconomic pathway of SSP585.

[0019] 3. A method for estimating design floods in stages under the influence of climate change and reservoir regulation as described in claim 2, characterized in that 35 global climate models are selected.

[0020] Preferably, step S2 specifically includes:

[0021] 2.1 Based on the reservoir inflow and outflow daily runoff series obtained in step S1, the reservoir's engineering characteristics and operation-related data, the water diversion data of the reservoir's basin, and the basic data of industrial and agricultural water use, the water balance method is used to invert the reservoir inflow runoff series under natural conditions;

[0022] 2.2 The Thiessen polygon method is used to derive the basin-mean daily series of meteorological variables from the ERA5-Land reanalysis dataset and global climate model output data;

[0023] 2.3 Based on the reservoir runoff series under natural conditions and the daily precipitation, daily maximum temperature and daily minimum temperature data of ERA5-Land during the same period, the GR4J-9 hydrological model with snowmelt module was calibrated;

[0024] 2.4 The daily precipitation, daily maximum temperature, and daily minimum temperature data of ERA5-Land during the period 1950-2022 were used to drive the calibrated GR4J-9 hydrological model to obtain the simulated natural inflow runoff series from 1950 to 2022.

[0025] Preferably, step S3 specifically includes:

[0026] 3.1 Based on the observed daily runoff series and the natural runoff series simulated in step S2, the runoff series affected by human activities is derived as follows:

[0027] Q hum =Q in -Q nat ;

[0028] Where Q in is the inflow runoff data obtained from the reservoir management agency, Q nat is the natural inflow runoff series simulated in step S2, Q hum The inflow runoff series is affected by human activities;

[0029] 3.2 Derived the daily average relative humidity and specific humidity, and constructed a long short-term memory (LSTM) model to simulate the inflow runoff series affected by human activities;

[0030] 3.3 By calculating Q obtained in step 3.1 hum Statistical analysis of the series and basin-averaged daily precipitation was performed to determine the lag time that affects the long-short term memory model;

[0031] 3.4 Based on the data obtained in the above steps, a long short-term memory model LSTM with a three-layer neural network architecture was constructed, with the average daily meteorological data of the ERA5-Land basin as the explanatory variable and Q hum Taking the target variable as input, calibrate the long short-term memory model LSTM;

[0032] 3.5 The runoff series affected by human activities is simulated by the calibrated long short-term memory model LSTM. The simulation equation is: Q cor (t) = F LSTM [QM(t),QM(t-1),QM(t-2),…,QM(tN)]];

[0033] Where Q cor (t) represents the runoff series under the influence of human activities simulated at time t; QM(t) represents the input variables for calibrating the long short-term memory model LSTM, including the daily average temperature, daily average relative humidity, daily precipitation, daily snowfall, daily runoff depth, daily average specific humidity, daily shortwave radiation intensity, and daily longwave radiation of ERA5-Land; QM(t-1) represents the meteorological and hydrological variables at time t-1, and N represents the lag determined by the long short-term memory model LSTM; F LSTM Represents the calibrated long short-term memory model LSTM.

[0034] Preferably, the long short-term memory model LSTM is trained using the minimum batch gradient descent method.

[0035] Preferably, step S4 specifically includes:

[0036] 4.1 Obtain meteorological simulation series under climate change scenarios based on a global climate model ensemble and quantile bias correction methods;

[0037] 4.2 Calculate the difference between the global climate model output data and the meteorological and hydrological reanalysis data provided by ERA5-Land at each quantile of 0.01-0.99, and remove this difference from the global climate model output quantile of each future scenario to obtain the future calibrated global climate model prediction dataset;

[0038] 4.3 Correct the air temperature as follows: T adj,d =T GCM,d +(T obs,Q -T GCM,ref,Q ), the same method is used to correct specific humidity, relative humidity, wind speed, shortwave radiation and longwave radiation; the precipitation is corrected as follows: P adj,d =P GCM,d ×(P obs,Q / P GCM,ref,Q ), the same method is used to correct snowfall and runoff depth;

[0039] Where T and P represent temperature and precipitation, respectively, adj represents the corrected series, obs represents the ERA5-Land observation data, ref and fut represent the historical reference period and the future prediction period, respectively, d represents daily data, and Q represents each quantile;

[0040] 4.4 Use the calibrated global climate model prediction dataset to drive the GR4J-9 hydrological model calibrated in step S2 to simulate the natural inflow runoff process under future scenarios;

[0041] 4.5 Use the calibrated global climate model prediction dataset to drive the long short-term memory (LSTM) model calibrated in step S3 to simulate runoff series affected by human activities under future scenarios;

[0042] 4.6 The natural inflow runoff and the runoff affected by human activities in the future scenario are superimposed to obtain the inflow daily runoff series in the future scenario. The calculation formula is as follows:

[0043]

[0044] Where, represents the natural inflow runoff simulation series under climate change, The runoff series represents the impact of human activities under climate change, represents the reservoir daily runoff series for future scenarios;

[0045] 4.7 Daily runoff series for future scenarios The probability change point analysis method is used to divide the flood season into stages.

[0046] Preferably, step S5 specifically includes:

[0047] 5.1 Based on the phased maximum sampling method, extract the main flood season flood Q of the daily runoff series under climate change simulated in step S4, and extract the non-main flood season W i (i=1, 2, ..., n), where n represents the number of non-main flood seasons obtained in step S4;

[0048] 5.2 Using the P-III distribution function to construct the marginal distribution function F of floods during the main flood season Q (q) and the marginal distribution function of floods in non-main flood seasons and the probability density function of floods in the main flood season f Q (q), probability density function of floods during non-main flood seasons The linear moment method is used to estimate the shape, scale and location parameters of the P-III distribution function;

[0049] 5.3 Use Copula function to construct the joint distribution function of floods in the main flood season and non-main flood season;

[0050] Let F(q,w1,...,w n ) represents the main flood season flood Q and n non-main flood season floods W i (i=1,2,...,n), then there exists an n-Copula function C:

[0051]

[0052] Where: θ is the correlation parameter of the Copula function;

[0053] 5.4 The OR return period is used as a metric for reservoir flood control standards and is defined as follows:

[0054] T(q,w1,...,w n )=1 / [1-F(q,w1,...,w n )];

[0055] Where, T(q,w1,...,w n ) is the joint return period, in years, F(q,w1,...,w n ) is the joint distribution function of floods in the main flood season and different non-main flood seasons, q is the flood magnitude in the main flood season, w i is the magnitude of floods in non-main flood seasons, where i = 1, 2, ..., n;

[0056] 5.5 Pass Q and W iThe joint probability density function value f(q,w1,w2,…,w n-1 ,w n ) to measure the relative probability of different peak quantity combinations; the larger the joint probability density function value, the greater the probability of the peak quantity combination; solve f(q,w1,w2,…,w n-1 ,w n ) The most likely combination of flood peak and flood volume is obtained when the maximum value of the following formula is satisfied, namely:

[0057]

[0058] The joint distribution function of Q and Wi (i=1,2,…,n) is expressed as F(q,w1,w2,…,w n-1 ,w n ) means that the expression is as follows:

[0059] F(q,w1,w2,…,w n-1 ,w n )=P(Q≤q,W1≤w1,...,W n ≤w n );

[0060] By taking the derivative, the corresponding probability density function is:

[0061] f(q,w1,w2,…,w n-1 ,w n )=dF(q,w1,w2,…,w n-1 ,w n ) / dqdw1,...,dw n ;

[0062] In the above formula, d represents the derivative symbol;

[0063] With the help of Copula function, the joint probability density function is expressed as:

[0064]

[0065] Where, is the density function of the Copula function; f X 、 X and Y respectively i Probability density function of (i=1,2,…,n);

[0066] 5.6 Use the Lagrange multiplier method to solve the most likely combination of staged design floods and construct the following solution equation:

[0067]

[0068] Joint probability density function f(q,w1,w2,…,w n-1 ,w n ) is continuous in the domain of n-dimensional space; C is the Copula function, c(·) is the density function of the Copula function, and given the joint return period T, the Lagrangian function is constructed as follows:

[0069]

[0070] For q,w1,…,w respectively n The Lagrangian function can be obtained by taking the partial derivative of λ and setting it to 0. All extreme points of ;

[0071] When n is any positive integer, the following system of equations can be obtained:

[0072]

[0073] Where: λ is the Lagrange multiplier; c=c(u,v1,v2,...,v n-1 ,v n ), p T =1-1 / T; f Q '、 (i=1,2,...,n) are the derivative functions of the corresponding probability density functions.

[0074] Preferably, step S6 specifically includes:

[0075] 6.1 The maximum daily average dew point temperature for each global climate model during the historical period 1985-2014 was calculated using the daily average temperature and daily average relative humidity data from each global climate model.

[0076] 6.2 Substitute the daily average temperature into the Clausius-Clapeyron equation to obtain the saturated water vapor pressure, and then use the saturated water vapor pressure and relative humidity to obtain the actual water vapor pressure: e act =e sa RH,

[0077] Where: e sa is the saturated water vapor pressure obtained using the daily average temperature of the global climate model, RH is the daily average relative humidity output by the global climate model, and e act is the calculated actual water vapor pressure;

[0078] 6.3 e actSubstitute the Clausius-Clapeyron equation to calculate the change trend Temptrend of the maximum daily average dew point temperature in each historical period of each global climate model, and calculate the change trend T of the average daily average temperature in each historical period of each global climate model. 2m trend;

[0079] 6.4 Integrate data from multiple global climate models to obtain the trend of maximum daily average dew point temperature changes in multiple historical periods, the trend of annual average daily average temperature changes, and the design flood WA for future periods. Fut1 The emergent constraint model constructed by the paired combination is as follows: WA Fut1 =a1·log(|Temptrend|)+a2·(T 2m trend) 2 +b;

[0080] Wherein, a1, a2 and b represent the parameters of the emergence constraint model; the least square method is used to solve the parameters a1, a2 and b of the emergence constraint model;

[0081] 6.5 Based on the ERA5-Land reanalysis dataset, the changing trend of the maximum daily dew point temperature in each year during the period 1985-2014 was calculated, and the changing trend of the average daily temperature in each year was calculated. The trend term was substituted into the constructed emergence constraint model, as follows: WA Fut =a1·log(|OBStrend|)+a2·(OBS 2m trend) 2 +b;

[0082] In the formula, WA Fut represents the corrected future phased design flood, OBStrend represents the trend of the annual maximum daily average dew point temperature obtained from the ERA5-Land dataset, and OBS 2m trend represents the changing trend of the annual average daily temperature obtained from the ERA5-Land dataset.

[0083] Compared with the existing technology, the present invention has the following beneficial effects: the present invention fully considers the combined impact of climate change and reservoir regulation, and adopts the emergence constraint method to reduce the uncertainty of the estimation of staged design floods; in addition, when deriving the joint design value of floods under different joint return periods, the present invention fully considers the inherent correlation of staged floods, conforms to the objective laws of flood events, and provides an important and highly operational reference basis for deriving staged design floods under changing environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] Figure 1This is a flow chart of a method for estimating floods by phased design under the influence of climate change and reservoir storage according to an embodiment of the present invention;

[0085] Figure 2 Schematic diagram of the change in the correlation coefficient between the reservoir inflow and the water flow with different lag times according to an embodiment of the present invention;

[0086] Figure 3 A schematic diagram of the marginal distribution function of floods during the main flood season according to an embodiment of the present invention;

[0087] Figure 4 A schematic diagram of the structure of a flood prediction system with phased design according to an embodiment of the present invention;

[0088] Figure 5 A schematic structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0089] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the embodiments described are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0090] In the description of this application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the described features. In the description of this application, "plurality" means two or more, unless otherwise specifically defined.

[0091] like Figures 1 to 5 As shown, this embodiment provides a method for estimating phased design floods under the influence of climate change and reservoir regulation. First, meteorological and hydrological data and human water use data are collected to deduce the restored natural inflow runoff series, and the GR4J-9 hydrological model is calibrated to realize the natural inflow runoff simulation; then, a long short-term memory model LSTM is constructed to simulate the influence of human activities such as reservoir regulation, and the GR4J-9 model is driven by the calibrated global climate model output data to obtain the inflow daily runoff series under climate change; further, the probabilistic change point analysis method is used to divide the flood season into stages, and a phased design flood calculation model for future scenarios is established based on the Copula function to deduce the most likely phased design flood; finally, an emergence constraint model based on dew point temperature is constructed to obtain the calibrated phased design flood results. The specific process is detailed in [1]. Figure 1 .

[0092] The steps are as follows:

[0093] S1. Collect daily runoff data for reservoirs, including engineering characteristics and operational data, and extract meteorological and hydrological reanalysis data, water diversion data, industrial and agricultural water use data, and global climate model output data for the reservoir basin.

[0094] S2. Based on the water balance method, the restored natural inflow runoff series is obtained and the GR4J-9 hydrological model is calibrated to simulate the natural inflow runoff.

[0095] S3. Based on natural inflow runoff, actual inflow runoff, and meteorological and hydrological reanalysis data, a long short-term memory (LSTM) model was constructed to simulate the impact of human activities on runoff and its parameters were calibrated.

[0096] S4. Based on a global climate model ensemble and quantile bias correction methods, a meteorological simulation series under climate change is obtained. This is then used to drive the GR4J-9 hydrological model and the long short-term memory (LSTM) model to obtain a series of daily inflows under climate change and stratify the flood season.

[0097] S5. Based on the daily runoff series obtained in step S4, a Copula function is used to establish a staged design flood calculation model for future scenarios to derive the most likely staged design flood;

[0098] S6. Based on the most likely staged design flood value derived in step S5 and the global climate model output data, an emergence constraint model based on dew point temperature is constructed to obtain a corrected staged design flood result.

[0099] The details are as follows:

[0100] Step S1 specifically includes the following steps:

[0101] Obtain the reservoir's daily runoff series from the reservoir management agency, collect data related to the reservoir's engineering characteristics and scheduling operations, and extract basic data on water diversion and industrial and agricultural water use in the basin where the reservoir is located.

[0102] For each grid point in the basin where the reservoir is located, daily meteorological and hydrological observation data from 1950 to 2022 were collected from the fifth-generation atmospheric reanalysis dataset (ERA5-Land) of the European Centre for Medium-Range Weather Forecasts, including daily maximum temperature, daily minimum temperature, daily mean temperature, air pressure, dew point temperature, snowfall, precipitation, runoff depth, shortwave radiation and longwave radiation.

[0103] Furthermore, the daily maximum temperature, daily minimum temperature, daily average temperature, daily average relative humidity, daily precipitation, daily snowfall, daily runoff depth, daily average specific humidity, daily shortwave radiation intensity, and daily longwave radiation were obtained for each global climate model for the historical period (1950-2014). For the future period (2015-2100), the daily temperature, relative humidity, precipitation, snowfall, runoff depth, specific humidity, shortwave radiation intensity, and longwave radiation were obtained for each global climate model under the shared socioeconomic pathway (SSP585). Specifically, 35 global climate models were selected.

[0104] Step S2 specifically includes the following steps:

[0105] First, based on the basic data such as the reservoir inflow and outflow daily runoff series obtained in step S1, the engineering characteristics and scheduling operation related data of the reservoir, the water diversion data of the basin where the reservoir is located, and the industrial and agricultural water use data, the water balance method is used to invert the reservoir inflow runoff series under natural conditions; and the Thiessen polygon method is used to derive the basin average daily series of meteorological variables of the ERA5-Land and global climate model output data.

[0106] The water balance method for restoring natural inflow runoff is a common technique in this field and will not be described in detail.

[0107] Furthermore, the GR4J-9 hydrological model considering the snowmelt module was calibrated based on the reservoir's natural inflow daily runoff series and ERA5-Land daily precipitation, daily maximum temperature, and daily minimum temperature data for the same period.

[0108] The GR4J hydrological model is a lumped conceptual hydrological model with only four parameters. This model has the characteristics of simple structure, fewer parameters, and high precision, and has been widely used. The model is mainly composed of two nonlinear reservoirs, namely the runoff reservoir and the confluence reservoir. Based on the GR4J model, this embodiment further considers the snowmelt module, wherein the snowmelt module is the CemaNeige module, to improve the accuracy of hydrological simulation, thereby forming the GR4J-9 hydrological model. The GR4J-9 hydrological model is a commonly used model in this field. Its input is a relatively short series of precipitation and temperature, and its output is the inflow runoff. Therefore, after calibrating the model, a long series of natural runoff processes can be simulated by inputting a long series of precipitation and temperature data.

[0109] This embodiment uses a composite hybrid evolution (SCE-UA) algorithm to optimize the hydrological model parameters. This algorithm is a global optimization algorithm that integrates the advantages of random search algorithm, simplex method, cluster analysis, and biological competitive evolution. It can effectively deal with problems such as insensitivity and non-convexity of the objective function reflection surface, and is not affected by local optimal points.

[0110] The daily precipitation, daily maximum temperature and daily minimum temperature data of ERA5-Land during the period of 1950-2022 were used to drive the calibrated GR4J-9 hydrological model to obtain the simulated natural inflow runoff series from 1950 to 2022.

[0111] Step S3 specifically includes the following steps:

[0112] First, based on the observed daily inflow series and the natural inflow series simulated in step S2, the inflow series affected by human activities is derived as follows:

[0113] Q hum =Q in -Q nat (1)

[0114] Where: Q in is the inflow runoff data obtained from the reservoir management agency, Q nat is the natural inflow runoff series simulated in step S2, Q hum It is the inflow runoff series affected by human activities.

[0115] Furthermore, a long short-term memory (LSTM) model was constructed to simulate the daily runoff series affected by human activities. The model requires daily average relative humidity and specific humidity as input, so the relative humidity and specific humidity are first derived.

[0116] The average daily temperature (T mean ) and dew point temperature (T dew ), deduce the relative humidity series; specifically, the atmospheric saturated water vapor pressure e can be obtained by the Clausius-Clapeyron equation and the given temperature T s :

[0117]

[0118] Where: T0 and e s0 are integration constants, which are 273.16K and 611Pa, respectively. v is the latent heat of vaporization (2.5×10 6 J kg -1 ), R v The steam gas constant (461 J kg -1 K -1 ); relative humidity RH = e s (T dew ) / e s (T mean ), where T mean represents the average daily temperature, T dew Indicates the dew point temperature, the obtained T mean and Tdew Substitute the corresponding saturated water vapor pressure e into the atmosphere s , the relative humidity RH can be solved.

[0119] Specific humidity q is the ratio of water vapor mass to the total mass of the air mass. It is derived using ERA5-Land ground pressure p and dew point temperature. The formula is as follows:

[0120]

[0121] Furthermore, by calculating the Q hum Statistical analysis of the series and basin-averaged daily precipitation was performed to determine the lag time that affects the long short-term memory model (LSTM).

[0122] like Figure 2 As shown in the figure, a schematic diagram of the change in the correlation coefficient between the reservoir inflow runoff and the precipitation at different lag times is given; the correlation coefficient between runoff and precipitation generally decreases gradually with the extension of the lag time; further, a correlation threshold that meets the characteristics of the underlying surface of the study basin is selected to determine the simulated runoff duration of the long short-term memory model LSTM; for example, 0.5 can be taken.

[0123] Based on the data obtained in the above steps, a long short-term memory (LSTM) model with a three-layer neural network architecture was constructed. The long short-term memory (LSTM) model was used to calibrate the LSTM model, with the average daily meteorological data of the ERA5-Land watershed as the explanatory variable and Qhum as the target variable as input.

[0124] This embodiment constructs a long short-term memory model LSTM with a three-layer neural network architecture to generalize the regulating and storage effects of dams, reservoirs or water diversion projects on watersheds and improve the accuracy of hydrological simulations. This embodiment uses a neural network interval simulation mean method to independently run the neural network model multiple times and take the average value as the final simulation result to reduce uncertainty.

[0125] To address the exploding and vanishing gradient problems of nonlinear autoregressive exogenous input (NARX) dynamic neural networks during deep learning (hidden layers ≥ 2), the Long Short-Term Memory (LSTM) model enhances the long-term memory capacity of the NARX neural network by introducing storage units—input gates, forget gates, internal feedback connections, and output gates—into the hidden layers of the NARX neural network. These units selectively remember current information or forget past information (such as the rainfall-runoff mapping). In short, the LSTM model replaces each hidden layer in the NARX dynamic neural network with a storage unit with memory capabilities, referred to as an LSTM unit. Its input and output layers are identical to those of the NARX dynamic neural network.

[0126] The average daily meteorological data of the ERA5-Land watershed is used as the explanatory variable, and Qhum is used as the target variable as input. After calibrating the LSTM model, the runoff series affected by human activities is simulated using the calibrated long short-term memory model LSTM. The simulation equation can be expressed as follows:

[0127] Q cor (t) = F LSTM [QM(t),QM(t-1),QM(t-2),…,QM(tN)]]; (4)

[0128] Where Q cor (t) represents the runoff series under the influence of human activities simulated at time t, QM(t) represents the input variables for calibrating the LSTM model, including the daily average temperature, daily average relative humidity, daily precipitation, daily snowfall, daily runoff depth, daily average specific humidity, daily shortwave radiation intensity, and daily longwave radiation of ERA5-Land; QM(t-1) represents the meteorological and hydrological variables at time t-1, N represents the lag determined by the LSTM model; F LSTM represents the calibrated Long Short-Term Memory (LSTM) model. To optimize the model parameters, the LSTM model is trained using the conventional technique in the field, the minimum batch gradient descent method.

[0129] Step S4 specifically includes the following steps:

[0130] Based on the global climate model ensemble and quantile bias correction method, a series of meteorological simulations under climate change scenarios are obtained;

[0131] The differences between the global climate model output data and the observed meteorological variables at each quantile (0.01-0.99) were calculated, and the differences were removed from the quantiles of the global climate model output of future scenarios to obtain the future revised global climate model prediction data set.

[0132] The air temperature (the same method is used for specific humidity, relative humidity, wind speed, shortwave radiation and longwave radiation) is corrected as follows:

[0133] T adj,d =T GCM,d +(T obs,Q -T GCM,ref,Q ) (5)

[0134] The correction for precipitation (snowfall and runoff depth are done in the same way) is as follows:

[0135] P adj,d =P GCM,d ×(P obs,Q / P GCM,ref,Q ) (6)

[0136] where T and P represent air temperature (specific humidity, relative humidity, wind speed, shortwave radiation, and longwave radiation) and precipitation (as well as snowfall and runoff depth), respectively; adj represents the corrected series; obs represents the ERA5-Land observational data; ref and fut represent the historical reference period and the future forecast period, respectively; d represents daily data; and Q represents each quantile.

[0137] Then, the calibrated global climate model prediction dataset is used to drive the GR4J-9 hydrological model calibrated in step S2 to simulate the natural inflow runoff process under future scenarios.

[0138] The calibrated global climate model prediction dataset is used to drive the long short-term memory (LSTM) model calibrated in step S3 to simulate the runoff series affected by human activities in the future scenario. Furthermore, the natural inflow runoff and the runoff affected by human activities in the future scenario are superimposed to obtain the inflow daily runoff series in the future scenario. The calculation formula is as follows:

[0139]

[0140] Where, represents the natural inflow runoff simulation series under climate change, represents the runoff series affected by human activities under climate change, Represents the daily runoff series into the reservoir for future scenarios.

[0141] Finally, for the daily runoff series under future climate change The flood season is divided into stages using a probability change point analysis method; the probability change point analysis method is a conventional technique in this field and will not be described in detail.

[0142] Step S5 specifically includes the following steps:

[0143] Based on the flood season stage results in step S4, the stage maximum sampling method is used to obtain the main flood season and non-main flood season floods under the climate change scenario. Based on the Copula function, a staged design flood calculation model for future scenarios is established, and the multivariate same-frequency combination and the most likely staged design flood under the climate change scenario are derived as follows:

[0144] Step 5.1: Based on the maximum sampling method, extract the flood Q of the reservoir daily runoff series under climate change simulated in step S4 during the main flood season, and extract the non-main flood season W. i (i=1, 2, ..., n), where n represents the number of non-main flood seasons obtained in step S4;

[0145] Step 5.2: Use the P-III distribution function to construct the marginal distribution function F of floods during the main flood season. Q (q) and the marginal distribution function of floods in non-main flood seasons and the probability density function of floods in the main flood season fQ (q), probability density function of floods during non-main flood seasons The linear moment method is used to estimate the shape, scale and location parameters of the P-III distribution function. Figure 3 As shown in the figure, a schematic diagram of the marginal distribution function of floods during the main flood season is given.

[0146] Step 5.3: Use the Copula function to construct the joint distribution function of floods during the main flood season and non-main flood season;

[0147] Let F(q,w1,...,w n ) represents the main flood season flood Q and n non-main flood season floods W i (i=1,2,...,n), then there exists an n-Copula function C:

[0148]

[0149] Where: θ is the correlation parameter of the Copula function.

[0150] In this embodiment, the Gumbel-Hougaard (GH) Copula function in the Archimedean Copula function family is used as the joint distribution function to construct the joint distribution function F(q, w1, ..., w n ).

[0151] Step 5.4, define flood protection standards based on OR (or) return period T;

[0152] The OR return period is used as a measurement indicator for reservoir flood control standards, and its definition is as follows:

[0153] T(q,w1,...,w n )=1 / [1-F(q,w1,...,w n )]; (9)

[0154] Where, T(q,w1,...,w n ) is the joint return period, in years, F(q,w1,...,w n ) is the joint distribution function of floods in the main flood season and different non-main flood seasons, q is the flood magnitude in the main flood season, w i The magnitude of floods during non-main flood seasons.

[0155] Step 5.5, through the joint probability density function value f(q,w1,w2,…,w n-1 ,w n) is used to measure the relative likelihood of different peak volume combinations. The larger the joint probability density function value, the greater the likelihood of the peak volume combination. To obtain the most likely combination of flood peaks and volumes, it is necessary to solve f(q, w1, w2, …, w n-1 ,w n ) satisfies the maximum value of the following formula, namely:

[0156]

[0157] Q and W i The joint distribution function of (i=1,2,…,n) is F(q,w1,w2,…,w n-1 ,w n ) means that the expression is as follows:

[0158] F(q,w1,w2,…,w n-1 ,w n )=P(Q≤q,W1≤w1,...,W n ≤w n ) (11)

[0159] By taking the derivative, the corresponding probability density function is:

[0160] f(q,w1,w2,…,w n-1 ,w n )=dF(q,w1,w2,…,w n-1 ,w n ) / dqdw1,...,dw n (12)

[0161] In the above formula, d represents the derivative symbol.

[0162] With the help of Copula function, the joint probability density function can be expressed as:

[0163]

[0164] Where, is the density function of the Copula function; f X 、 X and Y respectively i The probability density function of (i=1,2,…,n).

[0165] In step 5.6, the Lagrange multiplier method is used to solve the most likely combination of staged design floods, and the following solution equation is constructed:

[0166]

[0167] Joint probability density function f(q,w1,w2,…,w n-1 ,wn ) is continuous in the domain of n-dimensional space; C is the Copula function, c(·) is the density function of the Copula function, and given the joint return period T, the Lagrangian function is constructed as follows:

[0168]

[0169] For q,w1,…,w respectively n The Lagrangian function can be obtained by taking the partial derivative of λ and setting it to 0. All extreme points of .

[0170] When n is any positive integer, the following system of equations can be obtained:

[0171]

[0172] Where λ is the Lagrange multiplier; c=c(u,v1,v2,...,v n-1 ,v n ), p T =1-1 / T; f Q '、 are the derivatives of the corresponding probability density functions respectively.

[0173] The nonlinear equations of formula (16) above are the general formulas that should be satisfied by the most likely combination method of peak quantity under the joint return period level derived based on the Copula function. The present invention uses the peak quantity design results obtained by multivariable same-frequency combination calculation as the initial solution, and uses the harmonic mean Newton method for iterative solution to obtain the most likely combination of staged design flood (q*, w1*, w2*, ..., w n-1 *,w n *).

[0174] Step S6 specifically includes the following steps:

[0175] First, the maximum daily average dew point temperature of each global climate model in the historical period (1985-2014) is calculated, specifically using the daily average temperature and daily average relative humidity data of each global climate model.

[0176] The average daily temperature is substituted into the Clausius-Clapeyron equation to obtain the saturated water vapor pressure, and then the saturated water vapor pressure and relative humidity are used to obtain the actual water vapor pressure:

[0177] e act =e sa ·RH (17)

[0178] Where, e sais the saturated water vapor pressure obtained using the daily average temperature of the global climate model, RH is the daily average relative humidity output by the global climate model, and e act is the calculated actual water vapor pressure.

[0179] Further, e act Substituting the Clausius-Clapeyron equation, the daily average dew point temperature series for the historical period output by the global climate model is derived through numerical calculation methods. This method is a conventional technique in this field.

[0180] Furthermore, the variation trend of the maximum daily average dew point temperature in each historical period of each global climate model (denoted as Temptrend) is calculated; and the variation trend of the average daily average temperature in each historical period of each global climate model (denoted as T 2m Furthermore, the data of 35 global climate models were integrated to obtain 35 sets of historical annual maximum daily average dew point temperature change trends, annual average daily average temperature change trends and future period stage design flood (denoted as WA Fut1 ) pairing combination, the constructed emergence constraint model is specifically as follows:

[0181] WA Fut1 =a1·log(|Temptrend|)+a2·(T 2m trend) 2 +b (18)

[0182] Where a1, a2, and b represent the parameters of the emergence constraint model.

[0183] The least squares method was used to solve the parameters a1, a2, and b of the emergence constraint model. The trend of the maximum daily average dew point temperature during the period 1985-2014 was calculated based on the ERA5-Land reanalysis dataset, and the trend of the annual average daily temperature was calculated. The trend term was then substituted into the constructed emergence constraint model as follows:

[0184] WA Fut =a1·log(|OBStrend|)+a2·(OBS 2m trend) 2 +b (19)

[0185] In the formula, WA Fut represents the corrected future phased design flood, OBStrend represents the trend of the annual maximum daily average dew point temperature obtained from the ERA5-Land dataset, and OBS 2m trend represents the changing trend of the annual average daily temperature obtained from the ERA5-Land dataset.

[0186] Finally, the above method was used to obtain the corrected phased design flood results for the main flood season and non-main flood season, which were used to guide reservoir operation scheduling and integrated basin management.

[0187] It can be seen that existing methods often directly use global climate models to predict future flood processes, without considering human influences such as reservoir regulation, which is somewhat irrational. The present invention not only considers the combined impact of climate change and reservoir regulation, but also uses an emergence constraint method to reduce the uncertainty of staged design flood estimates. Furthermore, when deriving the combined design flood values ​​under different combined return periods, the present invention fully considers the inherent correlation of staged floods, conforming to the objective laws of flood events, and providing an important and highly operational reference for deriving staged design floods under changing environments.

[0188] The following describes the staged design flood estimation system provided by the present invention. The staged design flood estimation system described below and the staged design flood estimation method described above can be referenced to each other.

[0189] Figure 4 1 is a structural diagram of a staged design flood estimation system provided by an embodiment of the present invention, which includes an acquisition module 51, a training module 52, a correction module 53, a combination module 54, an establishment module 55 and a prediction module 56 connected in sequence.

[0190] Among them, the acquisition module 51 is used to collect meteorological and hydrological data to form a meteorological and hydrological data set; the training module 52 is used to determine the relative humidity and specific humidity at a set time, and train the watershed hydrological model and the long-short-term memory model based on the relative humidity and the specific humidity and the meteorological and hydrological data set; the correction module 53 is used to obtain the meteorological simulation data of the global climate model ensemble under the climate change scenario and correct the meteorological simulation data of the global climate model ensemble, and use the corrected meteorological simulation data of the global climate model ensemble to train the hydrological model and the long-short-term memory model to obtain the inflow runoff series under the future scenario; the combination module 54 is used to use the global climate model ensemble and the quantile deviation correction method to obtain the meteorological simulation series under climate change, and drive the GR4J-9 model and the long-short-term memory model to obtain the inflow runoff series under climate change and perform flood season phasing; the establishment module 55 is used to use the Copula function to establish a staged design flood calculation model for the future scenario and deduce the most likely staged design flood; the prediction module 56 is used to construct an emergence constraint model based on dew point temperature to obtain the corrected staged design flood results. It should be noted that the specific implementation of the staged design flood estimation system according to the embodiment of the present invention can refer to the specific implementation of the flood risk prediction method described above, and will not be described again here to avoid redundancy.

[0191] Figure 5An example of a physical structure diagram of an electronic device is provided. The electronic device may include: a processor 810, a communications interface 820, a memory 830, and a communications bus 840, wherein the processor 810, the communications interface 820, and the memory 830 communicate with each other via the communications bus 840. The processor 810 may call the logic instructions in the memory 830 to execute a phased design flood prediction method, which includes: collecting the inflow and outflow runoff observation data of the reservoir, collecting the engineering characteristics and scheduling operation related information of the reservoir, and extracting the meteorological data, water diversion data, industrial and agricultural water use data, and global climate model output data of the basin where the reservoir is located; obtaining the restored natural inflow runoff series based on the water balance method, and calibrating the GR4J hydrological model to realize the natural inflow runoff simulation; constructing a long-short-term memory model based on the natural inflow runoff, actual inflow runoff, and meteorological data to simulate the impact of human activities such as reservoir regulation on runoff. effect, and the minimum batch gradient descent method is used for parameter calibration; based on the global climate model ensemble and the quantile bias correction method, the meteorological simulation series under climate change is obtained, and the GR4J-9 model and the long short-term memory model are driven to obtain the reservoir runoff series under climate change and perform flood season staging; based on the daily runoff series of future scenarios, a staged design flood calculation model for future scenarios is established based on the Copula function, and the most likely staged design flood is deduced; based on the most likely staged design flood value and the global climate model output series, an emergence constraint model based on dew point temperature is constructed to obtain the corrected staged design flood results.

[0192] In addition, the logic instructions in the above-mentioned memory 830 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0193] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0194] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, or of course, by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the existing technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or certain parts of the embodiments.

[0195] In summary, the staged design flood estimation method, system and electronic equipment provided by the present invention, by combining global climate models, river basin hydrological models, most likely combination methods, deviation correction methods and emergence constraint methods, provide an important and highly operational reference basis for river basin flood risk assessment and early warning under climate change scenarios, and provide engineering reference value for responding to future climate disasters and scientifically formulating emission reduction strategies.

[0196] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for estimating staged design floods under the influence of climate change and reservoir regulation, characterized in that: The steps include: S1. Collect daily runoff data for reservoirs, including engineering characteristics and operational data, and extract meteorological and hydrological reanalysis data, water diversion data, industrial and agricultural water use data, and global climate model output data for the reservoir basin. S2. Based on the water balance method, the restored natural inflow runoff series is obtained and the GR4J-9 hydrological model is calibrated to simulate the natural inflow runoff. S3. Based on natural inflow, actual inflow, and meteorological and hydrological reanalysis data, a long-short-term memory model is constructed to simulate the impact of human activities on runoff and parameter calibration is performed. S4. Based on a global climate model ensemble and quantile bias correction methods, a meteorological simulation series under climate change is obtained. This is then used to drive the GR4J-9 hydrological model and the long short-term memory (LSTM) model to obtain a series of daily inflows under climate change and stratify the flood season. S5. Based on the daily runoff series obtained in step S4, a Copula function is used to establish a staged design flood calculation model for future scenarios to derive the most likely staged design flood; S6. Based on the most likely staged design flood value derived in step S5 and the global climate model output data, an emergence constraint model based on dew point temperature is constructed to obtain a corrected staged design flood result.

2. A method for estimating design floods in stages under the influence of climate change and reservoir regulation as claimed in claim 1, characterized in that: Step S1 specifically includes: 1.1 Obtain the reservoir's daily runoff series from the reservoir management agency, collect data on the reservoir's engineering characteristics and operation, and extract basic data on water diversion and industrial and agricultural water use in the basin where the reservoir is located; 1.2 For each grid point in the reservoir basin, daily meteorological and hydrological reanalysis data were collected from the fifth-generation atmospheric reanalysis dataset ERA5-Land of the European Centre for Medium-Range Weather Forecasts for the period 1950–2022. Specifically, these data included: daily maximum and minimum temperatures, daily mean temperatures, air pressure, dew point temperature, snowfall, precipitation, runoff depth, shortwave radiation, and longwave radiation. 1.3 Obtain daily maximum temperature, daily minimum temperature, daily average temperature, daily average relative humidity, daily precipitation, daily snowfall, daily runoff depth, daily average specific humidity, daily shortwave radiation intensity, and daily longwave radiation for the historical period 1950-2014 under multiple global climate models; 1.4 Obtain daily temperature, relative humidity, precipitation, snowfall, runoff depth, specific humidity, shortwave radiation intensity, and longwave radiation for multiple global climate models for the future period 2015–2100 under the shared socioeconomic pathway of SSP585.

3. A method for estimating design floods in stages under the influence of climate change and reservoir regulation as claimed in claim 2, characterized in that: Thirty-five global climate models were selected.

4. A method for estimating design floods in stages under the influence of climate change and reservoir regulation as claimed in claim 2, characterized in that: Step S2 specifically includes: 2.1 Based on the reservoir inflow and outflow daily runoff series obtained in step S1, the reservoir's engineering characteristics and operation-related data, the water diversion data of the reservoir's basin, and the basic data of industrial and agricultural water use, the water balance method is used to invert the reservoir inflow runoff series under natural conditions; 2.2 The Thiessen polygon method is used to derive the basin-mean daily series of meteorological variables from the ERA5-Land reanalysis dataset and global climate model output data; 2.3 Based on the reservoir runoff series under natural conditions and the daily precipitation, daily maximum temperature and daily minimum temperature data of ERA5-Land during the same period, the GR4J-9 hydrological model with snowmelt module was calibrated; 2.4 The daily precipitation, daily maximum temperature, and daily minimum temperature data of ERA5-Land during the period 1950-2022 were used to drive the calibrated GR4J-9 hydrological model to obtain the simulated natural inflow runoff series from 1950 to 2022.

5. The method for estimating design floods in stages under the influence of climate change and reservoir regulation as claimed in claim 1, characterized in that: Step S3 specifically includes: 3.1 Based on the observed daily runoff series and the natural runoff series simulated in step S2, the runoff series affected by human activities is derived as follows: Q hum =Q in -Q nat ; Where Q in is the inflow runoff data obtained from the reservoir management agency, Q nat is the natural inflow runoff series simulated in step S2, Q hum The inflow runoff series is affected by human activities; 3.2 Derived the daily average relative humidity and specific humidity, and constructed a long short-term memory (LSTM) model to simulate the inflow runoff series affected by human activities; 3.3 By calculating Q obtained in step 3.1 hum Statistical analysis of the series and basin-averaged daily precipitation was performed to determine the lag time that affects the long short-term memory model (LSTM); 3.4 Based on the data obtained in the above steps, a long short-term memory model LSTM with a three-layer neural network architecture was constructed, with the average daily meteorological data of the ERA5-Land basin as the explanatory variable and Q hum Taking the target variable as input, calibrate the long short-term memory model LSTM; 3.5 The runoff series affected by human activities is simulated by the calibrated long short-term memory model LSTM. The simulation equation is: Q cor (t) = F LSTM [QM(t),QM(t-1),QM(t-2),…,QM(tN)]]; Where Q cor (t) represents the runoff series under the influence of human activities simulated at time t; QM(t) represents the input variables for calibrating the long short-term memory model LSTM, including the daily average temperature, daily average relative humidity, daily precipitation, daily snowfall, daily runoff depth, daily average specific humidity, daily shortwave radiation intensity, and daily longwave radiation of ERA5-Land; QM(t-1) represents the meteorological and hydrological variables at time t-1, and N represents the lag determined by the long short-term memory model LSTM; F LSTM Represents the calibrated long short-term memory model LSTM.

6. A method for estimating design floods in stages under the influence of climate change and reservoir regulation as claimed in claim 4, characterized in that: The long short-term memory model LSTM is trained using the minimum batch gradient descent method.

7. The method for estimating design floods in stages under the influence of climate change and reservoir regulation as claimed in claim 1, characterized in that: Step S4 specifically includes: 4.1 Obtain meteorological simulation series under climate change scenarios based on a global climate model ensemble and quantile bias correction methods; 4.2 Calculate the difference between the global climate model output data and the meteorological and hydrological reanalysis data provided by ERA5-Land at each quantile of 0.01-0.99, and remove this difference from the global climate model output quantile of each future scenario to obtain the future calibrated global climate model prediction dataset; 4.3 Correct the air temperature as follows: T adj,d =T GCM,d +(T obs,Q -T GCM,ref,Q ), the same method is used to correct specific humidity, relative humidity, wind speed, shortwave radiation and longwave radiation; the precipitation is corrected as follows: P adj,d =P GCM,d ×(P obs,Q / P GCM,ref,Q ), the same method is used to correct snowfall and runoff depth; Where T and P represent temperature and precipitation, respectively, adj represents the corrected series, obs represents the ERA5-Land observation data, ref and fut represent the historical reference period and the future prediction period, respectively, d represents daily data, and Q represents each quantile; 4.4 Use the calibrated global climate model prediction dataset to drive the GR4J-9 hydrological model calibrated in step S2 to simulate the natural inflow runoff process under future scenarios; 4.5 Use the calibrated global climate model prediction dataset to drive the long short-term memory (LSTM) model calibrated in step S3 to simulate runoff series affected by human activities under future scenarios; 4.6 The natural inflow runoff and the runoff affected by human activities in the future scenario are superimposed to obtain the inflow daily runoff series in the future scenario. The calculation formula is as follows: Where, represents the natural inflow runoff simulation series under climate change, represents the runoff series affected by human activities under climate change, represents the reservoir daily runoff series for future scenarios; 4.7 Daily runoff series for future scenarios The probability change point analysis method is used to divide the flood season into stages.

8. The method for estimating design floods in stages under the influence of climate change and reservoir regulation as claimed in claim 1, characterized in that: Step S5 specifically includes: 5.1 Based on the phased maximum sampling method, extract the main flood season flood Q of the daily runoff series under climate change simulated in step S4, and extract the non-main flood season flood W i (i=1, 2, ..., n), where n represents the number of non-main flood seasons obtained in step S4; 5.2 Using the P-III distribution function to construct the marginal distribution function F of floods during the main flood season Q (q) and the marginal distribution function of floods in non-main flood seasons and the probability density function of floods in the main flood season f Q (q), probability density function of floods during non-main flood seasons The linear moment method is used to estimate the shape, scale and location parameters of the P-III distribution function; 5.3 Use Copula function to construct the joint distribution function of floods in the main flood season and non-main flood season; Let F(q,w1,...,w n ) represents the main flood season flood Q and n non-main flood season floods W i (i=1,2,...,n), then there exists an n-Copula function C: Where: θ is the correlation parameter of the Copula function; 5.4 The OR return period is used as a metric for reservoir flood control standards and is defined as follows: T(q,w1,...,w n )=1 / [1-F(q,w1,...,w n )]; Where, T(q,w1,...,w n ) is the joint return period, in years, F(q,w1,...,w n ) is the joint distribution function of floods in the main flood season and different non-main flood seasons, q is the flood magnitude in the main flood season, w i is the magnitude of floods in non-main flood seasons, where i = 1, 2, ..., n; 5.5 Pass Q and W i The joint probability density function value f(q,w1,w2,…,w n-1 ,w n ) to measure the relative probability of different peak quantity combinations; the larger the value of the joint probability density function, the greater the probability of the peak quantity combination; solve f(q,w1,w2,…,w n-1 ,w n ) The most likely combination of flood peak and flood volume is obtained when the maximum value of the following formula is satisfied, namely: Q and W i The joint distribution function of (i=1,2,…,n) is F(q,w1,w2,…,w n-1 ,w n ) means that the expression is as follows: F(q,w1,w2,…,w n-1 ,w n )=P(Q≤q,W1≤w1,...,W n ≤w n ); By taking the derivative, the corresponding probability density function is: f(q,w1,w2,…,w n-1 ,w n )=dF(q,w1,w2,…,w n-1 ,w n ) / dqdw1,...,dw n ; In the above formula, d represents the derivative symbol; With the help of Copula function, the joint probability density function is expressed as: Where, is the density function of the Copula function; f X 、 X and Y respectively i Probability density function of (i=1,2,…,n); 5.6 Use the Lagrange multiplier method to solve the most likely combination of staged design floods and construct the following solution equation: Joint probability density function f(q,w1,w2,…,w n-1 ,w n ) is continuous in the domain of n-dimensional space; C is the Copula function, c(·) is the density function of the Copula function, and given the joint return period T, the Lagrangian function is constructed as follows: For q,w1,…,w respectively n The Lagrangian function can be obtained by taking the partial derivative of λ and setting it to 0. All extreme points of ; When n is any positive integer, the following system of equations is obtained: Where: λ is the Lagrange multiplier; c=c(u,v1,v2,...,v n-1 ,v n ), p T =1-1 / T; f Q '、 are the derivatives of the corresponding probability density functions respectively.

9. The method for estimating design floods in stages under the influence of climate change and reservoir regulation as claimed in claim 1, characterized in that: Step S6 specifically includes: 6.1 The maximum daily average dew point temperature for each global climate model during the historical period 1985-2014 was calculated using the daily average temperature and daily average relative humidity data from each global climate model. 6.2 Substitute the daily average temperature into the Clausius-Clapeyron equation to obtain the saturated water vapor pressure, and then use the saturated water vapor pressure and relative humidity to obtain the actual water vapor pressure: e act =e sa RH, Where: e sa is the saturated water vapor pressure obtained using the daily average temperature of the global climate model, RH is the daily average relative humidity output by the global climate model, and e act is the calculated actual water vapor pressure; 6.3 e act Substitute the Clausius-Clapeyron equation to calculate the change trend Temptrend of the maximum daily average dew point temperature in each historical period of each global climate model, and calculate the change trend T of the average daily average temperature in each historical period of each global climate model. 2m trend; 6.4 Integrate data from multiple global climate models to obtain the trend of maximum daily average dew point temperature changes in multiple historical periods, the trend of annual average daily average temperature changes, and the design flood WA for future periods. Fut1 The emergent constraint model constructed by the paired combination is as follows: WA Fut1 =a1·log(|Temptrend|)+a2·(T 2m trend) 2 +b; Wherein, a1, a2 and b represent the parameters of the emergence constraint model; the least square method is used to solve the parameters a1, a2 and b of the emergence constraint model; 6.5 Based on the ERA5-Land reanalysis dataset, the changing trend of the maximum daily dew point temperature in each year during the period 1985-2014 was calculated, and the changing trend of the average daily temperature in each year was calculated. The trend term was substituted into the constructed emergence constraint model, as follows: WA Fut =a1·log(|OBStrend|)+a2·(OBS 2m trend) 2 +b; In the formula, WA Fut represents the corrected future phased design flood, OBStrend represents the trend of the annual maximum daily average dew point temperature obtained from the ERA5-Land dataset, and OBS 2m trend represents the changing trend of the annual average daily temperature obtained from the ERA5-Land dataset.

Images