Method and system for estimating design floods considering climate and human activity non-stationarity

Abstract

The application discloses a kind of non-uniform design flood estimation method and system considering climate and human activity, comprising: obtaining the hydro-meteorological data of research basin and reservoir related data;Based on the data obtained in step 1, GAMLSS method and sliding window method are used to derive time-varying probability distribution model;Based on the annual average reliability, the non-uniform design flood values under two methods in the engineering design period T1-T2 are estimated according to the time-varying probability distribution model derived in step 2;Based on the actual data acquisition condition of basin, the design flood values of GAMLSS method and sliding window method are compared, and the non-uniform design flood value more suitable for the design requirement of research basin is selected, which provides reference for engineering design under changing environment.The application comprehensively considers the difference between GAMLSS method and sliding window method in determining non-uniform design flood value, and selects more reliable non-uniform design flood value according to actual basin condition, so that the design is more scientific and reasonable.

Description

Technical Field

[0001] This invention belongs to the technical field of hydraulic design, specifically relating to a non-uniform design flood estimation method and system that takes into account climate and human activities. Background Technology

[0002] Due to the influence of climate conditions and human activities, the probability distribution of hydrological variables and the design values ​​corresponding to a certain return period may change over time. Using design values ​​under consistent conditions will increase the risks of engineering construction or operation. The derivation of time-varying probability distributions can be achieved using statistical models or watershed hydrological models. Different models have different methodologies and input information. Therefore, in practical engineering, it is essential to select reliable non-consistent design flood results to guide the design of water conservancy projects under changing environments. Considering the increasingly severe disturbances to flood sequences caused by climate change and human activities, constructing a non-consistent design flood estimation method and system that considers climate and human activities for calculating engineering hydrological design values ​​under non-consistent conditions is of great significance for the design of water conservancy projects under changing environments. Summary of the Invention

[0003] One objective of this invention is to address the shortcomings of existing technologies by providing a non-consistent design flood estimation method that takes into account climate and human activities. This method comprehensively considers the differences in determining non-consistent design flood values ​​using different models and selects a more reliable non-consistent design flood value based on the actual conditions of the watershed, thereby making the design more scientific and reasonable.

[0004] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0005] A method for estimating non-uniform design floods that takes into account climate and human activities includes the following steps:

[0006] Step 1: Obtain hydrological and meteorological data and reservoir-related data for the study watershed;

[0007] Step 2: Based on the data obtained in Step 1, the time-varying probability distribution model is derived using the GAMLSS method and the sliding window method, respectively;

[0008] Step 3: Based on the annual average reliability, estimate the non-consistent design flood values ​​under the two methods within the engineering design life T1~T2 according to the time-varying probability distribution model derived in Step 2;

[0009] Step 4: Based on the actual data obtained in the watershed, compare the non-consistent design flood values ​​of the GAMLSS method and the sliding window method, and select the non-consistent design flood value that is more in line with the watershed conditions, so as to provide a reference for engineering design under changing environments.

[0010] Furthermore, the data obtained in step 1 includes:

[0011] The study investigates the historical runoff of the basin, meteorological data from the entire historical period of the basin, future meteorological data simulated by climate models, as well as the controlled area and total storage capacity of large and medium-sized reservoirs within the basin, and the planning and construction of reservoir projects in the basin plan.

[0012] Furthermore, the method for obtaining future meteorological data in step 1 includes: using the CMIP6 global climate model GCM to simulate meteorological data, and then downscaling it to obtain future rainfall data for the basin.

[0013] Furthermore, methods for deriving time-varying probability distribution models using the GAMLSS method include:

[0014] Based on the collected basin meteorological data and reservoir data, the summer precipitation anomaly SPA, the annual precipitation anomaly TPA, and the reservoir index RI were calculated. The calculated summer precipitation anomaly SPA and annual precipitation anomaly TPA were used as basin climate covariates, and the reservoir index RI was used as a basin human activity covariate.

[0015] Constructing a non-consistent probability distribution model based on the GAMLSS model And the maximum likelihood method is used to estimate its distribution parameters;

[0016] Assuming that each observation is independent, the log-likelihood function is expressed as follows under consistent and inconsistent conditions:

[0017]

[0018]

[0019] In the formula, n is the measured length of the flood sequence; The flood variable Y t The probability density function y t This represents the observed value of the flood variable at a certain time t; These are covariates, including climate covariate X1 and human activity covariate X2. It is the model parameter vector, under the consistency condition. θ is a constant, under non-uniform conditions t =(μ t , σ t , ξ t ), θ t μ represents the model distribution parameters at a certain time t. t σ represents the position parameter at a certain time t. t ξ represents the scale parameter at a given time t. t The shape parameter represents a certain time t;

[0020] That is, the functional relationship between the distribution parameters and the covariates is: ;

[0021] Substituting the future climate covariate and the human activity covariate into the functional relationship between the distribution parameters and the covariates mentioned above, we obtain the probability distribution parameters for future years. The obtained probability distribution parameters Substituting into the above probability distribution model Thus, the time-varying probability distribution of floods within the engineering design life is obtained. .

[0022] Furthermore, methods for deriving time-varying probability distribution models using the sliding window method include:

[0023] The watershed with significant variations in the flood sequence is divided into two periods: the natural period and the variable period. Hydrological runoff data from both periods are used to calibrate and validate the hydrological model.

[0024] The meteorological data output from the climate model GCM was downscaled and then input into the hydrological model that had been validated during the aforementioned period of change to simulate the future daily runoff sequence of the watershed.

[0025] Selecting a sliding window length of m years, for the future daily runoff sequence, taking each year within the engineering design life as the window center, select the sequences of each year before and after the window center for m / 2 years to form the flood sample within that window. Distribution fitting can then yield the probability distribution corresponding to each window. This leads to the time-varying probability distribution of floods within the engineering design life. .

[0026] Furthermore, the formula for calculating the non-uniform design flood value in step 3 is as follows:

[0027]

[0028] In the formula, p is the probability of exceeding the limit, and T1 and T2 are the start and end years of the project. This represents the time-varying probability distribution derived in step 2.

[0029] Furthermore, in step 4, when determining the non-consistent design flood value for the study area, the existence of historical flood data and non-consistent influencing factor data for the study watershed is first considered. If they exist, the design value result of the GAMLSS method is selected as the non-consistent design flood value; if they do not exist, the uncertainty of the non-consistent design value corresponding to different exceedance probabilities of the GAMLSS method and the sliding window method is compared, and the design value result with smaller uncertainty among the two methods is selected.

[0030] Another object of the present invention is to provide a system for implementing the above-described non-uniform design flood estimation method that takes into account climate and human activities, comprising:

[0031] The data acquisition module is used to acquire hydrological and meteorological data of the research watershed, as well as reservoir-related data;

[0032] The time-varying probability distribution model derivation module is used to derive the time-varying probability distribution model based on the acquired data, using the GAMLSS method and the sliding window method respectively.

[0033] The non-consistent design flood value calculation module is used to estimate the non-consistent design flood values ​​under two methods within the engineering design life T1~T2 based on the annual average reliability and the derived time-varying probability distribution model.

[0034] The module for determining the non-consistent design flood value of the study watershed is used to compare the design flood values ​​of the GAMLSS method and the sliding window method based on the actual data obtained in the watershed, and select the non-consistent design flood value that better meets the design requirements of the study watershed.

[0035] An electronic device includes a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the above method.

[0036] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0037] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention comprehensively considers the influence of different model structures and input information on the estimated design flood value, and compares the differences between the GAMLSS method and the sliding window method in determining the non-consistent design flood value. Among them, the GAMLSS method has better applicability in handling the nonlinear relationship and inconsistency of the sequence. If the covariates are selected appropriately, the model construction based on long sequence data will also be more reliable. The sliding window method analyzes the changing trend of data by continuously moving the time window, which can better adapt to the dynamic changes of data. However, both methods have certain uncertainties in terms of model structure and parameter estimation. Therefore, in order to provide the accuracy of the results, this invention compares the two methods and selects the more suitable method to calculate the non-consistent design flood value according to the actual situation of the watershed. This can provide more reliable design results for design flood estimation and water conservancy engineering design under changing environments. In addition, against the background of increasingly significant impacts of climate change and human activities, the calculated design flood results are very important for flood control and disaster reduction, and can be used for flood estimation and risk management of different regions and different types of rivers. Attached Figure Description

[0038] Figure 1This is a flowchart of the non-uniform design flood estimation method and system considering climate and human activities in an example of the present invention;

[0039] Figure 2 This is a quantile curve of the annual maximum daily flow corresponding to the consistent and optimal non-consistent probability distribution models in the example of this invention;

[0040] Figure 3 This is the future climate covariate sequence of the Wujiang River Basin in this invention example;

[0041] Figure 4 This is the sequence of the maximum daily flow in the Wujiang River basin in future years, as described in this invention example.

[0042] Figure 5 These are the estimation results of non-consistent design flood values ​​and consistent design flood values ​​corresponding to the GAMLSS method and sliding window method in the examples of this invention;

[0043] Figure 6 The GAMLSS method and sliding window method in this invention example correspond to different probabilities corresponding to the 5%~95% quantile range of the non-consistent design flood value. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0045] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0046] The present invention will be further described below with reference to specific embodiments, but these are not intended to limit the scope of the invention.

[0047] This invention discloses a method for estimating non-uniform design floods that takes into account climate and human activities, comprising the following steps:

[0048] Step 1: Obtain hydrological and meteorological data and reservoir-related data for the study watershed; in this step, the collected data includes:

[0049] Collect and organize historical runoff and meteorological data of the basin, as well as information on the controlled area and total storage capacity of existing large and medium-sized reservoirs within the basin;

[0050] Download the meteorological data simulated by the CMIP6 Global Climate Model (GCM) (including three shared socioeconomic pathways: SSP1-2.6, SSP2-4.5, and SSP5-8.5), downscale them to obtain the daily rainfall for future periods in the basin, and calculate the future climate covariates of the basin.

[0051] Collect information on the construction and operation of future reservoir projects and corresponding reservoir information from watershed planning in order to calculate covariates of future human activities in the watershed.

[0052] Step 2: Based on the data obtained in Step 1, the time-varying probability distribution model is derived using the GAMLSS method and the sliding window method, respectively;

[0053] The methods for deriving time-varying probability distribution models using the GAMLSS method include:

[0054] (1) First, based on the collected basin meteorological data and reservoir data, calculate the precipitation anomaly factors: summer precipitation anomaly (SPA), annual precipitation anomaly (TPA), and reservoir impact factor indicators: reservoir index (RI).

[0055]

[0056]

[0057]

[0058] In the formula, This is the average precipitation during the summer months (May-September), in mm / day; It is the multi-year average of summer precipitation, in mm / day; It is the annual average precipitation, in mm / day; It is the multi-year average annual precipitation, mm / day; N is the total number of reservoirs upstream of the hydrological station; A i It is the controlled area of ​​the i-th upstream reservoir, in km². 2 A T It refers to the area controlled by the hydrological station, in km². 2 V i The total storage capacity of the i-th reservoir is m. 3 ; It is the sum of the total storage capacity of all reservoirs upstream of the hydrological station, m 3 ;

[0059] The calculated summer precipitation anomaly (SPA) and annual precipitation anomaly (TPA) were used as covariates of watershed climate, and the reservoir index (RI) was used as a covariate of human activities in the watershed.

[0060] (2) Construct a non-consistent probability distribution model based on the GAMLSS model (the distribution types include 6 two-parameter distributions: Weibull distribution, Gumbel distribution, Gamma distribution, Logistic distribution, Normal distribution, Lognormal distribution, and 2 three-parameter distributions: Generalized extreme value distribution and Pearson type III distribution). Taking the Lognormal distribution as an example, its probability density function is as follows:

[0061]

[0062] In the formula, These are the location parameter and the scale parameter, respectively, and y represents the measured sequence;

[0063] The maximum likelihood method is used to estimate the distribution parameters. Assuming that each observation is independent, the log-likelihood functions under consistent and inconsistent conditions are expressed as follows:

[0064]

[0065]

[0066] In the formula, n is the measured length of the flood sequence; The flood variable Y t The probability density function y t This represents the observed value of the flood variable at a certain time t; These are covariates, including climate covariates X1 (such as SPA and TPA) and human activity covariates X2 (such as RI). It is the model parameter vector, under the consistency condition. θ is a constant, under non-uniform conditions t =(μ t , σ t , ξ t ), θ t μ represents the model distribution parameters at a certain time t. t σ represents the position parameter at a certain time t. t ξ represents the scale parameter at a given time t. t The shape parameter represents a certain time t;

[0067] Based on historical data, the distribution parameters change with the covariates: ;

[0068] (3) Substitute the future climate covariate and the human activity covariate into the functional relationship between the distribution parameters and covariates constructed based on the historical data to obtain the probability distribution parameters for future years. The obtained probability distribution parameters Substituting into the above probability density function Thus, the time-varying probability distribution of floods within the engineering design life is obtained. The GAMLSS method is characterized by making full use of hydrological and meteorological information for each future year while ensuring that the length of the engineering design life is not limited by the available data. It can also capture the sequence and magnitude of future extreme events, but is therefore susceptible to the influence of individual extreme values.

[0069] Methods for deriving time-varying probability distribution models using the sliding window method include:

[0070] (1) The variation diagnosis of flood sequences is performed using non-consistency diagnostic methods such as the Mann-Kendall trend test or the Pettitt change point test. The basin with a significant variation (e.g., 10% significance level) in the flood sequence is divided into two periods: the natural period (assuming no impact from climate change and human activities) and the change period (affected by climate change and human activities). The determination of the natural period must be before the non-consistency change point and the construction of the first large reservoir in the basin; the determination of the change period must be after the non-consistency change point, and the sequence lengths of the natural period and the change period should be as similar as possible. In this embodiment, the Pettitt (PT) change point test is used to diagnose the variation of the flood sequence to determine the non-consistency change point. Specifically:

[0071] For a sequence with n observations Pettitt test statistic K t The definition is as follows:

[0072]

[0073] In the formula,

[0074]

[0075] In the formula,

[0076]

[0077] The corresponding probability value of the Pettitt test ( )for:

[0078]

[0079] The null hypothesis of the PT test is that there are no significant mutations in the measured sample sequence. When, accept the null hypothesis, assuming that there is no significant mutation point in the sequence at position t; when When the null hypothesis is rejected, it is assumed that there is a significant mutation point in the series at position t, where α represents the significance level.

[0080] Hydrological models (Xin'anjiang model, GR4J model, etc.) were calibrated and validated using hydrological runoff data from the defined natural and variable periods, respectively. The objective function was to maximize the KGE efficiency coefficient, and its calculation formula is as follows:

[0081]

[0082] In the formula, These are measured runoff and model-simulated runoff sequences, respectively. The correlation coefficient between the measured runoff series and the simulated runoff series. and These are the average values ​​of the measured runoff sequence and the simulated runoff sequence, respectively. and These represent the standard deviations of the measured runoff series and the simulated runoff series, respectively. The range of the KGE efficiency coefficient is... The closer the KGE value is to 1.0, the better the runoff simulation effect of the model.

[0083] (2) The meteorological data (daily rainfall, evaporation, etc.) output by the climate model GCM are downscaled and then input into the hydrological model after the change period is verified to simulate the future daily runoff sequence of the watershed under non-uniform conditions.

[0084] (3) A 30-year sliding window is selected to process the future daily runoff sequence (e.g., the runoff sequence from 2021 to 2100). Each year within the engineering design period (e.g., the engineering design period is 66 years from 2021 to 2086) is taken as the center of the window. The annual maximum flow sequence of the 15 years before and after the center of the window is selected to form the flood sample within the window. The probability distribution corresponding to each window can be obtained by distribution fitting. This leads to the time-varying probability distribution of floods within the engineering design life. The sliding window method obtains the probability distribution for each future year by fitting a sliding window. The sliding window allows for significant overlap between adjacent years, avoiding the extreme impact of outliers in short time periods on the distribution fit. However, the use of a sliding window limits the length of the engineering design period. The sliding window length can be adjusted appropriately according to research needs, but there are also issues such as insufficient sample size when fitting the sequence distribution within the window for shorter windows, and reduced specificity of the results for each year within the engineering design period for longer windows.

[0085] Step 3: Based on the annual average reliability, estimate the non-consistent design flood values ​​under the two methods within the engineering design life T1~T2 according to the time-varying probability distribution model derived in Step 2;

[0086] In this step, the method for calculating the non-consistent design flood value within the engineering design life based on the annual average reliability estimate is as follows:

[0087]

[0088] In the formula, p is the probability of exceeding the limit; T1 and T2 are the start and end years of the project.

[0089] Step 4: Based on the actual data obtained in the watershed, compare the uncertainty of the design flood value between the GAMLSS method and the sliding window method, and select the non-consistent design flood value that better meets the design requirements of the watershed under study, so as to provide a reference for engineering design under changing environments;

[0090] When selecting the non-consistent design flood value, the first consideration is whether the studied watershed has historical flood data for the survey period (obtained through field surveys and literature verification of major floods that occurred in a certain river section in history, such as the flood discharges at the Yichang hydrological station since 1470 in 1560, 1613, 1788, 1796, 1860, and 1870) and corresponding non-consistent influencing factor data (such as reconstructed rainfall data from 1470 to 1950 within the watershed controlled by the Yichang hydrological station). If such data exists, the historical survey data increases the representativeness of the hydrological data, making the calculation of the non-consistent design flood value obtained based on the GAMLSS method more reliable. The sliding window method, however, cannot establish a model due to the difficulty in collecting detailed and continuous historical flood and meteorological data. Therefore, in this case, the design value result from the GAMLSS method is selected as the non-consistent design flood value. If the data does not meet this condition (i.e., the aforementioned data does not exist), the uncertainty of the non-consistent design value corresponding to different exceedance probabilities of the GAMLSS method and the sliding window method is compared, and the design value result with lower uncertainty between the two methods is selected. In this embodiment, the magnitude of uncertainty is quantified by the size of the quantile interval (e.g., the 5% to 95% quantile interval) of the design values ​​calculated by multiple climate models. That is, one climate model can calculate one design value corresponding to a certain probability of exceedance, and multiple climate models can obtain multiple design values ​​corresponding to a certain probability of exceedance, thus obtaining the quantile interval of the design value.

[0091] <Specific Example>:

[0092] The Wujiang River is the largest right-bank tributary of the upper reaches of the Yangtze River. Its basin has a significant elevation difference and is influenced by the alternating East Asian and South Asian monsoons, resulting in abundant rainfall and relatively rich hydropower resources, providing conditions for large-scale cascade reservoir development and operation. Due to the abundant rainfall and numerous cascade reservoirs in the Wujiang River basin, this invention considers a non-uniform design flood estimation method and system based on the Wujiang River basin as an example. By comparing the uncertainties of non-uniform design values ​​corresponding to different exceedance probabilities using the GAMLSS method and the sliding window method, the design value with lower uncertainty from the two methods is selected to guide the design of water conservancy projects in basins under changing environments. The specific process is as follows:

[0093] like Figure 1 As shown, historical runoff and meteorological data from Wulong Station, as well as information on the control area and total capacity of 11 large and medium-sized reservoirs within the basin, were first collected and organized. Meteorological data simulated by the CMIP6 Global Climate Model (GCM) (including three shared socio-economic pathways: SSP1-2.6, SSP2-4.5, and SSP5-8.5) were downloaded and downscaled to obtain daily meteorological data for future periods in the basin. Based on the basin planning, future new construction projects in the Wujiang River basin were not considered for the time being. Then, a time-varying probability distribution model was derived using both statistical modeling methods (GAMLSS method) and basin hydrological modeling methods (sliding window method). The GAMLSS method was used to calculate summer precipitation anomalies (SPA), annual precipitation anomalies (TPA), and reservoir index (RI). Based on the GAMLSS model, a non-uniform probability distribution model was constructed. Future climate covariates and human activity covariates were substituted into the optimal non-uniform probability distribution model selected based on the SBC criterion to obtain the probability distribution parameters for future years. Thus, the time-varying probability distribution of floods within the engineering design life is obtained. The sliding window method, based on the diagnostic results of flood sequence inconsistency, divides the watershed into two periods: the natural period (assuming no impact from climate change and human activities) and the variable period (affected by climate change and human activities). This is used for the calibration and validation of the Xin'anjiang hydrological model. In this example, the division of the Wujiang watershed, where the flood sequence exhibits significant variation, into these two periods is shown in Table 1. The model's effectiveness is judged based on the KGE coefficient. Future meteorological data is input into the calibrated hydrological model to simulate the future daily runoff of the watershed. Each year within the engineering design life is used as the window center, and the sequences of 15 years before and after the window center are selected to constitute the flood samples within that window. Distribution fitting yields the probability distribution corresponding to each window. This leads to the time-varying probability distribution of floods within the engineering design life. Next, the annual average reliability method is used to estimate the non-consistent design flood values ​​for the engineering design life corresponding to the GAMLSS method and the sliding window method, respectively. Finally, considering the unavailability of flood data and accurate non-consistent influence factor data for a longer historical survey period in the Wujiang River basin, the uncertainty of the non-consistent design values ​​corresponding to different exceedance probabilities of the GAMLSS method and the sliding window method is compared. The design value result with less uncertainty from the two methods is selected to guide the design of water conservancy projects in the basin under changing environments.

[0094] Table 1. Division of natural and variable periods in watersheds with significant variations in flood sequences.

[0095]

[0096] The correlation coefficients between the annual maximum daily discharge and precipitation anomaly factors and reservoir influencing factors were calculated. Table 2 shows that the correlation between the annual maximum daily discharge Q at Wulong station and the summer precipitation anomaly SPA is greater than the correlation between Q and the annual precipitation anomaly TPA. Q also shows a high negative correlation with the reservoir index RI. Therefore, SPA and RI were selected as the climate covariate and human activity covariate affecting flood discharge at Wulong station, respectively. A non-uniform probability distribution model of the annual maximum daily discharge was constructed based on the GAMLSS model. The maximum likelihood method was used to estimate the model parameters. The optimal non-uniform probability distribution model was selected based on the SBC criterion. Figure 2 As a result, the inconsistent model significantly outperformed the consistent model in fitting the flood samples. According to... Figure 3 The future SPA sequence at Wulong Station shows an upward trend. The sequence changes of SSP1-2.6 and SSP2-4.5 are similar. The upward trend of the sequence in the far future under the SSP5-8.5 scenario is significantly greater than that under the SSP1-2.6 and SSP2-4.5 scenarios.

[0097] Table 2. Correlation coefficients between annual maximum daily flow and SPA, TPA, and RI

[0098]

[0099] Table 3 shows that the Xin'anjiang model performs well in simulating runoff in the Wujiang River basin. According to... Figure 4 The future flood magnitude at Wulong Station is significantly lower than that of historical periods. This is because the flood sequence within the basin during the change period is greatly affected by human activities (especially the construction of large reservoir groups), thus reducing the magnitude. The changes in the flood sequence under the three scenarios SSP1-2.6, SSP2-4.5, and SSP5-8.5 are not significantly different.

[0100] Table 3 Evaluation of the Model Runoff Simulation Results of the Xin'anjiang River in the Wujiang River Basin

[0101]

[0102] according to Figure 5 At Wulong Station, under inconsistent conditions, the design flood values ​​for some scenarios are greater than the consistent design values ​​under very rare probability conditions. The 5%–95th quantile range for some scenarios corresponding to inconsistent design values ​​is relatively large, indicating that the flood design values ​​respond differently to different climate models. Comparing the design values ​​corresponding to a 0.1% exceedance probability, it can be seen that the inconsistent design values ​​corresponding to both the GAMLSS method and the sliding window method are smaller than the consistent design values. Except for the SSP1-2.6 scenario, where there is a difference, the design values ​​corresponding to the SSP2-4.5 and SSP5-8.5 scenarios are quite similar. Since the future SPA sequence shows an upward trend, it indicates that human activities, mainly reservoir construction, have a significant impact on the reduction of downstream design floods.

[0103] according to Figure 6 The values ​​of 0.001, 0.01, 0.1, and 0.2 exceed the 5% to 95% quantile range of the non-consistent design values ​​corresponding to the probability sliding window method, which is significantly larger than that of the GAMLSS method. Therefore, the non-consistent design flood values ​​of the GAMLSS method are selected as theoretical values ​​to guide the design of water conservancy projects in the Wujiang River Basin under changing environments.

[0104] The GAMLSS method explores the impacts of climate change and human activities through covariates, while the sliding window method considers the impacts of climate change and human activities on flood runoff more generally by dividing the natural period into natural and variable periods. Because the GAMLSS method and the sliding window method use different models and hydrological and meteorological information, their results are not strictly comparable. However, comparing their results can increase the persuasiveness of calculations for inconsistent design flood values.

[0105] The present invention also provides a system for implementing the above-described non-uniform design flood estimation method that takes into account climate and human activities, comprising:

[0106] The data acquisition module is used to acquire hydrological and meteorological data of the research watershed, as well as reservoir-related data;

[0107] The time-varying probability distribution model derivation module is used to derive the time-varying probability distribution model based on the acquired data, using the GAMLSS method and the sliding window method respectively.

[0108] The non-consistent design flood value calculation module is used to estimate the non-consistent design flood values ​​under two methods within the engineering design life T1~T2 based on the annual average reliability and the derived time-varying probability distribution model.

[0109] The module for determining the non-consistent design flood value of the study watershed is used to compare the design flood values ​​of the GAMLSS method and the sliding window method based on the actual data obtained in the watershed, and select the non-consistent design flood value that better meets the design requirements of the study watershed.

[0110] An electronic device includes a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the above method.

[0111] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0112] The above are merely preferred embodiments of the present invention and are not intended to limit the implementation methods and scope of protection of the present invention. Those skilled in the art should recognize that any equivalent substitutions and obvious changes made based on the content of this specification should be included within the scope of protection of the present invention.

Claims

1. A method for estimating non-uniform design floods considering climate and human activities, characterized in that, Includes the following steps: Step 1: Obtain hydrological and meteorological data and reservoir-related data for the study watershed; Step 2: Based on the data obtained in Step 1, the time-varying probability distribution model is derived using the GAMLSS method and the sliding window method, respectively; Step 3: Based on the annual average reliability, estimate the non-consistent design flood values ​​under the two methods within the engineering design life T1~T2 according to the time-varying probability distribution model derived in Step 2; Step 4: Based on the actual data obtained in the watershed, compare the non-consistent design flood values ​​of the GAMLSS method and the sliding window method, and select the non-consistent design flood value that is more in line with the situation of the watershed under study, so as to provide a reference for engineering design under changing environment; The methods for deriving time-varying probability distribution models using the sliding window method include: The watershed with significant variations in flood sequence is divided into two periods: the natural period and the variable period. The hydrological model is calibrated and validated in these two periods. The meteorological data output by the climate model GCMs was downscaled and then input into the above-mentioned validated hydrological model to simulate the future daily runoff sequence of the watershed. Selecting a sliding window length of m years, for the future daily runoff sequence, taking each year within the engineering design life as the window center, select the sequences of each year before and after the window center for m / 2 years to constitute the flood sample within that window. Distribution fitting can then yield the probability distribution corresponding to each window. This leads to the time-varying probability distribution of floods within the engineering design life. ; In step 4, when determining the non-consistent design flood value for the study area, the first consideration is whether there is flood data and non-consistent influencing factor data for a certain historical survey period in the studied watershed. If so, the design value result of the GAMLSS method is selected as the non-consistent design flood value; if not, the uncertainty of the non-consistent design value corresponding to different exceedance probabilities of the GAMLSS method and the sliding window method is compared, and the design value result with smaller uncertainty of the two methods is selected.

2. The non-uniform design flood estimation method considering climate and human activities according to claim 1, characterized in that, The data obtained in step 1 includes: The study investigates the historical runoff of the basin, meteorological data from the entire historical period of the basin, future meteorological data simulated by climate models, as well as the controlled area and total storage capacity of large and medium-sized reservoirs within the basin, and the planning and construction of reservoir projects in the basin plan.

3. The non-uniform design flood estimation method considering climate and human activities according to claim 2, characterized in that, The method for obtaining future meteorological data in step 1 includes: using the CMIP6 global climate model GCM to simulate meteorological data, and then downscaling it to obtain future rainfall data for the watershed.

4. The non-uniform design flood estimation method considering climate and human activities according to claim 1, characterized in that, Methods for deriving time-varying probability distribution models using the GAMLSS method include: Based on the collected basin meteorological data and reservoir data, the summer precipitation anomaly SPA, the annual precipitation anomaly TPA, and the reservoir index RI were calculated. The calculated summer precipitation anomaly SPA and annual precipitation anomaly TPA were used as basin climate covariates, and the reservoir index RI was used as a basin human activity covariate. Constructing a non-consistent probability distribution model based on the GAMLSS model And the maximum likelihood method is used to estimate its distribution parameters; Assuming that each observation is independent, the log-likelihood function is expressed as follows under consistent and inconsistent conditions: ； ； In the formula, n is the measured length of the flood sequence; The flood variable Y t The probability density function y t This represents the observed value of the flood variable at a certain time t; These are covariates, including climate covariate X1 and human activity covariate X2; It is the model parameter vector, under the consistency condition. θ is a constant, under non-uniform conditions t =(μ t , σ t , ξ t ), μ t σ represents the position parameter. t ξ represents the scale parameter. t The functional relationship between the shape parameter, i.e., the distribution parameter, and the covariates is as follows: ; Substituting the future climate covariate and the human activity covariate into the functional relationship between the distribution parameters and the covariates mentioned above, we obtain the probability distribution parameters for future years. Thus, the time-varying probability distribution of floods within the engineering design life is obtained. .

5. The non-uniform design flood estimation method considering climate and human activities according to claim 1, characterized in that, The formula for calculating the non-uniform design flood value in step 3 is: ； In the formula, p is the probability of exceeding the limit, and T1 and T2 are the start and end years of the project. This represents the time-varying probability distribution derived in step 2.

6. A system for implementing the non-uniform design flood estimation method considering climate and human activities as described in any one of claims 1-5, characterized in that, include: The data acquisition module is used to acquire hydrological and meteorological data of the research watershed, as well as reservoir-related data; The time-varying probability distribution model derivation module is used to derive the time-varying probability distribution model based on the acquired data, using the GAMLSS method and the sliding window method respectively. The non-consistent design flood value calculation module is used to estimate the non-consistent design flood values ​​under two methods within the engineering design life T1~T2 based on the annual average reliability and the derived time-varying probability distribution model. The module for determining the non-consistent design flood value of the study watershed is used to compare the design flood values ​​of the GAMLSS method and the sliding window method based on the actual data obtained in the watershed, and select the non-consistent design flood value that better meets the design requirements of the study watershed.

7. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.

Images