A design flood calculation method for data-scarce basins considering the impact of climate change

Abstract

The application discloses a design flood calculation method of a data-deficient basin considering the influence of climate change, and belongs to the technical field of hydrology, water resources and climate change adaptability assessment. Through collecting the measured precipitation and temperature data of a target basin and a surrounding region, deeply fusing global climate model data and regional standardized hydrology manual results, and constructing a double-path change rate analysis mechanism, two technical schemes are provided, i.e., a rainstorm change rate based on directly output precipitation data of the global climate model and a rainstorm change rate based on a temperature-precipitation physical relationship equation. The scientificity and robustness of the calculation results are enhanced, the water conservancy planning and design and flood control safety assessment ability of the data-deficient basin under the background of climate change are significantly improved, and the application value and popularization of the method are important.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary field of hydrology, water resources and climate change adaptation assessment, and specifically relates to a method for calculating design floods in data-scarce watersheds that takes into account the impact of climate change. Background Technology

[0002] Design floods are a crucial basis for the planning, design, operation, and management of water conservancy projects, and their reliability directly affects the safety of these projects and the flood control management of river basins. However, against the backdrop of global climate change, extreme precipitation events are becoming more frequent, especially with a significant increase in the intensity of short-duration rainfall. This poses a severe challenge to design flood results based on historical climate steady-state assumptions, profoundly impacting the flood control safety of small and medium-sized river basins.

[0003] Current methods for calculating design floods mainly fall into two categories: one is based on flood data and uses frequency analysis to derive the design flood for a specified standard; the other is based on rainfall series and uses a coupling approach between the design rainfall and a hydrological model to derive the design flood. Both methods require that the rainfall and flood data meet stationarity requirements and cannot account for design flood variations caused by future climate change. This is particularly problematic for small and medium-sized river basins, which often lack long-term rainfall data series and sufficient measured flood data for hydrological model construction. This makes it difficult to perform design rainfall calculations and construct hydrological models based on the basin's own rainfall and flood data, resulting in significant technical challenges in calculating design floods for small and medium-sized river basins under the influence of climate change. This hinders the scientific planning of water conservancy projects and the reliable assessment of flood control safety in such basins. Summary of the Invention

[0004] To address the aforementioned problems, this invention proposes a design flood calculation method for data-scarce watersheds that considers the impact of climate change. It employs a dual-path approach—using direct precipitation output and the temperature-precipitation physical relationship—to derive the rate of change in rainfall. This method couples climate model predictions with standardized hydrological calculation methods, systematically solving the design flood calculation challenges in data-scarce watersheds under the background of climate change. It can significantly improve the design and flood control safety assessment capabilities of water conservancy projects in small and medium-sized watersheds with no or scarce data under changing environments, enhancing the scientific rigor and robustness of the calculation results.

[0005] To solve the above problems, the present invention adopts the following technical solution:

[0006] A method for calculating design floods in data-scarce watersheds that takes into account the impacts of climate change, the steps of which are as follows:

[0007] (1) Collect measured precipitation and temperature data of the target watershed and surrounding areas to form a long series of measured precipitation and temperature data;

[0008] (2) Select m climate models from the global climate models, extract historical and future precipitation and temperature data, and perform downscaling to generate a precipitation and temperature dataset suitable for the target watershed.

[0009] (3) At least two different methods are used to correct the bias of historical and future rainfall and temperature data, and one of them is selected as the best. By sampling the annual maximum value, a series of extreme rainstorm samples with different durations are obtained, and the design values ​​of rainstorms with different durations in the historical period and the future period are calculated under different climate models and different return periods.

[0010] (4) Compare and analyze the design values ​​of rainstorms in the future period with those in the historical period, and calculate the rate of change of the design values ​​of rainstorms in the future period relative to the corresponding design values ​​of rainstorms in the historical period under different climate models, different scenarios, and different return periods.

[0011] (5) Based on measured precipitation and temperature data, establish the relationship between the annual maximum one-day heavy rainfall and the daily average temperature, and quantify the rate of change of extreme heavy rainfall with temperature variation, including:

[0012] Temperature binning technology was used to calculate the 99th percentile precipitation value and the average value of all temperatures within the corresponding interval, and extreme precipitation was paired with the corresponding temperature interval.

[0013] An exponential approach was used to analyze and fit the relationship between precipitation and temperature. ;

[0014] Determine the temperature corresponding to the inflection point Regression relationships were established between extreme rainfall events and corresponding temperatures on both sides of the turning point temperature:

[0015]

[0016] in, This is an extreme precipitation value. Let b be the temperature, b be the intercept, and k be the slope. For regression parameters;

[0017] (6) Based on the corrected historical and future temperature data from step (3), calculate the average watershed temperature for the historical period and the average watershed temperature for the future under different models and scenarios.

[0018] (7) Based on the average temperature of historical periods and future periods, and according to the established logarithmic linear regression equation of extreme rainfall and temperature, calculate the extreme rainfall values ​​of historical periods and future periods, and obtain the rate of change of extreme rainfall in the future relative to extreme rainfall in historical periods under different climate models and scenarios.

[0019] (8) Based on the design storm parameters of the target watershed, select the mean, coefficient of variation and skewness coefficient corresponding to the storm in the watershed, and calculate the design storm values ​​of the watershed for different durations under the current conditions;

[0020] (9) Calculate the design values ​​of rainstorms under different models and scenarios in the future period. :

[0021]

[0022] in, Design rainfall values ​​for different durations of the watershed under current conditions. Design the rate of change of rainstorm under different modes m and different scenarios s;

[0023] (10) Based on the design rainfall value Based on hydrological data, the typical time-history distribution ratio of the design storm in the target watershed is determined, and the design storm process of the watershed under different models and scenarios in the future is estimated.

[0024] (11) Based on step (10) and with reference to the hydrological data results, determine the runoff generation and confluence scheme of the target watershed, and deduce the watershed design flood under different models and scenarios in the future.

[0025] In each scenario, the mean or median of the set of design flood results corresponding to m climate models is used as the representative value, and the uncertainty of the design flood results is quantified by using the variance of the set or the interval composed of the 5th and 95th percentiles.

[0026] Furthermore, in step (2), a convolutional neural network model and quantile mapping method are used to correct the bias in the historical and future rainfall data of GCMs after downscaling.

[0027] Among them, the quantile mapping method adopts the Gamma distribution mapping method, and the Gamma distribution function is as follows:

[0028]

[0029]

[0030] In the formula, Let Γ(α) be the probability density function of the Gamma distribution; Γ(α) be the Gamma function; α be the shape parameter; and β be the scale parameter. and These are the cumulative function and inverse function of the Gamma distribution, respectively; These are the shape and scale parameters of the raw data for m-month GCMs, respectively. These are the shape and scale parameters of the measured precipitation in month m, respectively. This is the corrected value (mm) for GCMs precipitation on day d of month m after correction. This is the original value (mm) for GCMs product on day d of month m. The probability density function The independent variable, It is a natural constant.

[0031] Furthermore, in step (2), the linear scaling method and the quantile mapping method are used to correct the bias of the historical and future temperature data of the GCM after downscaling.

[0032] The linear scaling method can be expressed as follows: Based on the ratio of measured precipitation data to the multi-year monthly average of GCM precipitation data, a correction is made in terms of precipitation amount. The calculation formula is as follows:

[0033]

[0034] In the formula, , The values ​​are the measured and GCMs precipitation averages (mm) for the month of Mon, respectively.

[0035] Furthermore, step (5) specifically involves:

[0036] Data pairing begins by extracting all dates with precipitation > 0.1 mm from historical precipitation data, defining them as "wet days." Then, the precipitation on each wet day is paired with its corresponding daily average temperature to form a dataset.

[0037]

[0038] in: Rainfall (mm) For the corresponding temperature (°C). , for paired datasets

[0039] Temperatures are divided into boxes at 1°C intervals, ignoring temperature ranges with less than 10 wet days.

[0040] Calculate the percentile for the daily wet precipitation P in each temperature chamber, and sort all precipitation values ​​in the current temperature chamber in ascending order: Where 'a' represents the number of wet days in the temperature chamber; for the target percentile 'q', calculate the location index. If k is an integer, then the percentile is If k is not an integer, then it is calculated using linear interpolation:

[0041]

[0042] Establish a log-linear regression equation for the relationship between extreme rainfall and temperature.

[0043] Furthermore, the specific method for step (7) is as follows:

[0044] Calculate the average temperature of historical and future periods in GCMs data. and ;

[0045] Substituting this into the regression equation constructed above, and calculating the corresponding extreme rainfall, the rate of change under different models m and different scenarios s is then obtained. for:

[0046]

[0047] , This represents the extreme rainfall amounts under different future scenarios (f), different future scenarios (s), and different climate models (m) in the future period.

[0048] Furthermore, in step (2), the precipitation and temperature data for the future period include four shared socioeconomic path scenarios. An adversarial neural network model is used to downscale the historical data of m GCMs and the model data for the future period under the four scenarios to generate a precipitation and temperature dataset suitable for the target watershed.

[0049] Furthermore, in step (1), the measured precipitation and temperature data collected in the target watershed and surrounding areas include at least: data from the watershed's own observation stations, the ERA5 dataset, and precipitation and temperature data from nearby areas.

[0050] Furthermore, the hydrological data used are from the "Hydrological Handbook" issued by the province / region where the data is obtained.

[0051] The beneficial effects of this invention are:

[0052] (1) An innovative technical framework is proposed to deeply integrate global climate model data with the results of regional standardized hydrological manuals, and a new method for design flood calculation in data-scarce watersheds that takes into account the impact of climate change is established, providing a new approach for design storm and design flood calculation under the background of climate change.

[0053] (2) A dual-path rate of change analysis mechanism was constructed, providing two technical solutions: one is to estimate the rate of change of heavy rainfall based on precipitation data directly output from the global climate model, and the other is to estimate the rate of change of heavy rainfall indirectly based on the temperature-precipitation physical relationship equation. This enhances the scientificity and robustness of the calculation results.

[0054] (3) It effectively solves the technical bottleneck of design flood calculation in watersheds with scarce data under the background of climate change, and provides a reliable way to estimate design floods for small and medium-sized watersheds that cannot obtain long-term observation data.

[0055] (4) It significantly improves the planning and design of water conservancy projects and the flood control safety assessment capabilities of watersheds with no or scarce data under the background of climate change, and has important engineering application value and promotion prospects.

[0056] The present invention will be further described in detail below with reference to specific embodiments. Detailed Implementation

[0057] Example 1

[0058] This embodiment presents a method for calculating design floods in data-scarce watersheds that takes into account the impact of climate change. The method steps are as follows:

[0059] (1) Collect measured precipitation and temperature data of the target basin and surrounding areas to form a long series of measured precipitation and temperature data.

[0060] (2) Select m climate models from the global climate models, extract historical and future precipitation and temperature data, and perform downscaling to generate a precipitation and temperature dataset suitable for the target watershed.

[0061] (3) At least two different methods are used to correct the deviation of historical and future rainfall and temperature data, and one of them is selected as the best. By sampling the annual maximum value, a series of extreme rainstorm samples with different durations are obtained, and the design values ​​of rainstorms with different durations in the historical period and the future period are calculated under different climate models and different return periods.

[0062] (4) Compare and analyze the design values ​​of rainstorms in the future period with those in the historical period, and calculate the rate of change of the design values ​​of rainstorms in the future period relative to the corresponding design values ​​of rainstorms in the historical period under different climate models, different scenarios, and different return periods.

[0063] (5) Based on measured precipitation and temperature data, establish the relationship between the annual maximum one-day heavy rainfall and the daily average temperature, and quantify the rate of change of extreme heavy rainfall with temperature variation, including:

[0064] Temperature binning technology was used to calculate the 99th percentile precipitation value and the average value of all temperatures within the corresponding interval, and extreme precipitation was paired with the corresponding temperature interval.

[0065] An exponential approach was used to analyze and fit the relationship between precipitation and temperature. ;

[0066] Determine the temperature corresponding to the inflection point Regression relationships were established between extreme rainfall events and corresponding temperatures on both sides of the turning point temperature:

[0067]

[0068] in, This is an extreme precipitation value. Let b be the temperature, b be the intercept, and k be the slope. These are the regression parameters.

[0069] (6) Based on the corrected historical and future temperature data from step (3), calculate the average watershed temperature for the historical period and the average watershed temperature for the future under different models and scenarios.

[0070] (7) Based on the average temperature of historical periods and future periods, and according to the established logarithmic linear regression equation of extreme rainfall and temperature, calculate the extreme rainfall values ​​of historical periods and future periods, and obtain the rate of change of extreme rainfall in future periods relative to extreme rainfall in historical periods under different climate models and scenarios.

[0071] (8) Based on the design storm parameters of the target watershed, select the mean, coefficient of variation and skewness coefficient corresponding to the storm in the watershed, and calculate the design storm value of the watershed for different durations under the current conditions.

[0072] (9) Calculate the design values ​​of rainstorms under different models and scenarios in the future period. :

[0073]

[0074] in, Design rainfall values ​​for different durations of the watershed under current conditions. Design the rate of change of rainstorm under different modes m and different scenarios s.

[0075] (10) Based on the design rainfall value Based on hydrological data, the typical time-history distribution ratio of design storms in the target watershed is determined, and the design storm processes of the watershed under different models and scenarios in the future are estimated.

[0076] (11) Based on step (10) and with reference to hydrological data results, determine the runoff generation and runoff scheme of the target basin and deduce the basin design flood under different models and scenarios in the future.

[0077] In each scenario, the mean or median of the set of design flood results corresponding to m climate models is used as the representative value, and the uncertainty of the design flood results is quantified by using the variance of the set or the interval composed of the 5th and 95th percentiles.

[0078] Example 2:

[0079] This embodiment is a refinement of Embodiment 1. Taking watershed A, where data is scarce, as an example, this embodiment further explains how to use the technical solution of the present invention to calculate the design flood process of watershed A with a 50-year return period and a duration of 1 day under the influence of climate change:

[0080] A method for calculating design floods in data-scarce watersheds that takes into account the impacts of climate change, the method comprising the following steps:

[0081] (1) Collect precipitation and temperature data for watershed A. Since watershed A is a data-scarce watershed, in addition to collecting the limited data from the watershed's own observation stations, supplementary data were also collected through the ERA5 dataset, publicly available grid datasets based on station interpolation at home and abroad, and precipitation and temperature data from nearby areas to form a long series of measured precipitation and temperature data.

[0082] (2) Select m climate models from the global climate models and extract precipitation and temperature data for historical and future periods. The future period data includes four shared socioeconomic path scenarios (SSP126, SSP245, SSP370, and SSP585). Use an adversarial neural network model to downscale the historical data of the m GCMs and the future data of the four scenarios to generate a precipitation and temperature dataset suitable for the target watershed.

[0083] (3) Based on the measured rainfall data of the watershed obtained in step (1), the historical and future rainfall data of GCMs after downscaling in step (2) are corrected for bias using a convolutional neural network model and a quantile mapping method. The correction effects of the two methods are analyzed, and the correction result of one method is selected as the final result. For example, the quantile mapping method can be represented as follows:

[0084] The basis of the Gamma distribution mapping method is that precipitation time series follow a gamma distribution function. Based on the distribution of measured data, the precipitation distribution is adjusted by correcting the mean, standard deviation, and quantiles of the GCMs. The Gamma distribution function is as follows:

[0085]

[0086]

[0087] In the formula, Let Γ(α) be the probability density function of the Gamma distribution; Γ(α) be the Gamma function; α be the shape parameter; and β be the scale parameter. and These are the cumulative function and inverse function of the Gamma distribution, respectively; These are the shape and scale parameters of the raw data for m-month GCMs, respectively. These are the shape and scale parameters of the measured precipitation in month m, respectively. This is the corrected value (mm) for GCMs precipitation on day d of month m after correction. This is the original value (mm) for GCMs product on day d of month m. The probability density function The independent variable, It is the natural constant (e=2.71828…).

[0088] (4) Based on the measured temperature data of the watershed obtained in step (1), the historical and future temperature data of the GCMs after downscaling in step (2) are corrected for bias using linear scaling and quantile mapping methods, respectively. The correction effects of the two methods are analyzed, and the correction result of one method is selected as the final result. For example, the linear scaling method can be expressed as follows: Based on the ratio of the measured precipitation data to the multi-year monthly average of the GCMs precipitation data, the precipitation is corrected. The calculation formula is as follows:

[0089]

[0090] In the formula, The corrected value (mm) for GCMs precipitation on day d of month mon; The original value (mm) for GCMs product on day d of month mon; , The values ​​are the measured and GCMs precipitation averages (mm) for the month of Mon, respectively.

[0091] (5) Based on the corrected historical rainfall data of GCMs in step (3), extreme rainfall sample series with different durations are obtained by sampling the annual maximum value. The hydrological frequency analysis method based on Pearson's three-type curve is adopted, and the parameters are estimated by the linear moment method. The design value of the maximum 1-day rainfall in a 50-year return period is calculated for each of the m climate models in the historical period. ).

[0092] (6) Based on the corrected GCMs future period rainfall data in step (3), the extreme rainfall sample series with different durations are obtained by sampling the annual maximum value. The hydrological frequency analysis method based on Pearson's three-type curve is adopted, and the parameters are estimated by the linear moment method. The design value of the maximum 1-day rainfall in a 50-year return period is calculated for each of the m climate models and 4 scenarios (s) in the historical period. ).

[0093] (7) By comparing and analyzing the future period design value of heavy rainfall calculated in step (6) with the historical period design value of heavy rainfall calculated in step (5), calculate the rate of change of the future period design value of heavy rainfall relative to the corresponding historical period design value under different climate models, scenarios, and return periods. The rate of change of the design heavy rainfall under different model m and different scenario s. It can be calculated using the following formula:

[0094]

[0095] , This represents the extreme rainfall amounts under different future scenarios (f), different future scenarios (s), and different climate models (m) in the future period.

[0096] (8) Based on the measured rainfall and temperature data of the watershed obtained in step (1), establish the relationship between the annual maximum one-day heavy rainfall and the daily average temperature, and quantify the rate of change of extreme heavy rainfall with temperature. First, use temperature binning technology to calculate the 99th percentile precipitation value and the average value of all temperatures in the corresponding interval, and pair extreme precipitation with the corresponding temperature interval. Considering that there is a hook structure between extreme heavy rainfall and temperature, first determine the temperature corresponding to the turning point. Logarithmic regression equations for precipitation and temperature were established for the two temperature segments:

[0097]

[0098] in, This is an extreme precipitation value. Let b be the temperature, b be the intercept, and k be the slope. For regression parameters;

[0099] The specific steps are as follows:

[0100] ① Data pairing. First, extract all dates with precipitation > 0.1 mm from historical precipitation data, defining them as "wet days". Then, pair the precipitation of each wet day with its corresponding daily average temperature to form a dataset:

[0101]

[0102] in: Rainfall (mm) For the corresponding temperature (°C), , for paired datasets .

[0103] ② Perform temperature binning. Divide temperatures into bins at 1°C intervals (e.g., 0–1°C, 1–2°C, … 30–31°C) (if the temperature on a given day is 12.3°C, it should be placed in the 12–13°C range). Ignore temperature ranges with fewer than 10 wet days (to ensure statistical reliability).

[0104] ③ Calculate the percentile for the daily wet precipitation P in each temperature chamber (e.g., 12–13°C). Sort all precipitation values ​​in the current temperature chamber in ascending order: , where a is the number of wet days in the temperature chamber. For the target percentile q (e.g., 75) th 90 th 95 th 99 th ), calculate location index If k is an integer, then the percentile is If k is not an integer, then it is calculated using linear interpolation: (Rounding down and rounding up, respectively).

[0105] ④ Establish a regression relationship between precipitation and corresponding temperature. Using the 99th percentile of each temperature interval as the extreme precipitation value, re-pair the extreme precipitation with the corresponding temperature interval (99th percentile precipitation value and the average of all temperatures within the corresponding interval). Since precipitation and temperature exhibit a peak structure, use statistical methods to find the temperature corresponding to the inflection point. For the two temperature ranges, regression equations for logarithmic precipitation and temperature are established using the following formulas, and quasi-linear regression calculations are performed:

[0106]

[0107] Where b is the intercept and k is the slope, the left and right sets of values ​​are obtained from regression calculations. and .

[0108] (9) Based on the corrected historical and future temperature data of GCMs in step (4), calculate the average watershed temperature in the historical period and the average watershed temperature under different models and scenarios in the future period.

[0109] (10) Based on the historical and future average temperatures obtained in step (9), and according to the extreme rainfall and temperature relationship equation established in step (8), calculate the extreme rainfall values ​​for the historical and future periods, and obtain the rate of change of extreme rainfall in the future relative to the historical extreme rainfall under different climate models and scenarios. The specific scheme is as follows: First, calculate the historical and future average temperatures of the GCMs data. and Substitute the annual average temperature or summer average temperature into the rainfall-temperature relationship equation constructed above to calculate the corresponding extreme rainfall. The rate of change under different modes m and different scenarios s for:

[0110]

[0111] (11) Since the target basin A lacks data, the mean, coefficient of variation and skewness coefficient of the corresponding rainstorm in the basin are found based on the results of the Hydrological Handbook issued by the provinces / regions where the basin is located. Based on this, the maximum annual design rainstorm value of the basin under the current conditions can be calculated.

[0112] (12) Design storm based on the current conditions obtained in step (11) Combining the rates of change of future storm design values ​​calculated using two different methods in steps (7) and (10), the storm design values ​​under different models and scenarios in the future are calculated according to the following formula. :

[0113]

[0114] (13) Comprehensive analysis of the design values ​​of rainstorms in the future under different modes and scenarios obtained by the two approaches in (12). Under each scenario, m future design rainstorm values ​​can be obtained, and their uncertainty can be analyzed accordingly.

[0115] (14) Based on the future design rainfall values ​​obtained in step (14), and in accordance with the results of the Hydrological Manual issued by the province / region where Basin A is located, determine the typical time history distribution scheme of the design rainfall of Basin A, and deduce the basin design rainfall process under m climate models and 4 different scenarios in the future period.

[0116] (15) Based on the future basin design rainstorm process obtained in step (14), determine the runoff generation and runoff scheme of the target basin according to the results of the Hydrological Manual issued by the province / region where basin A is located, and deduce the basin design flood under m climate models and 4 different scenarios in the future period.

[0117] (16) In each scenario (SSP126, or SSP245, or SSP370 and or SSP585), based on the set of design flood results corresponding to m climate models, the mean or median of the set is used as the possible value of the design flood in the future period, and the variance of the set and the interval composed of the 5th and 95th quantiles of the set are used to quantify the uncertainty of the future design flood.

[0118] Through the above steps (1)-(16), the design flood process of the A basin under the influence of climate change can be calculated.

[0119] Example 3

[0120] Based on the above embodiments, this embodiment is an application example:

[0121] Taking a specific watershed as an application example, and combining ERA5 and GCM data, this invention employs a method for calculating design floods in data-scarce watersheds that considers the impact of climate change to calculate the future design floods for this watershed. The specific steps are as follows:

[0122] First, ERA5 hourly precipitation and temperature data for the watershed from 1991 to 2020 were collected. After moving average calculations, 24-hour cumulative precipitation and corresponding 24-hour average temperature series were obtained. Then, a GCM climate model (name: FGOALS-g3) under the SSP245 scenario in the NEX-GDDP-CMIP6 downscaling dataset was selected to extract historical (1951-2000) and future (2051-2100) precipitation and temperature data for the watershed. Using quantile mapping as a benchmark, bias corrections were applied to the GCM data to obtain corrected GCM data, including precipitation and temperature.

[0123] The obtained 24-hour cumulative precipitation series was paired with the corresponding 24-hour average temperature series using "wet day" pairing to form a dataset. Temperature intervals of 1℃ were selected (0-1℃, 1-2℃, ..., 29-30℃), discarding temperature intervals with fewer than 10 samples. The average temperature (x-axis) and the natural logarithm of the 99th precipitation level (y-axis) for each temperature interval were used as binning values. Piecewise linear fitting was performed on both sides of the temperature interval based on statistical relationships. The relationship between extreme precipitation and air temperature established through temperature binning is as follows:

[0124]

[0125] Based on the bias-corrected GCM data, the historical and future average watershed temperatures were calculated to be 13.74℃ and 15.09℃, respectively. Substituting the rainfall-temperature relationship obtained from the temperature binning, the design storm rainfall variation rate calculated based on this GCM under the SSP245 scenario was obtained. The results are as follows:

[0126] 16.01%

[0127] Using the reasoning formula provided in the local Hydrological Manual, the calculated future design rainfall was converted into the future design flood, and the result was 1530 m³ / s.

[0128] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features, and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for calculating design floods in data-scarce watersheds considering the impacts of climate change, characterized in that, The steps are as follows: (1) Collect measured precipitation and temperature data of the target watershed and surrounding areas to form a long series of measured precipitation and temperature data; (2) Select m climate models from the global climate models, extract historical and future precipitation and temperature data, and perform downscaling to generate a precipitation and temperature dataset suitable for the target watershed. (3) At least two different methods are used to correct the bias of historical and future rainfall and temperature data, and one of them is selected as the best. By sampling the annual maximum value, a series of extreme rainstorm samples with different durations are obtained, and the design values ​​of rainstorms with different durations in the historical period and the future period are calculated under different climate models and different return periods. (4) Compare and analyze the design values ​​of rainstorms in the future period with those in the historical period, and calculate the rate of change of the design values ​​of rainstorms in the future period relative to the corresponding design values ​​of rainstorms in the historical period under different climate models, different scenarios, and different return periods. (5) Based on measured precipitation and temperature data, establish the relationship between the annual maximum one-day heavy rainfall and the daily average temperature, and quantify the rate of change of extreme heavy rainfall with temperature variation, including: Temperature binning technology was used to calculate the 99th percentile precipitation value and the average value of all temperatures within the corresponding interval, and extreme precipitation was paired with the corresponding temperature interval. An exponential approach was used to analyze and fit the relationship between precipitation and temperature. ; Determine the temperature corresponding to the inflection point Regression relationships were established between extreme rainfall events and corresponding temperatures on both sides of the turning point temperature: in, This is an extreme precipitation value. Let b be the temperature, b be the intercept, and k be the slope. For regression parameters; (6) Based on the corrected historical and future temperature data from step (3), calculate the average watershed temperature for the historical period and the average watershed temperature for the future under different models and scenarios. (7) Based on the average temperature of historical periods and future periods, and according to the established logarithmic linear regression equation of extreme rainfall and temperature, calculate the extreme rainfall values ​​of historical periods and future periods, and obtain the rate of change of extreme rainfall in the future relative to extreme rainfall in historical periods under different climate models and scenarios. (8) Based on the design storm parameters of the target watershed, select the mean, coefficient of variation and skewness coefficient corresponding to the storm in the watershed, and calculate the design storm values ​​of the watershed for different durations under the current conditions; (9) Calculate the design values ​​of rainstorms under different models and scenarios in the future period. : in, Design rainfall values ​​for different durations of the watershed under current conditions. Design the rate of change of rainstorm under different modes m and different scenarios s; (10) Based on the design rainfall value Based on hydrological data, the typical time-history distribution ratio of the design storm in the target watershed is determined, and the design storm process of the watershed under different models and scenarios in the future is estimated. (11) Based on step (10) and with reference to the hydrological data results, determine the runoff generation and confluence scheme of the target watershed, and deduce the watershed design flood under different models and scenarios in the future. In each scenario, the mean or median of the set of design flood results corresponding to m climate models is used as the representative value, and the uncertainty of the design flood results is quantified by using the variance of the set or the interval composed of the 5th and 95th percentiles.

2. The method for calculating design floods in data-scarce watersheds considering the impact of climate change, as described in claim 1, is characterized in that... In step (2), a convolutional neural network model and quantile mapping method are used to correct the bias in the historical and future rainfall data of GCMs after downscaling. Among them, the quantile mapping method adopts the Gamma distribution mapping method, and the Gamma distribution function is as follows: In the formula, Let Γ(α) be the probability density function of the Gamma distribution; Γ(α) be the Gamma function; α be the shape parameter; and β be the scale parameter. and These are the cumulative function and inverse function of the Gamma distribution, respectively; These are the shape and scale parameters of the raw data for m-month GCMs, respectively. These are the shape and scale parameters of the measured precipitation in month m, respectively. This is the corrected value (mm) for GCMs precipitation on day d of month m after correction. This is the original value (mm) for GCMs product on day d of month m. The probability density function The independent variable, It is a natural constant.

3. The method for calculating design floods in data-scarce watersheds considering the impact of climate change, as described in claim 1, is characterized in that... In step (2), the linear scaling method and the quantile mapping method are used to correct the bias of the historical and future temperature data of GCM after downscaling. The linear scaling method can be expressed as follows: Based on the ratio of measured precipitation data to the multi-year monthly average of GCM precipitation data, a correction is made in terms of precipitation amount. The calculation formula is as follows: In the formula, , The values ​​are the measured and GCMs precipitation averages (mm) for the month of Mon, respectively.

4. The method for calculating design floods in data-scarce watersheds considering the impact of climate change, as described in claim 1, is characterized in that... Step (5) specifically involves: Data pairing begins by extracting all dates with precipitation > 0.1 mm from historical precipitation data, defining them as "wet days." Then, the precipitation on each wet day is paired with its corresponding daily average temperature to form a dataset. in: Rainfall (mm) For the corresponding temperature (°C). ; Temperatures are divided into boxes at 1°C intervals, ignoring temperature ranges with less than 10 wet days. Calculate the percentile for the daily wet precipitation P in each temperature chamber, and sort all precipitation values ​​in the current temperature chamber in ascending order: Where 'a' represents the number of wet days in the temperature chamber; for the target percentile 'q', calculate the location index. If k is an integer, then the percentile is If k is not an integer, then it is calculated using linear interpolation: Establish a log-linear regression equation for the relationship between extreme rainfall and temperature.

5. The method for calculating design floods in data-scarce watersheds considering the impact of climate change, as described in claim 1, is characterized in that... The specific method for step (7) is as follows: Calculate the average temperature of historical and future periods in GCMs data. and ; Substituting this into the regression equation constructed above, and calculating the corresponding extreme rainfall, the rate of change under different models m and different scenarios s is then obtained. for: , This represents the extreme rainfall amounts under different future scenarios (f), different future scenarios (s), and different climate models (m) in the future period.

6. The method for calculating design floods in data-scarce watersheds considering the impact of climate change, as described in claim 1, is characterized in that... In step (2), the precipitation and temperature data for the future period include four shared socioeconomic path scenarios. An adversarial neural network model is used to downscale the historical data of m GCMs and the model data for the future period under the four scenarios to generate a precipitation and temperature dataset suitable for the target watershed.

7. The method for calculating design floods in data-scarce watersheds considering the impact of climate change, as described in claim 1, is characterized in that... In step (1), the measured precipitation and temperature data collected in the target watershed and surrounding areas include at least: data from the watershed's own observation stations, the ERA5 dataset, and precipitation and temperature data from nearby areas.

8. The method for calculating design floods in data-scarce watersheds considering the impact of climate change, as described in claim 1, is characterized in that... The hydrological data used are from the "Hydrological Handbook" issued by the province / region where the data is obtained.

9. The method for calculating design floods in data-scarce watersheds considering the impact of climate change, as described in any one of claims 1-8, is characterized in that... It can be used to calculate the design flood process of a target watershed with a 50-year return period under the influence of climate change.

Images