Economic risk prediction method, system and equipment based on heat wave and flood combined disaster

By combining Clausius-Clapperon thermodynamic equations and machine learning models, considering the uncertainty of global climate pattern ensembles, and evaluating the socio-economic impact of high-temperature heat wave-flood composite disasters, solving the failure to effectively evaluate the existing technology. The issue of humidity and pattern uncertainty has achieved more accurate climate disaster risk assessment and early warning.

CN118504994BActive Publication Date: 2024-11-15WUHAN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410970135.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-19
Publication Date
2024-11-15
Estimated Expiration
2044-07-19

AI Technical Summary

Technical Problem

The existing technology fails to effectively consider the uncertainty of atmospheric humidity and global climate pattern collections when predicting high-temperature heat wave-flood composite disasters, resulting in insufficient accuracy and reliability of simulation results, and fails to scientifically evaluate extreme disasters for human health and Socio-economic impact.

Method used

The Clausius-Clapperon thermodynamic equation is used to determine the relative humidity and specific humidity in meteorological and hydrological data, combined with machine learning models and global climate model ensembles, predict future scenarios through quantile deviation correction and multi-model weighted averaging method. and establish a disease-high temperature regression model and risk assessment model to evaluate the impact of wet stress heat wave-flood composite disasters on GDP, population and agriculture.

Benefits of technology

It provides more accurate and operational climate disaster risk assessment and early warning, helps to scientifically formulate emission reduction strategies, improves the predictive capacity of hydrological disasters in future climate change, and enhances the socio-economic risk assessment of compound disasters.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118504994B_ABST
    Figure CN118504994B_ABST
Patent Text Reader

Abstract

The present invention provides an economic risk prediction method, system and equipment based on heat wave and flood compound disasters, including: collecting meteorological, hydrological and social economic data sets, and calibrating multiple machine learning models; using the corrected global climate model set meteorological simulation data to train the machine learning model, and obtain the daily runoff process under the future scenario; combining the global climate model set with the machine learning model, and determining the weight parameters of each combined scenario in the multi-model weighted average method; extracting the wet-bulb temperature and pathogenic high temperature stress index based on reanalysis data, establishing a disease-high temperature regression model in the flood season, and solving the model parameters; predicting the population, GDP and agricultural risks caused by future compound disasters based on the weight parameters of each combined scenario in the multi-model weighted average method, and calculating the comprehensive social and economic risks. The present invention provides an important and highly operational reference for extreme climate disaster risk assessment and early warning.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of data prediction and processing, and in particular to a method, system and device for predicting economic risks based on combined disasters of heat waves and floods. Background Art

[0002] Since the Industrial Revolution, the emission of greenhouse gases such as carbon dioxide in the atmosphere has increased, the energy balance and material circulation process of the earth's land and atmosphere have changed, the global land temperature has continued to rise, coral reefs and tropical rainforests have died out on a large scale, extreme disasters such as heat waves, floods, droughts, and wildfires have occurred frequently, and the imbalance and inequality of the social and economic system and the ecological environment have become increasingly severe. In recent years, heat waves have had a significant impact on social and economic development and human health, and have received widespread attention from scholars at home and abroad. Heat wave events often mean higher atmospheric water vapor content, which may trigger extreme precipitation after heat wave events, and then cause flood disasters. In order to meet the above challenges, it is crucial to study the evolution and prediction of high temperature heat wave-flood compound disasters under climate change.

[0003] At present, the research on future climate change mainly relies on global climate models. Global climate models allow simulation of the physical processes and interactions of changes in various components of the Earth's climate system (including the atmosphere, ocean, land surface processes, cryosphere and biosphere). Due to the complexity of the global climate system, each climate model makes assumptions or approximations to varying degrees when simulating changes in these components. In recent years, domestic and foreign scholars have combined the global climate model ensemble under the Sixth International Coupled Model Intercomparison Program with basin hydrological models and machine learning models to study the evolution of future hydrological disasters. However, current studies usually use the arithmetic mean of the global climate model ensemble to demonstrate the simulation results, without considering the impact of the uncertainty of the global climate model ensemble itself on the simulation results. In addition, existing studies pay little attention to high temperature heat wave-flood compound disasters considering atmospheric humidity, and there is no relevant literature that uses high temperature-disease models to quantitatively diagnose the impact of high temperature heat wave events on human health, which is not conducive to scientifically assessing the risks of future extreme disasters and their socioeconomic impacts. Summary of the invention

[0004] The invention provides a method, system and device for predicting economic risks based on heat wave and flood combined disasters, so as to solve the defect of the existing technology that there is no uncertainty of atmospheric humidity and global climate model set for combined disaster prediction.

[0005] In a first aspect, the present invention provides an economic risk prediction method based on heat wave and flood compound disasters, comprising the following steps:

[0006] Step 100: Collect meteorological and hydrological data sets, disease samples and socio-economic data of the study basin, and deduce the flood season of the study basin;

[0007] Step 200: using the Clausius-Clapeyron thermodynamic equation to determine the relative humidity and specific humidity of each grid point in the meteorological and hydrological data set, and training five machine learning models based on the meteorological and hydrological data set;

[0008] Step 300: Based on the quantile bias correction method, obtain the global climate model ensemble meteorological simulation data under the climate change scenario, use the corrected global climate model ensemble meteorological simulation data to train the five machine learning models, and obtain the daily runoff series under the future scenario;

[0009] Step 400: combining the global climate model ensemble meteorological simulation data with the five machine learning models to determine the weight parameters of each combined scenario in the multi-model weighted average method;

[0010] Step 500: Based on the relative humidity of each grid point calculated in step 200 and the meteorological and hydrological data set of step 100, the wet-bulb temperature is deduced, and the pathogenic high temperature stress index and the pathogenic high temperature stress index are deduced through the wet-bulb temperature and relative humidity, a disease-high temperature regression model for the flood season is established, and the disease-high temperature regression model parameters are solved;

[0011] Step 600: Based on the daily runoff series under the future scenario, a super-quantitative sampling method is used to extract flood events; then heat wave events are extracted based on the disease-high temperature regression model, and wet stress heat wave-flood compound disasters are extracted;

[0012] Step 700: Using the flood season precipitation as a covariate, a risk assessment model based on the Copula function and the same frequency combination is established, and based on the weight parameters of each combination scenario in the multi-model weighted average method, the GDP, population and agricultural risks caused by future wet stress heat wave-flood disasters are predicted, and the comprehensive socio-economic risks are calculated.

[0013] Further, step 100 includes:

[0014] Obtain the daily runoff series of the control water level station in the study basin, and obtain the meteorological data of the ERA5 reanalysis data set. Obtain precipitation, near-surface air pressure, 2m air temperature, 2m dew point temperature, shortwave radiation, near-surface wind speed, longwave radiation, and snowfall data from the ERA5 reanalysis data set, and obtain the daily series data after time scale conversion;

[0015] Collect meteorological data from five pre-designed global climate models in the sixth phase of the International Coupled Model Intercomparison Project;

[0016] Access to population and GDP data from the Shared Socioeconomic Pathways dataset;

[0017] Obtain cropland area data for historical periods and under different SSP scenarios;

[0018] Obtain data on illness and death cases during the flood season from disease and health management departments, hospitals, and medical centers in the study basin;

[0019] Based on the daily precipitation data extracted from the ERA5 reanalysis dataset, the Thiessen polygon method was used to obtain the surface average daily precipitation series of the study basin, which was then integrated to obtain the monthly precipitation series. Finally, the cumulative precipitation for any five consecutive months in a year was solved, and the five consecutive months with the largest precipitation were divided into the flood season.

[0020] Further, step 200 includes:

[0021] Determine the latent heat of vaporization constant, the water vapor gas constant, the first integral constant and the second integral constant in the Clausius-Clapeyron thermodynamic equation to obtain a nonlinear function of saturated water vapor pressure and air temperature;

[0022] Substitute the hourly 2m surface air temperature and hourly dew point temperature of each grid point in the ERA5 reanalysis data set into the nonlinear function of saturated water vapor pressure and air temperature to obtain the hourly relative humidity, and then calculate the daily average relative humidity;

[0023] Substituting the hourly dew point temperature into the nonlinear function of saturated water vapor pressure and air temperature and the ground pressure in the ERA5 reanalysis data set to obtain the hourly specific humidity, and then calculating the daily average specific humidity;

[0024] Based on the daily precipitation, daily average 2m temperature, daily maximum 2m temperature, daily minimum 2m temperature, shortwave radiation, near-ground wind speed, longwave radiation and daily snowfall at each grid point in the ERA5 reanalysis dataset, as well as the inferred daily average specific humidity and relative humidity as input, and taking the measured runoff data of the basin control hydrological station as the target variable, 5 machine learning models were established, the 5 machine learning models included convolutional neural network, long short-term memory model, support vector machine, random forest and artificial neural network; and the 5 machine learning models were calibrated by optimizing the training through the gradient descent method.

[0025] Further, step 300 includes:

[0026] The first step is to obtain a global climate model ensemble meteorological simulation series data under a climate change scenario based on a global climate model ensemble and a quantile deviation correction method, calculate the difference between the output variable and the observed meteorological variable in the global climate model ensemble meteorological simulation series data at each quantile, remove the difference in the output future scenario of the global climate model ensemble meteorological simulation series data, and obtain the corrected global climate model ensemble meteorological simulation data;

[0027] In the second step, the calibrated five machine learning models are driven by the corrected global climate model ensemble meteorological simulation data to output the daily runoff series under the future scenario.

[0028] Further, step 400 includes:

[0029] The weight parameters of the multi-model weighted average method are derived based on five machine learning models, including:

[0030] In the weight derivation scheme, the weights for January to December are deduced separately, as follows:

[0031] For each month, the weight parameters of the 5 machine learning models satisfy:

[0032]

[0033] in, i Indicates the first i A machine learning model; Indicates i The weight parameters of the machine learning model LM Indicates the number of machine learning models, which is 5;

[0034] The weight is calculated using the following formula :

[0035]

[0036] Among them, RB i Characterization i The relative deviation between the runoff simulated by the machine learning model and the measured runoff; RMS i No. i The absolute deviation between the runoff simulated by the machine learning model and the measured runoff, and exp() represents the natural exponential function.

[0037] Further, step 500 includes:

[0038] In the first step, for each grid point, the hourly wet-bulb temperature is calculated based on the hourly surface 2m air temperature provided by the ERA5 reanalysis dataset and the hourly relative humidity calculated in step 200:

[0039]

[0040] Where: RH is the relative humidity, T dry The average daily temperature at 2m T w is the estimated wet bulb temperature, atan is the inverse tangent function;

[0041] In the second step, for each grid point, the hourly wet-bulb temperature and hourly relative humidity are used to deduce the hourly pathogenic high temperature stress index:

[0042]

[0043] Where: T h is the pathogenic high temperature stress index;

[0044] The third step is to calculate the daily average wet-bulb temperature and the daily average pathogenic high temperature stress index based on the hourly wet-bulb temperature and the hourly pathogenic high temperature stress index of each grid point, and calculate the spatial average daily wet-bulb temperature and the daily pathogenic high temperature stress index of the study basin by the Thiessen polygon method;

[0045] The fourth step is to establish a disease-high temperature regression model during the flood season based on the disease and death case data during the flood season obtained in step 100, as follows:

[0046]

[0047] Where: Indicates the first y The total number of illnesses and deaths during the annual flood season, DD y Indicates y Number of days in the annual flood season; Indicates y The flood season of the year dd The pathogenic high temperature stress index of the day, Indicates y The flood season of the year dd daily wet-bulb temperature; is a parameter of the disease-high temperature regression model; the calculation result of step 100 is used in the flood season;

[0048] The least squares method was used to solve the disease-high temperature regression model during the flood season and the parameters were obtained. .

[0049] Furthermore, the step 600 specifically includes:

[0050] The first step is to select the identification threshold of flood events and extract a certain quantile of the daily runoff series in the historical period as the flood standard. The daily runoff exceeding the threshold in the historical period and future period is a flood event. The daily runoff under the scenario exceeding the threshold is used as the peak flow and the standard flood magnitude.

[0051] In the second step, the meteorological data of the global climate model under different shared socioeconomic pathways are used to deduce the daily average wet-bulb temperature and the daily average pathogenic high temperature stress index of each grid point, and the parameters of the disease-high temperature regression model are used to calculate the normalized high temperature index for the historical period and the future period, which is:

[0052]

[0053] Where: HI is the normalized high temperature index;

[0054] In the third step, a certain quantile of the historical wet-bulb temperature series is used as the threshold. Events exceeding this threshold are defined as high temperature events. If a high temperature event lasts for more than three days, it is defined as a high temperature heat wave event. When a high temperature heat wave event occurs within 15 days before a flood event, it is determined to be a wet stress heat wave-flood composite disaster. The flood magnitude and intensity of the high temperature heat wave in the composite disaster event can then be extracted.

[0055] Further, step 700 includes:

[0056] The first step is to construct the joint probability distribution function of heat wave intensity and flood magnitude based on the Copula function; based on the combination scenario of the same frequency combination and optimal composite disaster attributes, the optimal combination function is constructed;

[0057] For any scenario under the climate scenario, the Gumbel Copula function is selected as the joint probability distribution function of the heat wave intensity and flood magnitude, and the parameters of the Copula function are Replace with time-varying parameters :

[0058]

[0059] in, is the Copula joint distribution function, Range is ; u t , v t Heat wave intensity D , Flood magnitude S Probability density function of marginal distribution;

[0060] Based on the definition of Copula function, the non-uniform two-variable Copula function is expressed as:

[0061]

[0062] in, represents the time-varying joint distribution function of D and S; C() is the abbreviation of Copula function, , Respectively D and S The time-varying marginal distribution function of the variable, , Respectively D and S time-varying parameters of variables;

[0063] The parameters of the time-varying Copula function are expressed as covariates w It is expressed as:

[0064]

[0065] in, represents the link function of the copula function, when hour, ; b 0 , b 1 are the parameters of the model respectively; w To study the spatial average flood season precipitation in the basin;

[0066] Then the same frequency combination function is:

[0067]

[0068] in, represents a joint return period T or High temperature heat wave intensity d , Flood magnitude s The most likely combination of scenarios; μ is the average interval time of wet stress heat wave-flood composite disaster; T AND is the return period of “and”;

[0069] The second step is to deduce the GDP, population and agricultural risks caused by the increase of moisture stress heat wave-flood compound disasters under future climate change based on the results of multi-model weighted average;

[0070] Newton iteration method is used to solve the same frequency combination function to obtain a certain return period in the historical period T h Corresponding heat wave intensity and flood magnitude ( , ); Use the historical period length as the sliding window to construct the time-varying marginal distribution and Copula function of the future period, and ( , ) are substituted into the future time period k The time-varying distribution function of a sliding window is used to calculate the new return period ;

[0071] After obtaining the return period for each combination scenario, the weighted average method is used to obtain M In the combined scenario k The average return period of the sliding window :

[0072]

[0073] in, i Indicates a combined scenario; represents the weight parameter of the combined scenario; Indicates a combination scenario i Next k The return period of a sliding window; if This indicates that k The composite disaster risk of each window increases, and vice versa;

[0074] GDP, population and agricultural risks in future periods are measured by the following formula:

[0075]

[0076]

[0077]

[0078] in, , and Respectively represent the population, GDP and agricultural risks caused by compound disasters, , and CROP k Respectively k Population, GDP, and farmland area in the year; I (•) is the indicator function, When it is, it is recorded as 1, otherwise it is recorded as 0; N 1 and N 2 Respectively represent the start and end years of the study period;

[0079] Using the results of GDP, population and agricultural risks, the comprehensive socio-economic risk is calculated:

[0080]

[0081] in: Represents the comprehensive socio-economic risks affected by compound hazards; They represent the weights of population, GDP and agricultural risks respectively, and are determined according to the characteristics of socio-economic development in the study area.

[0082] In another aspect, the present invention provides an economic risk prediction system based on heat wave and flood compound disasters, comprising:

[0083] A collection module is used to collect meteorological and hydrological data sets;

[0084] A training module, using the Clausius-Clapeyron thermodynamic equation to determine relative humidity and specific humidity, and training five machine learning models based on the meteorological and hydrological data set;

[0085] A correction module, for obtaining global climate model ensemble meteorological simulation data under a climate change scenario based on a quantile bias correction method, and using the corrected global climate model ensemble meteorological simulation data to train the five machine learning models to obtain a daily runoff series under a future scenario;

[0086] A combination module, used to combine the global climate model set with the machine learning model to determine the weight parameters of each combination scenario in the multi-model weighted average method;

[0087] Establish a module to build a disease-high temperature regression model during the flood season based on the acquired disease-related death data in the study basin, and solve the model parameters;

[0088] A calculation module is used to extract flood events under future climate change scenarios through the hydrological process under the future scenarios, further extract moisture stress heat wave-flood compound events, and obtain high temperature heat wave intensity and flood magnitude;

[0089] The prediction module is used to establish a two-dimensional risk assessment model based on Copula functions and same-frequency combinations under inconsistent conditions, predict the social population, GDP and agricultural risks caused by future compound disasters based on the weight parameters of each combination scenario in the multi-model weighted average method, and calculate the comprehensive social and economic risks.

[0090] On the other hand, the present invention provides a device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements an economic risk prediction method based on heat wave and flood combined disasters when executing the program.

[0091] Compared with the prior art, the present invention has the following beneficial effects:

[0092] The economic risk prediction method, system and electronic equipment based on the heat wave and flood combined disasters provided by the present invention combine the earth system model, the climate-health model considering disease risks, the water-heat balance equation, the global hydrological model, the machine learning model, and the moisture stress heat wave and flood combined disasters, to provide an important and highly operational reference basis for the risk assessment and early warning of hydrological disasters under climate change, and provide engineering reference value for responding to future climate disasters and scientifically formulating emission reduction strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0093] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0094] Figure 1 This is one of the flow diagrams of the method for predicting economic risks of heat wave-flood compound disasters based on moisture stress provided by the present invention;

[0095] Figure 2 It is a schematic diagram of the population and GDP forecast results for the future period provided by the present invention;

[0096] Figure 3 It is a structural schematic diagram of the flood risk prediction system provided by the present invention;

[0097] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0098] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0099] In view of the shortcomings of existing heat wave-flood combined disaster prediction technologies, such as generally less consideration of atmospheric humidity, failure to comprehensively consider the uncertainty of the global climate model itself, and failure to consider the impact of the abnormal sensitivity of the global climate model on the prediction results, the present invention proposes a prediction method for moisture stress heat wave-flood combined disasters driven by a multi-model ensemble.

[0100] Example 1

[0101] Figure 1This is one of the flow charts of the economic risk prediction method based on heat wave and flood compound disasters provided by the present invention, such as Figure 1 As shown, including:

[0102] Step 100: Collect meteorological and hydrological data sets, disease samples and socio-economic data of the study basin, and deduce the flood season of the study basin;

[0103] Step 200: using the Clausius-Clapeyron thermodynamic equation to determine the relative humidity and specific humidity of each grid point in the meteorological and hydrological data set, and training five machine learning models based on the meteorological and hydrological data set;

[0104] Step 300: Based on the quantile bias correction method, obtain the global climate model ensemble meteorological simulation data under the climate change scenario, use the corrected global climate model ensemble meteorological simulation data to train the five machine learning models, and obtain the daily runoff series under the future scenario;

[0105] Step 400: combining the global climate model ensemble meteorological simulation data with the five machine learning models to determine the weight parameters of each combined scenario in the multi-model weighted average method;

[0106] Step 500: Based on the relative humidity of each grid point calculated in step 200 and the meteorological and hydrological data set of step 100, the wet-bulb temperature is deduced, and the pathogenic high temperature stress index and the pathogenic high temperature stress index are deduced through the wet-bulb temperature and relative humidity, a disease-high temperature regression model for the flood season is established, and the disease-high temperature regression model parameters are solved;

[0107] Step 600: Based on the daily runoff series under the future scenario, a super-quantitative sampling method is used to extract flood events; then heat wave events are extracted based on the disease-high temperature regression model, and wet stress heat wave-flood compound disasters are extracted;

[0108] Step 700: Using the flood season precipitation as a covariate, a risk assessment model based on the Copula function and the same frequency combination is established, and based on the weight parameters of each combination scenario in the multi-model weighted average method, the GDP, population and agricultural risks caused by future wet stress heat wave-flood disasters are predicted, and the comprehensive socio-economic risks are calculated.

[0109] Specifically, we first collected meteorological and hydrological data sets, disease samples and socioeconomic data, including meteorological data from the European Centre for Medium-Range Weather Forecasts' fifth-generation atmospheric reanalysis data set (ERA5) reanalysis data set, daily flow series of hydrological stations controlling the study basin, meteorological data from the Earth system model in the historical period and under three shared socioeconomic pathways, and dynamic population, GDP and farmland area data from the shared socioeconomic pathway data set; it also included disease sample data involved in the study basin; and the precipitation series from the ERA5 data set was used to infer the flood season.

[0110] Then, relative humidity and specific humidity were derived, and five machine learning models were calibrated using the ERA5 reanalysis dataset. Based on the quantile bias correction method, the global climate model ensemble meteorological simulation series under the climate change scenario was obtained, and the corrected simulation series was used to drive the five machine learning models to obtain the daily runoff series under the future scenario.

[0111] Furthermore, based on the combined scenarios of the global climate model ensemble and the five machine learning models, the weight parameters of each combined scenario in the multi-model weighted average method were derived; based on the reanalysis data, the wet-bulb temperature and pathogenic high temperature stress index were extracted, a disease-high temperature regression model for the flood season was established, and the model parameters were solved.

[0112] Based on the daily runoff series under the future scenario, the super-quantitative sampling method was used to extract flood events; the high temperature index was further calculated based on the above-mentioned disease-high temperature regression model. If there were three consecutive high temperature events within 15 days before the flood event, it was defined as a wet stress heat wave-flood compound disaster. Finally, the flood season precipitation was used as a covariate to establish a two-dimensional risk assessment model based on Copula and the same frequency combination, and the results of multi-model weighted average were used to assess the socioeconomic risks caused by the future wet stress heat wave-flood compound disaster.

[0113] Taking the collected meteorological and hydrological data sets, the derived relative humidity and specific humidity as input, 5 machine learning models were calibrated, and quantile bias correction was used to obtain the simulated meteorological series. Through the combination of multiple models, the future daily runoff simulation series was obtained; further, the wet-bulb temperature and pathogenic high temperature stress index were extracted based on the reanalysis data, and a disease-high temperature regression model for the flood season was established, and the model parameters were solved; flood events were extracted through the future simulation series, and the wet stress heat wave-flood composite disaster was determined by further combining the disease-high temperature regression model. Based on the high temperature intensity and peak flow of the composite disaster, a time-varying Copula joint distribution function was constructed, and the same frequency combination scenario was derived to evaluate the risk of future wet stress heat wave-flood composite disasters.

[0114] The present invention combines the earth system model, human health, machine learning model, same-frequency combination scenario method, multi-model weighted average method and basin heat wave-flood compound disaster, to provide an important and highly operational reference basis for climate disaster 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.

[0115] Based on the above embodiment, step 100 includes:

[0116] Daily runoff series from control water level stations in the study area were obtained, and meteorological data from the fifth generation atmospheric reanalysis dataset ERA5 of the European Centre for Medium-Range Weather Forecasts were obtained;

[0117] Collect meteorological data from five pre-designed global climate models in the sixth phase of the International Coupled Model Intercomparison Project;

[0118] Get population and GDP data from the Shared Socioeconomic Pathways dataset.

[0119] Specifically, meteorological data such as precipitation, near-surface air pressure, 2m air temperature, 2m dew point temperature, shortwave radiation, near-surface wind speed, longwave radiation, and snowfall are obtained from the ERA5 reanalysis dataset. ERA5 is the fifth-generation atmospheric reanalysis dataset of the European Centre for Medium-Range Weather Forecasts, with a spatial resolution of 0.25°, providing hourly meteorological data covering the globe since 1940. The embodiment of the present invention obtains hourly precipitation, 2m air temperature, 2m dew point temperature, shortwave radiation, near-surface wind speed, longwave radiation, and snowfall data of the ERA5 dataset in the study basin during 1979-2022, and obtains a daily series after time scale conversion.

[0120] Meteorological data of 10 GCMs under three SSPs are collected: To estimate future climate scenarios, the 10 global climate models newly released by the sixth phase of the International Coupled Model Intercomparison Program (CMIP6) are used. CMIP6 uses a matrix framework of shared socioeconomic paths. In order to assess the socioeconomic risks that drought events may cause, the population data and GDP data of three shared socioeconomic paths, namely sustainable development (SSP1), competitive development (SSP3) and conventional development (SSP5), are considered, combined with the corresponding greenhouse gas emission scenario RCP, and the output data of three matrix frameworks, SSP126, SSP370 and SSP585, are used. The meteorological variables selected in the embodiment of the present invention are daily precipitation, daily average temperature, daily maximum temperature, daily minimum temperature, specific humidity, relative humidity, shortwave radiation, near-ground wind speed, longwave radiation and daily snowfall data. The historical period is set to 1985-2014, and the future period is set to 2015-2100.

[0121] The population and GDP data of the shared socio-economic pathway dataset are obtained, and the output data of three matrix frameworks, SSP126, SSP370 and SSP585, are used. The embodiment of the present invention uses a certain population policy estimation dataset released by the Collaborative Innovation Center for Meteorological Disaster Forecasting and Assessment of the School of Geography of a certain university. This dataset takes into account the results of previous domestic population and economic censuses, as well as the annual statistical yearbooks. The economic data is estimated by using the Cobb-Douglas model and the population-environment-development model to estimate the domestic socio-economic index from 2010 to 2100. This dataset has been widely used to assess the socio-economic impact of extreme hydrological events.

[0122] The farmland area data under historical periods and different SSP scenarios are obtained. The data are farmland area simulation data obtained based on the global dynamic vegetation model (LPJmL) under the ISIMIP3b framework. The data can provide the farmland area ratio of each grid point in the world, and then consider the area of ​​each grid point to obtain the farmland area.

[0123] Based on the daily precipitation data extracted from the ERA5 reanalysis dataset, the Thiessen polygon method was used to obtain the surface average daily precipitation series of the study basin, which was then integrated to obtain the monthly precipitation series. Finally, the cumulative precipitation for five consecutive months was solved, and the five consecutive months with the largest precipitation were divided into the flood season. The data on pathogenic and fatal cases in the flood season were obtained from disease and health management departments, hospitals, medical centers and other institutions in the study basin.

[0124] Based on the above embodiment, step 200 includes:

[0125] Determine the latent heat of vaporization constant, the water vapor gas constant, the first integral constant and the second integral constant in the Clausius-Clapeyron thermodynamic equation to obtain a nonlinear function of saturated water vapor pressure and air temperature;

[0126] Substitute the hourly 2m surface air temperature and dew point temperature of each grid point in ERA5 into the nonlinear function of saturated water vapor pressure and air temperature to obtain the hourly relative humidity, and then calculate the daily average relative humidity;

[0127] Substituting the hourly dew point temperature into the nonlinear function of saturated water vapor pressure and air temperature, and the ground pressure in ERA5, the hourly specific humidity is obtained, and then the daily average specific humidity is calculated;

[0128] Based on the daily precipitation, daily average 2m temperature, daily maximum 2m temperature, daily minimum 2m temperature, shortwave radiation, near-ground wind speed, longwave radiation and daily snowfall of each grid point in ERA5, as well as the derived daily average specific humidity and relative humidity as input, and taking the measured runoff data of the control hydrological station in the study basin as the target variable, 5 machine learning models were established, and the 5 machine learning models after calibration were obtained by optimizing the training through the gradient descent method, including:

[0129] Firstly, the hourly relative humidity and specific humidity are calculated based on the meteorological data of the ERA5 dataset;

[0130] The Clausius-Clapeyron thermodynamic equation can quantitatively describe the saturated water vapor pressure With temperature T The nonlinear relationship:

[0131] (1)

[0132] in, and are the first and second integral constants, which are 273.16 K and 611 Pa respectively; L v is the latent heat of vaporization constant, taken as 2.5×10 6 J kg -1 ; R v is the water vapor gas constant, which is 461 J kg -1 K -1 .

[0133] The dew point temperature represents the temperature at which the air is cooled to water vapor saturation under the condition of constant water vapor content and air pressure. Substituting it into the Clausius-Clapeyron equation can measure the actual water vapor pressure. T 2m ) and dew point temperature ( T dew ) are substituted into formula (1) to deduce the near-ground relative humidity .

[0134] Specific humidity q is the ratio of water vapor mass to total air mass mass, using ERA5 ground pressure p And dew point temperature, the formula is as follows:

[0135] (2)

[0136] After obtaining the hourly relative humidity and specific humidity of each grid point, the daily average relative humidity and specific humidity of each grid point are calculated, and then based on the measured daily runoff series of the hydrological station and the daily meteorological data of each grid point in the ERA5 reanalysis data set, 5 machine learning models are calibrated; the 5 machine learning models are specifically convolutional neural network, long short-term memory model, support vector machine, random forest and artificial neural network; the 5 machine learning models are trained by gradient descent method to obtain the 5 calibrated machine learning models. The calibration of machine learning models is a conventional technology in the field.

[0137] Specifically, in order to fully consider the spatial heterogeneity of hydrological processes, the present invention uses the meteorological variables of all grid points in the study basin as input for calibrating the machine learning model; traditional machine learning simulation methods generally calculate the regional average meteorological variables first, and then use them to calibrate the machine learning model. The method used in this embodiment is more in line with physical laws.

[0138] Based on the above embodiment, step 300 includes:

[0139] Calculating the difference between the output variable and the observed meteorological variable in the global climate model ensemble meteorological simulation data at each quantile, removing the difference in the output future scenario of the global climate model ensemble meteorological simulation data, and obtaining the corrected global climate model ensemble meteorological simulation data;

[0140] The calibrated five machine learning models are driven by the corrected global climate model ensemble meteorological simulation data to output the daily runoff series under the future scenario.

[0141] Specifically, the embodiment of the present invention obtains a meteorological simulation series under a climate change scenario based on a global climate model ensemble and a quantile bias correction method, and drives the 5 calibrated machine learning models to simulate a daily runoff series under a future scenario. Specifically, it includes:

[0142] The first step is to obtain a series of meteorological simulations under climate change scenarios based on a global climate model ensemble and quantile bias correction methods;

[0143] The differences between GCMs output variables and observed meteorological variables at each quantile (0.01-0.99) were calculated, and the differences were removed at each quantile of the GCMs output future scenarios to obtain the revised GCMs climate forecasts for the future.

[0144] Corrections for air temperature (as well as specific humidity, relative humidity, wind speed, shortwave radiation and longwave radiation) are as follows:

[0145] (3)

[0146] The correction for precipitation (and snowfall) is as follows:

[0147] (4)

[0148] in, T and P Respectively represent air temperature (specific humidity, relative humidity, wind speed, shortwave radiation and longwave radiation) and precipitation (snowfall), adj represents the corrected series, obs represents the observed data, ref and fut Represent the historical reference period and the future forecast period respectively, d Represents daily data, Q Represents each quantile.

[0149] In the second step, the calibrated simulated meteorological data (including daily precipitation, daily average 2m temperature, daily maximum 2m temperature, daily minimum 2m temperature, specific humidity, relative humidity, shortwave radiation, near-ground wind speed, longwave radiation and daily snowfall) are used to drive the five machine learning models calibrated in step 200 to simulate the hydrological process under future scenarios. The meteorological data under the climate change scenario calibrated in the first step are input into the five calibrated machine learning models to simulate the daily runoff series under the future climate change scenario.

[0150] Based on the above embodiment, step 400 includes:

[0151] Determine, based on the total number of multiple combined scenarios and any combined scenario weight parameter, that the sum of the combined scenario weight parameters is 1;

[0152] Specifically, the embodiment of the present invention uses five machine learning models to deduce the weight parameters of the multi-model weighted average method. Specifically including:

[0153] There are January to December in a year. In the weight derivation scheme, the weight of each month in different years is the same, so the weights of January to December need to be deduced separately, as follows:

[0154] For each month, the weight parameters of the 5 machine learning models satisfy:

[0155] (5)

[0156] in, i Indicates the first i A machine learning model; Indicates i The weight parameters of the machine learning model LM Indicates the number of machine learning models, which is 5;

[0157] The weights are calculated using the following formula:

[0158] (6)

[0159] Among them, RB i Characterization i The relative deviation between the runoff simulated by the machine learning model and the measured runoff; RMS i No. i The absolute deviation between the simulated runoff from the machine learning model and the measured runoff.

[0160] Based on the above embodiment, step 500 includes:

[0161] Based on the relative humidity of each grid point calculated in step 200 and the air temperature data in step 100, the wet-bulb temperature is calculated; and the pathogenic high temperature stress index is calculated through the wet-bulb temperature and the relative humidity;

[0162] The Thiessen polygon method was used to obtain the average wet-bulb temperature and pathogenic high temperature stress index in the study basin;

[0163] Based on the acquired disease-related death data in the study basin, a disease-high temperature regression model in the flood season was established, and the model parameters were solved;

[0164] The specific steps include:

[0165] In the first step, for each grid point, the hourly wet-bulb temperature is calculated based on the hourly 2-meter air temperature provided by the ERA5 dataset and the hourly relative humidity calculated in step 200:

[0166] (7)

[0167] Where: RH is relative humidity, T dry is the average daily temperature, T w is the wet-bulb temperature to be derived, and atan is the inverse tangent function.

[0168] In the second step, for each grid point, the hourly wet-bulb temperature and relative humidity are used to further deduce the hourly pathogenic high temperature stress index:

[0169] (8)

[0170] Where: T h To derive the pathogenic high temperature stress index.

[0171] The third step is to calculate the daily average wet-bulb temperature and the daily average pathogenic high temperature stress index based on the hourly wet-bulb temperature and the hourly pathogenic high temperature stress index of each grid point, and then calculate the spatial average daily wet-bulb temperature and the pathogenic high temperature stress index of the study basin by the Thiessen polygon method.

[0172] The fourth step is to establish a disease-high temperature regression model during the flood season based on the disease and death case data during the flood season obtained in step 100, as follows:

[0173] (9)

[0174] Where: Indicates the first y The total number of illnesses and deaths during the annual flood season, DD y Indicates y Number of days in the annual flood season; Indicates y The flood season of the year dd The pathogenic high temperature stress index of the day, Indicates y The flood season of the year dd daily wet-bulb temperature; are parameters of the disease-high temperature regression model; the calculation result of step 100 is used in the flood season.

[0175] The least square method is used to solve equation (9) and the parameters are obtained. .

[0176] Based on the above embodiment, step 600 includes:

[0177] Firstly, 80% was selected as the identification threshold of flood events, and the 80% quantile of the daily runoff series in the historical period (1985-2014) was extracted as the flood standard. The daily runoff exceeding the threshold in the historical and future periods was considered a flood event, and the daily runoff under the scenario exceeding the threshold was used as the peak flow and the standard flood magnitude.

[0178] In the second step, the meteorological data of the global climate model under different shared socioeconomic pathways are used to deduce the daily average wet-bulb temperature and the daily average pathogenic high temperature stress index of each grid point, and the parameters of the disease-high temperature regression model are used to calculate the normalized high temperature index for the historical period and the future period, which is:

[0179] (10)

[0180] Where: HI is the normalized high temperature index.

[0181] In the third step, the 95% quantile of the normalized high temperature index series of the historical period is used as the threshold. Events exceeding the threshold are defined as high temperature events. If a high temperature event lasts for more than three days, it is defined as a high temperature heat wave event. When a high temperature heat wave event occurs within 15 days before a flood event, it is determined to be a wet stress heat wave-flood composite disaster. Then, the flood magnitude and the intensity of the high temperature heat wave in the composite disaster event can be extracted. In this embodiment, the intensity of the high temperature heat wave is defined as the cumulative normalized high temperature index in the high temperature heat wave event.

[0182] Based on the above embodiment, step 700 includes:

[0183] The first step is to construct the joint probability distribution function of high temperature heat wave intensity and flood magnitude based on the Copula function; based on the combination scenario of the same frequency combination and optimal composite disaster attributes, the optimal combination function is constructed.

[0184] For any scenario under the climate scenario, the Gumbel Copula function is selected as the joint probability distribution function of the heat wave intensity and flood magnitude, and the parameters of the Copula function are Replace with time-varying parameters :

[0185] (11)

[0186] in, is the Copula joint distribution function, Range is ; u t , v t Heat wave intensity D , Flood magnitude S Probability density function of the marginal distribution.

[0187] Based on the definition of Copula function, the non-uniform two-variable Copula function can be expressed as:

[0188] (12)

[0189] in, represents the time-varying joint distribution function of D and S; , and , Respectively D and S The time-varying marginal distribution function and time-varying parameters of the variable. Furthermore, the parameters of the time-varying Copula function are expressed as covariates w It is expressed as:

[0190] (13)

[0191] in, represents the link function of the copula function, when When (for GH Copula), ; b 0 , b 1 are the parameters of the model respectively; w It is the spatial average flood season precipitation of the study basin.

[0192] Then the same frequency combination function is:

[0193] (14)

[0194] in, represents a joint return period T or High temperature heat wave intensity d , Flood magnitude s The most likely combination of scenarios; μ is the average interval time of wet stress heat wave-flood composite disaster; T ANDThe return period is “and”, indicating that the intensity of high temperature heat waves and flood magnitude exceed the threshold at the same time.

[0195] The second step is to deduce the GDP, population and agricultural risks caused by the increase of moisture stress heat wave-flood compound disasters under future climate change based on the weighted average results of multiple models;

[0196] The Newton iteration method is used to solve formula (14) and the return period of a certain period in the historical period (1985-2014) is obtained ( T h ) corresponding to the intensity of high temperature heat waves and flood magnitude ( , ); further, using 30 years as a sliding window (consistent with the length of the historical period), the time-varying marginal distribution and Copula function of the future period (2015-2100) are constructed. , ) are substituted into the future time period k The time-varying distribution function of a sliding window is used to calculate the new return period .

[0197] After obtaining the return period for each combination scenario, the weighted average method is used to obtain M In the combined scenario k The average return period of the sliding window :

[0198] (15)

[0199] in, i Indicates a combined scenario; represents the weight parameter of the combined scenario; Indicates a combination scenario i Next k The return period of a sliding window.

[0200] like This indicates that k The composite disaster risk of the first window increases, and vice versa. k The parameters of the marginal distribution and joint distribution are derived using the data of 15 years before and after the center point in a time window.

[0201] Figure 2 The GDP and population data for the future period of the study area are given. Based on the socio-economic scenario estimates, the GDP, population and agricultural risks in the future period are measured by the following formula:

[0202] (16)

[0203] (17)

[0204] (18)

[0205] in, , and Respectively represent the population, GDP and agricultural risks caused by compound disasters, , and CROP k Respectively k Population, GDP, and farmland area in the year; I (•) is the indicator function, When it is, it is recorded as 1, otherwise it is recorded as 0; N 1 and N 2 Respectively represent the start and end years of the study period.

[0206] Finally, the comprehensive socio-economic risk is calculated using the results of GDP, population and agricultural risks:

[0207] (19)

[0208] in: Represents the comprehensive socio-economic risks affected by compound hazards; They represent the weights of population, GDP and agricultural risks respectively, and are determined according to the characteristics of socio-economic development in the study area.

[0209] It can be seen that the existing methods usually directly use the arithmetic mean or median of the combined scenario set to evaluate drought risk in the future period, without considering the contribution of different combined scenarios to the overall fitting results, which is somewhat unreasonable. The present invention not only considers the inconsistency caused by future climate change, but also considers the uncertainty of the combined scenario itself, quantifies the weight of the combined scenario in the overall fitting results, and can reduce the error of future risk assessment.

[0210] Example 2

[0211] The economic risk prediction system based on the combined disaster of heat wave and flood provided by the present invention is described below. The economic risk prediction system based on the combined disaster of heat wave and flood described below and the economic risk prediction method based on the combined disaster of heat wave and flood described above can be referenced to each other.

[0212] Figure 3 is a schematic diagram of the structure of an economic risk prediction system based on heat wave and flood combined disasters provided by an embodiment of the present invention. Figure 3As shown, it includes: a collection module 71, a training module 72, a correction module 73, a combination module 74, a building module 75, a calculation module 76 and a prediction module 77, wherein:

[0213] The acquisition module 71 is used to collect meteorological and hydrological data sets; the training module 72 is used to determine relative humidity and specific humidity using the Clausius-Clapeyron thermodynamic equation, and train the machine learning model based on the meteorological and hydrological data sets; the correction module 73 is used to obtain the meteorological simulation data of the global climate model set under the climate change scenario based on the quantile deviation correction method, and use the corrected meteorological simulation data of the global climate model set to drive the machine learning model to obtain the daily runoff process under the future scenario; the combination module 74 is used to combine the global climate model set with the machine learning model to determine the weight parameters of each combination scenario in the multi-model weighted average method; establish Module 75 is used to establish a disease-high temperature regression model in the flood season based on the acquired disease-cause-death data in the research basin, and to solve the model parameters; the calculation module 76 is used to extract flood events under the future climate change scenario through the hydrological process under the future scenario, and further extract the moisture stress heat wave-flood compound events to obtain the high temperature heat wave intensity and flood magnitude; the prediction module 77 is used to establish a two-dimensional risk assessment model based on the Copula function and the same frequency combination under inconsistent conditions, and predict the social population, GDP and agricultural risks caused by future compound disasters based on the weight parameters of each combination scenario in the multi-model weighted average method, and calculate the comprehensive socio-economic risks.

[0214] Example 3

[0215] Figure 4 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830 and a communication bus 840, wherein the processor 810, the communication interface 820 and the memory 830 communicate with each other through the communication bus 840. The processor 810 may call the logic instructions in the memory 830 to execute the composite disaster risk prediction method, which includes: collecting a meteorological and hydrological data set; using the Clausius-Clapeyron thermodynamic equation to determine relative humidity and specific humidity, and training a machine learning model based on the meteorological and hydrological data set; based on the quantile deviation correction method, obtaining the meteorological simulation data of the global climate model set under the climate change scenario, using the corrected meteorological simulation data of the global climate model set to train the machine learning model considering the hydrological process, and obtaining the daily runoff series under the future scenario; combining the global climate model set with the machine learning model to determine the multi-model weighted The weight parameters of each combined scenario in the averaging method; based on the acquired disease-related death data in the research basin, a disease-high temperature regression model for the flood season is established, and the model parameters are solved; flood events are extracted through the daily runoff series under the future scenario, and the wet-bulb temperature is further deduced to extract the wet stress heat wave-flood compound disaster, and the high temperature heat wave intensity and flood magnitude are obtained; the preset characteristic parameters in the water-heat coupling balance equation of the basin are used as covariates, and a two-dimensional risk assessment model based on Copula function and same-frequency combination under inconsistent conditions is established, and the social population, GDP and agricultural risks caused by future compound disasters are predicted based on the weight parameters of each combined scenario in the multi-model weighted averaging method.

[0216] 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 it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention is essentially 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, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a 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: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0217] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0218] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for 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 some parts of the embodiments.

[0219] 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 aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting economic risks based on heat wave and flood combined disasters, characterized by: The steps include: Step 100: Collect meteorological and hydrological data sets, disease samples and socio-economic data of the study basin, and deduce the flood season of the study basin; Step 100 includes: Obtain the daily runoff series of the control water level station in the study basin, and obtain the meteorological data of the ERA5 reanalysis data set. Obtain precipitation, near-surface air pressure, 2m air temperature, 2m dew point temperature, shortwave radiation, near-surface wind speed, longwave radiation, and snowfall data from the ERA5 reanalysis data set, and obtain the daily series data after time scale conversion; Collect meteorological data from five pre-designed global climate models in the sixth phase of the International Coupled Model Intercomparison Project; Access to population and GDP data from the Shared Socioeconomic Pathways dataset; Obtain cropland area data for historical periods and under different SSP scenarios; The SSP scenario includes three shared socio-economic pathways: sustainable development, competitive development, and conventional development; Obtain data on illness and death cases during the flood season from disease and health management departments, hospitals, and medical centers in the study basin; Based on the daily precipitation data extracted from the ERA5 reanalysis dataset, the Thiessen polygon method was used to obtain the average daily precipitation series of the study basin, which was then integrated to obtain the monthly precipitation series. Finally, the cumulative precipitation of any five consecutive months in a year was solved, and the five consecutive months with the largest precipitation in May were divided into the flood season. Step 200: using the Clausius-Clapeyron thermodynamic equation to determine the relative humidity and specific humidity of each grid point in the meteorological and hydrological data set, and training five machine learning models based on the meteorological and hydrological data set; Step 300: Based on the quantile bias correction method, obtain the global climate model ensemble meteorological simulation data under the climate change scenario, use the corrected global climate model ensemble meteorological simulation data to train the five machine learning models, and obtain the daily runoff series under the future scenario; Step 400: combining the global climate model ensemble meteorological simulation data with the five machine learning models to determine the weight parameters of each combined scenario in the multi-model weighted average method; Step 500: Based on the relative humidity of each grid point calculated in step 200 and the meteorological and hydrological data set of step 100, the wet-bulb temperature is deduced, and the pathogenic high temperature stress index is deduced through the wet-bulb temperature and relative humidity, a disease-high temperature regression model for the flood season is established, and the disease-high temperature regression model parameters are solved; Step 500 includes: In the first step, for each grid point, the hourly wet-bulb temperature is calculated based on the hourly surface 2m air temperature provided by the ERA5 reanalysis dataset and the hourly relative humidity calculated in step 200: Where: RH is the relative humidity, The average daily temperature at 2m is the estimated wet bulb temperature, atan is the inverse tangent function; In the second step, for each grid point, the hourly wet-bulb temperature and hourly relative humidity are used to deduce the hourly pathogenic high temperature stress index: Where: is the pathogenic high temperature stress index; The third step is to calculate the daily average wet-bulb temperature and the daily average pathogenic high temperature stress index based on the hourly wet-bulb temperature and the hourly pathogenic high temperature stress index of each grid point, and calculate the spatial average daily wet-bulb temperature and the daily pathogenic high temperature stress index of the study basin by the Thiessen polygon method; The fourth step is to establish a disease-high temperature regression model during the flood season based on the disease and death case data during the flood season obtained in step 100, as follows: Where: Indicates the first y The total number of illnesses and deaths during the annual flood season, Indicates y Number of days in the annual flood season; Indicates y The flood season of the year dd The pathogenic high temperature stress index of the day, Indicates y The flood season of the year dd daily wet-bulb temperature; is a parameter of the disease-high temperature regression model; the calculation result of step 100 is used in the flood season; The least squares method was used to solve the disease-high temperature regression model during the flood season and the parameters were obtained. ; Step 600: Based on the daily runoff series under the future scenario, a super-quantitative sampling method is used to extract flood events; then heat wave events are extracted based on the disease-high temperature regression model, and wet stress heat wave-flood compound disasters are extracted; The step 600 specifically includes: The first step is to select the identification threshold of flood events and extract a certain quantile of the daily runoff series in the historical period as the flood standard. The daily runoff exceeding the threshold in the historical period and future period is a flood event. The daily runoff under the scenario exceeding the threshold is used as the peak flow and the standard flood magnitude. In the second step, the meteorological data of the global climate model under different shared socioeconomic pathways are used to deduce the daily average wet-bulb temperature and the daily average pathogenic high temperature stress index of each grid point, and the parameters of the disease-high temperature regression model are used to calculate the normalized high temperature index for the historical period and the future period, which is: Where: HI is the normalized high temperature index; The third step is to use a certain quantile of the historical wet-bulb temperature series as a threshold. Events exceeding this threshold are defined as high temperature events. If a high temperature event lasts for more than three days, it is defined as a high temperature heat wave event. If a high temperature heat wave event occurs within 15 days before a flood event, it is determined to be a wet stress heat wave-flood composite disaster. The flood magnitude and the intensity of the high temperature heat wave in the composite disaster event can then be extracted. Step 700: Using the precipitation in the flood season as a covariate, a risk assessment model based on the Copula function and the same frequency combination is established, and based on the weight parameters of each combination scenario in the multi-model weighted average method, GDP, population and agricultural risks caused by future wet stress heat wave-flood disasters are predicted, and comprehensive socio-economic risks are calculated; Step 700 includes: The first step is to construct the joint probability distribution function of heat wave intensity and flood magnitude based on the Copula function; based on the combined scenarios of composite disaster attributes with the same frequency combination, the optimal combination function is constructed; For any scenario under the climate scenario, the Gumbel Copula function is selected as the joint probability distribution function of the heat wave intensity and flood magnitude, and the parameters of the Copula function are Replace with time-varying parameters : in, is the Copula joint distribution function, Range is ; , Heat wave intensity D , Flood magnitude S Probability density function of marginal distribution; Based on the definition of Copula function, the non-uniform two-variable Copula function is expressed as: in, represents the time-varying joint distribution function of D and S; C() is the abbreviation of Copula function, , Respectively D and S The time-varying marginal distribution function of the variable, , Respectively D and S time-varying parameters of variables; The parameters of the time-varying Copula function are expressed as covariates w It is expressed as: in, represents the link function of the copula function, when hour, ; , are the parameters of the model respectively; w To study the spatial average flood season precipitation in the basin; Then the same frequency combination function is: in, μ is the average interval time of wet stress heat wave-flood composite disaster; "and" return period, indicating that both the intensity of high temperature heat waves and the magnitude of floods exceed the threshold at the same time; represents a joint return period High temperature heat wave intensity d , Flood magnitude s The most likely combination scenario is obtained by solving the above formula; The second step is to deduce the GDP, population and agricultural risks caused by the increase of moisture stress heat wave-flood compound disasters under future climate change based on the results of multi-model weighted average; Newton iteration method is used to solve the same frequency combination function to obtain a certain return period in the historical period The corresponding heat wave intensity and flood magnitude ( , ); Use the historical period length as the sliding window to construct the time-varying marginal distribution and Copula function of the future period, and ( , ) are substituted into the future time period k The time-varying distribution function of a sliding window is used to calculate the new return period ; After obtaining the return period for each combination scenario, the weighted average method is used to obtain M In the combined scenario k The average return period of the sliding window : in, i Indicates a combined scenario; represents the weight parameter of the combined scenario; Indicates a combination scenario i Next k The return period of a sliding window; if This indicates that k The composite disaster risk of each window increases, and vice versa; GDP, population and agricultural risks in future periods are measured by the following formula: in, , and Respectively represent the population, GDP and agricultural risks caused by compound disasters, , and Respectively k Population, GDP, and farmland area in the year; I (•) is the indicator function, When it is, it is recorded as 1, otherwise it is recorded as 0; and Respectively represent the start and end years of the study period; Using the results of GDP, population and agricultural risks, the comprehensive socio-economic risk is calculated: in: Represents the comprehensive socio-economic risks affected by compound hazards; They represent the weights of population, GDP and agricultural risks respectively, and are determined according to the characteristics of socio-economic development in the study area.

2. The economic risk prediction method based on heat wave and flood combined disaster according to claim 1 is characterized by: Step 200 includes: Determine the latent heat of vaporization constant, the water vapor gas constant, the first integral constant and the second integral constant in the Clausius-Clapeyron thermodynamic equation to obtain a nonlinear function of saturated water vapor pressure and air temperature; Substitute the hourly 2m surface air temperature and hourly dew point temperature of each grid point in the ERA5 reanalysis data set into the nonlinear function of saturated water vapor pressure and air temperature to obtain the hourly relative humidity, and then calculate the daily average relative humidity; Substituting the hourly dew point temperature into the nonlinear function of saturated water vapor pressure and air temperature and the ground pressure in the ERA5 reanalysis data set to obtain the hourly specific humidity, and then calculating the daily average specific humidity; Based on the daily precipitation, daily average 2m temperature, daily maximum 2m temperature, daily minimum 2m temperature, shortwave radiation, near-ground wind speed, longwave radiation and daily snowfall at each grid point in the ERA5 reanalysis dataset, as well as the inferred daily average specific humidity and relative humidity as input, and taking the measured runoff data of the basin control hydrological station as the target variable, 5 machine learning models were established, the 5 machine learning models included convolutional neural network, long short-term memory model, support vector machine, random forest and artificial neural network; and the 5 machine learning models were calibrated by optimizing the training through the gradient descent method.

3. The economic risk prediction method based on heat wave and flood combined disaster according to claim 2 is characterized by: Step 300 includes: The first step is to obtain a global climate model ensemble meteorological simulation series data under a climate change scenario based on a global climate model ensemble and a quantile deviation correction method, calculate the difference between the output variable and the observed meteorological variable in the global climate model ensemble meteorological simulation series data at each quantile, remove the difference in the output future scenario of the global climate model ensemble meteorological simulation series data, and obtain the corrected global climate model ensemble meteorological simulation data; In the second step, the calibrated five machine learning models are driven by the corrected global climate model ensemble meteorological simulation data to output the daily runoff series under the future scenario.

4. The economic risk prediction method based on heat wave and flood combined disaster according to claim 3 is characterized by: Step 400 includes: The weight parameters of the multi-model weighted average method are derived based on five machine learning models, including: In the weight derivation scheme, the weights for January to December are deduced separately, as follows: For each month, the weight parameters of the 5 machine learning models satisfy: in, i Indicates the first i A machine learning model; Indicates i The weight parameters of the machine learning model LM Indicates the number of machine learning models, which is 5; The weight is calculated using the following formula : in, Characterization i The relative deviation between the runoff simulated by the machine learning model and the measured runoff; No. i The absolute deviation between the runoff simulated by the machine learning model and the measured runoff, and exp() represents the natural exponential function.

5. An economic risk prediction system based on heat wave and flood compound disasters, characterized in that: include: A collection module is used to collect meteorological and hydrological data sets; A training module, using the Clausius-Clapeyron thermodynamic equation to determine relative humidity and specific humidity, and training five machine learning models based on the meteorological and hydrological data set; A correction module, for obtaining global climate model ensemble meteorological simulation data under a climate change scenario based on a quantile bias correction method, and using the corrected global climate model ensemble meteorological simulation data to train the five machine learning models to obtain a daily runoff series under a future scenario; A combination module, used to combine the global climate model set with the machine learning model to determine the weight parameters of each combination scenario in the multi-model weighted average method; Establish a module to build a disease-high temperature regression model during the flood season based on the acquired disease-related death data in the study basin, and solve the model parameters; A calculation module is used to extract flood events under future climate change scenarios through the hydrological process under the future scenarios, further extract moisture stress heat wave-flood compound events, and obtain high temperature heat wave intensity and flood magnitude; A prediction module is used to establish a two-dimensional risk assessment model based on Copula functions and same-frequency combinations under non-consistent conditions, predict the population, GDP and agricultural risks caused by future compound disasters based on the weight parameters of each combination scenario in the multi-model weighted average method, and calculate the comprehensive social and economic risks; The economic risk prediction system based on the combined disaster of heat waves and floods is used to execute the steps in the economic risk prediction method based on the combined disaster of heat waves and floods described in any one of claims 1-4.

6. A device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the economic risk prediction method based on the combined disaster of heat wave and flood as described in any one of claims 1 to 4 is implemented.