Weather type and wind power analysis coefficient optimization method for estimating ET0

Abstract

The invention discloses a weather type and wind power analysis coefficient optimization method for estimating ET0, which is used for realizing ET0 estimation under the condition of public weather forecast data, and comprises the following steps: acquiring day-by-day meteorological observation data for many years, and acquiring weather forecast data of a plurality of weather stations distributed in a plurality of agricultural areas in the same period as the day-by-day meteorological observation data; the method comprises the following steps: based on agricultural region division, uniformly selecting meteorological stations in each agricultural region, collecting historical meteorological observation data for many years, and carrying out regionalization calibration on solar radiation parameters as and bs by adopting a least square regression method to obtain monthly and annual calibration values of the solar radiation calculation parameters as and bs in different agricultural regions; based on public resources, three public weather forecast information analysis strategies are provided, weather types and wind power analysis coefficients are optimized, accurate quantitative analysis from weather conditions to solar radiation amount and from wind power grades to wind speed is achieved, and low-cost and high-precision ET0 estimation is completed.

Description

Technical Field

[0001] This invention relates to an optimization method for estimating weather type and wind force resolution coefficients for ET0, which is based on public weather forecasts. Background Art Background Technology

[0002] ETO (Emergency Tolerance) estimation using weather forecast information allows for real-time assessment of crop water requirements, enabling timely adjustments to irrigation strategies. Weather forecast information is primarily divided into public weather forecasts and numerical weather prediction. Public weather forecasts mainly utilize historical data, statistical models, and meteorologists' experience to predict future weather changes at specific locations, typically including information such as maximum and minimum temperatures, weather conditions, and wind speed. Numerical weather prediction uses mathematical physics models and computer technology to predict future weather changes by solving a system of weather motion equations. It generally includes relevant weather variables such as temperature, vapor pressure, and solar radiation, and its data format is gridded. Weather forecast data for any point on the grid is obtained through interpolation.

[0003] Numerical weather forecasting (NWMA) offers higher spatial resolution and more meteorological parameters, but biases in initial values, simplification of equations, or low predictability of certain factors can all lead to forecast failures. Therefore, when forecasting ETO, it is necessary to check each variable and each station individually, or to use a combination of forecasts from different forecasting systems to improve the limitations of a single NWMA model. Furthermore, NWMA suffers from forecast inconsistency, meaning significant differences exist between two forecasts. This makes preprocessing for NWMA complex, hindering its widespread use. Public weather forecasts are readily available in common media, and the forecast information can be used for ETO calculation through digital analysis. Cai et al. (2005) proposed an analytical method to transform public weather forecasts into weather variables needed to calculate daily ETO using the Penman-Monteith (PM) equation (FAO-56). This method categorizes weather types into five categories: sunny, sunny turning cloudy, cloudy turning overcast, overcast with occasional showers, and continuous overcast with rain. It assumes that these five weather types are evenly distributed between zero radiation and clear-sky radiation (Rso), and analyzes the corresponding sunshine duration and solar radiation values. Wind speed is fixed at the median of the 2m altitude wind speed range given in the "Ground Meteorological Observation Specifications." Through this processing, weather forecast information is transformed into complete PM equation inputs, and the feasibility of the method is verified by comparing it with measured meteorological station data. This analytical method simplifies the calculation process, saves costs, and is easy to promote and apply. However, when analyzing weather conditions into sunshine duration or solar radiation, it uses fixed, empirical correspondences, resulting in poor regional adaptability.

[0004] The main shortcomings of using public weather forecasts to estimate ET0 are as follows:

[0005] 1) When public weather forecasts interpret weather conditions as solar radiation and wind force level as wind speed for irrigation management, there are certain deviations in different agricultural regions, which leads to a decrease in the accuracy of ET0 estimation.

[0006] 2) When calculating solar radiation Rs, the key parameters as and bs in the Angstrom formula are often not calibrated regionally. Using default values ​​for calculation under different climate and geographical conditions will result in estimation bias, which will lead to a decrease in the accuracy of ET0. Summary of the Invention

[0007] The purpose of this invention is to overcome the problems of insufficient parameter universality and low ET0 estimation accuracy in the existing public weather forecast analysis process, and to provide an optimized method for estimating ET0 based on weather type and wind force analysis coefficients, so as to achieve accurate ET0 estimation for different agricultural areas under public weather forecast data conditions.

[0008] To achieve the above objectives, the solution of the present invention is as follows:

[0009] An optimization method for estimating weather type and wind force resolution coefficients for ET0 is proposed to achieve accurate ET0 estimation under public weather forecast data conditions. This method includes acquiring daily meteorological observation data over many years and acquiring weather forecast data from multiple meteorological stations distributed across multiple agricultural areas concurrent with the daily meteorological observation data.

[0010] The meteorological observation data includes maximum temperature, minimum temperature, relative humidity, solar radiation, sunshine duration, and wind speed;

[0011] The public weather forecast data includes maximum temperature, minimum temperature, weather conditions, and wind speed and direction; the optimization method is:

[0012] Based on daily meteorological observation data and the measured value of solar radiation R s The net radiation value R of the crop surface in the ET0 estimation formula was calculated. n Then, the estimated value of ET0 under the observed data conditions is calculated using the ET0 estimation formula, and denoted as ET0- PM ;

[0013] Under the condition of public weather forecast data contemporaneous with daily meteorological observation data, the calculated value of solar radiation R is determined by the information analysis value of public weather forecast data. s ’ The net radiation value R of the crop surface in the ET0 estimation formula was calculated. n Then, the estimated value of ET0 under the condition of public weather forecast data is calculated by the ET0 estimation formula, and denoted as ET0- PWF ;

[0014] Comparison with ET0- PWF With ET0- PM The accuracy of ET0 estimation is evaluated, and the optimal analytical coefficients for weather type and wind force in multiple agricultural areas under public weather forecast data are obtained, thereby achieving high-precision ET0 estimation under public weather forecast data.

[0015] The scheme further includes: the net radiation value R of the crop surface n The calculation formula is:

[0016]

[0017] In the formula: R n It is the net radiation on the crop surface; R ns Net solar radiation; R nl Net output longwave radiation;

[0018] Wherein: the net radiation value R of the crop surface n The determination is made using the following two methods, resulting in the value denoted as ET0- PWF The estimated value of ET0 is denoted as ET0- PM Estimated ET0 value:

[0019] I. Under the conditions of meteorological observation data:

[0020]

[0021] ;

[0022] ;

[0023] ;

[0024] In the formula: R s R represents the measured value of solar radiation. a For zenith radiation; R so Solar radiation under clear skies; This refers to reflectivity or canopy reflectance coefficient. T is the Stefan-Boltzmann constant; max,k The highest absolute temperature value within 24 hours; T min,k The lowest absolute temperature value within 24 hours; e a This is the actual water vapor pressure; a s a is the regression constant; s +b s The zenith radiation reaching the ground on a clear day; n is the actual duration of sunshine; N is the maximum possible duration or number of hours of sunshine; G sc d is the solar constant; r This refers to the relative distance between the Earth and the Sun. The angle of the sun; The latitude is the local latitude. Angle at sunset;

[0025] II. Under the condition of public weather forecast data:

[0026] ;

[0027] ;

[0028] ;

[0029] In the formula: R s ’ This is the calculated value for solar radiation; R so Solar radiation under clear skies; R sc It is a completely negative solar radiation; For weather type analysis coefficients;

[0030] The ET0 estimation formula is:

[0031] ;

[0032] In the formula: R n It is the net radiation value of the crop surface; G is the soil heat flux density; T is the air temperature at a height of 2m; u2 is the wind speed at a height of 2m; e s It is the saturated vapor pressure; e a It is the actual water vapor pressure; e s -e a It is the saturated water vapor pressure difference; It is the slope of the temperature-vapor pressure curve; It is the thermometer constant; This refers to reflectivity or canopy reflectance coefficient. This is the Stefan-Boltzmann constant.

[0033] The scheme further includes: the method further includes: adjusting parameter a s and b s Regional calibration was performed: based on daily observation data from multiple meteorological stations, using sunshine duration n and solar radiation R. s Measured values, obtained through the formula Using R s / R a The least squares regression method with n / N was used to analyze the solar radiation parameter a in multiple agricultural areas. s and b s Regional calibration was performed to obtain the a values ​​for multiple agricultural regions at monthly and annual scales. s and b s Parameter value.

[0034] The solution further includes three methods for obtaining the optimal analytical coefficients for the weather type and wind force:

[0035] I. Under fixed wind speed conditions, two methods, Monte Carlo iteration and differential evolution, are used to achieve quantitative analysis of solar radiation under weather conditions and optimize the weather type analysis coefficients.

[0036] II. Under the premise of fixed weather type resolution coefficients, two methods, Monte Carlo iteration and differential evolution, are used to achieve quantitative analysis of wind force level to wind speed and optimize wind force resolution coefficients.

[0037] Third, under the condition of coordinated change of two factors, weather type and wind force level, particle swarm optimization algorithm and adaptive particle swarm optimization algorithm are introduced to optimize the analytical coefficients of weather type and wind force.

[0038] The solution further includes the following steps: Under fixed wind speed conditions, the method of using Monte Carlo iteration to quantitatively analyze solar radiation from weather conditions and optimize the weather type analysis coefficient includes:

[0039] The first step is to fix the wind speed to the median of the wind speed range at a height of 2m, determine its initial value range based on the empirical relationship between weather type and weather type resolution coefficient, and set the weather type resolution coefficient within the corresponding interval according to the weather type classification.

[0040] The second step is to randomly select a dataset in a uniform distribution within the given corresponding interval, conduct 5000 Monte Carlo simulations, and obtain 5000 candidate values ​​for the resolving coefficients of weather types.

[0041] The third step is to substitute each candidate value into the ET0 estimation formula to calculate the estimated ET0 value under the public weather forecast data conditions, compare it with the estimated ET0 value of the actual observation data at the station, and select the value with the smallest mean absolute error as the optimal resolution coefficient for this weather type.

[0042] The solution further includes the following steps: Under fixed wind speed conditions, the method of using differential evolution to quantitatively analyze solar radiation from weather conditions and optimize the weather type analysis coefficient includes:

[0043] The first step is to fix the wind speed as the median of the wind speed range at a height of 2m, and determine the global search interval of the weather type resolution coefficient based on the physical meaning of the weather type resolution coefficient.

[0044] The second step is to initialize the parameter population within the defined global search interval, generate candidate parameters through differential mutation, reorganize parameter fragments through crossover operation, and then substitute the generated experimental vector into the ET0 estimation formula to calculate the objective function value, that is, the average error between the ET0 estimated value and the measured value. The better parameters are retained through selection operation, and the above evolution operation is repeated until the iteration termination condition is met.

[0045] The third step is to substitute the final optimal weather type resolution coefficients into the ET0 estimation formula to calculate the ET0 value, and compare it with the estimated ET0 value of the actual observation data at the station to verify its fitting effect. This value is the global optimal parameter under the corresponding weather type.

[0046] The solution further includes the following steps: Under the condition of fixed weather type resolution coefficients, the method of using Monte Carlo iteration to achieve quantitative analysis of wind speed at wind force level and optimizing wind force resolution coefficients includes:

[0047] The first step is to fix the weather type resolution coefficient to the median of its initial interval, and determine the initial interval of the wind force resolution coefficient based on the wind speed range at a height of 2m in the "Ground Meteorological Observation Specification";

[0048] The second step is to randomly select a dataset within the given initial range of wind resolution coefficients according to a uniform distribution, and conduct 5000 Monte Carlo simulations to obtain 5000 candidate values ​​for wind resolution coefficients.

[0049] The third step is to substitute each candidate value into the ET0 estimation formula to calculate the estimated ET0 value under the public weather forecast data conditions, compare it with the estimated ET0 value of the actual observation data of the station, and select the value with the smallest mean absolute error as the optimal resolution coefficient for the wind force level.

[0050] The solution further includes the following steps: Under the condition of fixed weather type resolution coefficients, the method of using differential evolution to achieve quantitative analysis of wind force level to wind speed and optimizing wind force resolution coefficients includes:

[0051] The first step is to fix the weather type resolution coefficient to the median of its initial interval. Based on the physical meaning of the wind resolution coefficient, under different wind force levels, the global search interval of the wind resolution coefficient is determined based on the wind speed range at a height of 2m given in the "Ground Meteorological Observation Specification" to cover a reasonable physical interval.

[0052] The second step is to initialize the parameter population within the defined search interval, generate candidate parameters through differential mutation, reorganize the parameter fragments through crossover operation, and then substitute the generated experimental vector into the ET0 estimation formula to calculate the objective function value, that is, the average error between the ET0 estimated value and the measured value. The better parameters are retained through selection operation, and the above evolution operation is repeated until the iteration termination condition is met.

[0053] The third step is to substitute the final optimal wind force resolution coefficient into the ET0 estimation formula and compare it with the estimated ET0 value of the actual observation data at the station to verify its fitting effect. This value is the global optimal parameter under the corresponding wind force level.

[0054] The scheme further includes the following steps of the particle swarm optimization algorithm under the condition of coordinated change of both weather type and wind force level:

[0055] The first step is to determine the search range of the two parameters based on the empirical constraints between weather type and weather type resolution coefficient, and between wind force level and wind force resolution coefficient. The range of values ​​for weather type resolution coefficient is the range corresponding to 10 weather types, and the range of values ​​for wind force resolution coefficient is the range of wind speed at a height of 2m given in the "Ground Meteorological Observation Specification".

[0056] The second step is to initialize the particle swarm within the range of values ​​mentioned above and the wind speed at a height of 2m, set fixed inertial weights and learning factors, and update the velocity and position of the particles by combining the "individual historical best position + the global best position of the population"; substitute the parameter combination of each particle into the ET0 estimation formula, use the mean absolute error (MAE) as the fitness function, and iterate continuously for 50 rounds to complete the optimization.

[0057] The third step, after the iteration is completed, select the parameter combination with the smallest fitness MAE in the population as the optimal weather type and wind force resolution coefficient under the condition of "weather type + wind force level".

[0058] The scheme further includes the following steps of the adaptive particle swarm optimization algorithm under the condition of coordinated change of both weather type and wind force level:

[0059] The first step is to determine the search range of the two parameters based on the physical constraints between weather type and weather type resolution coefficient, and between wind force level and wind force resolution coefficient. The range of values ​​for weather type resolution coefficient is the range corresponding to 10 weather types, and the range of values ​​for wind force resolution coefficient is the range of wind speed at a height of 2m given in the "Ground Meteorological Observation Specification".

[0060] The second step involves initializing the particle swarm within a defined range of values ​​and wind speeds at a height of 2 meters, and adaptively adjusting key parameters during the iteration process: the inertia weight is linearly reduced from 0.9 to 0.4, while the learning factor is dynamically adjusted based on particle fitness. Particles update their velocity and position by combining "individual optimality + global optimality". The corresponding parameter combinations are substituted into the ET0 estimation formula, with MAE as the fitness function, and the optimization is completed by iterating 50 times.

[0061] The third step, after the iteration, is to select the parameter combination with the smallest fitness MAE in the population, substitute it into the ET0 estimation formula, and perform a fitting test with the ET0 estimated value of the actual observation data of the station, which is used as the optimal weather type and wind force resolution coefficient under the condition of "weather type + wind force level".

[0062] The beneficial effects of this invention are:

[0063] 1. This invention is based on agricultural zoning, with meteorological stations evenly distributed within each agricultural zone to collect historical meteorological observation data over many years. The least squares regression method is used to analyze the solar radiation parameter a. s and b s Regional calibration was performed to obtain the solar radiation calculation parameter a for different agricultural regions. s and b s Monthly and annual calibration values; this invention relies on public weather forecast information to optimize weather type and wind speed resolution coefficients, achieving low-cost, high-precision ETO estimation; it optimizes weather type and wind speed resolution coefficients through multiple algorithms such as Monte Carlo iteration, differential evolution, particle swarm optimization, and adaptive particle swarm optimization, significantly improving the calculation accuracy of solar radiation and wind speed; moreover, weather forecast information is a free public resource, easy to use, and can achieve high-precision ETO estimation without actual measured data, saving the cost of acquiring meteorological data or building automatic weather stations, and can be widely used.

[0064] 2. When converting public weather forecast information into the input factors required for the ETO estimation (FAO-56 Penman-Monteith) formula, the system optimizes the analysis process by considering two key influencing factors: weather type and wind force level. First, Monte Carlo iteration and differential evolution algorithms are used to optimize the analytical coefficients for weather type and wind force respectively, selecting the optimal analytical coefficients for each single factor. Then, Particle Swarm Optimization (PSO) and Adaptive Particle Swarm Optimization (APSO) algorithms are introduced to jointly optimize the analytical coefficients, considering the synergistic effect of both weather type and wind force level. By comparing the ETO estimation accuracy under different conditions, the optimal weather forecast information analysis strategy is selected, and the optimal analytical coefficients for weather type and wind force suitable for application in China's nine major agricultural regions are determined. Regionalized parameter construction is implemented to enhance system adaptability. This invention considers spatial variability, dividing China into nine major agricultural regions and calculating the solar radiation parameter a. s b s By regionalizing and calibrating weather types and wind force analysis coefficients, more accurate local parameters have been established, which can truly reflect the actual water demand of different agricultural areas, providing technical support for precision irrigation decisions and smart water-saving agriculture, and realizing refined management of water resources.

[0065] 3. This invention constructs an ETO estimation accuracy optimization method and system based on public weather forecast data, enabling high-precision ETO calculation even in the absence of measured meteorological data. After obtaining regional weather forecast information, crop and soil distribution data, this method can effectively estimate crop evapotranspiration in real time, achieving real-time forecasting of crop evapotranspiration over large areas and providing relatively reliable parameters for field irrigation management and decision-making.

[0066] The invention will be further explained in detail below with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0067] Figure 1 This is a technical roadmap for the present invention;

[0068] Figure 2 Solar radiation parameters a for China's nine major agricultural regions s and b s Calibration technology roadmap;

[0069] Figure 3 Public Weather Forecast Information Analysis Strategy - Technical Roadmap;

[0070] Figure 4 Public Weather Forecast Information Analysis Strategy Two: Technical Roadmap;

[0071] Figure 5 Public Weather Forecast Information Analysis Strategy Three: Technical Roadmap Detailed Implementation

[0072] This embodiment is an improvement on the existing method mentioned in the background art for converting public weather forecasts into the FAO-56 PM equation to calculate the weather variables needed for daily ETO. In the traditionally known ETO estimation formula, the Angstrom formula is usually used to calculate the measured solar radiation value R. s Furthermore, the net radiation value R on the crop surface was estimated. n This embodiment improves upon the previous one, proposing a calculated value of solar radiation R. s ’ The calculation formula was derived, and the weather type and wind force resolution coefficient were optimized in the process of converting weather conditions into solar radiation and wind force level into wind speed, so as to realize the estimation of ET0 under the condition of public weather forecast data.

[0073] An optimization method for estimating weather type and wind force resolution coefficients for ET0 is proposed to achieve accurate ET0 estimation under public weather forecast data conditions. This method includes acquiring daily meteorological observation data over many years and acquiring weather forecast data from multiple meteorological stations distributed across multiple agricultural areas concurrent with the daily meteorological observation data.

[0074] The meteorological observation data includes maximum temperature, minimum temperature, relative humidity, solar radiation, sunshine duration, and wind speed;

[0075] The public weather forecast data includes maximum temperature, minimum temperature, weather conditions, and wind speed and direction; the optimization method is:

[0076] Based on daily meteorological observation data and the measured value of solar radiation R s The net radiation value R of the crop surface in the ET0 estimation formula was calculated. n Then, the estimated value of ET0 under the observed data conditions is calculated using the ET0 estimation formula, and denoted as ET0- PM ;

[0077] Under the condition of public weather forecast data contemporaneous with daily meteorological observation data, the calculated value of solar radiation R is determined by the information analysis value of public weather forecast data. s ’ The net radiation value R of the crop surface in the ET0 estimation formula was calculated. n Then, the estimated value of ET0 under the condition of public weather forecast data is calculated by the ET0 estimation formula, and denoted as ET0- PWF ;

[0078] Comparison with ET0- PWF With ET0- PM The accuracy of ET0 estimation is evaluated, and the optimal analytical coefficients for weather type and wind force in multiple agricultural areas under public weather forecast data are obtained, thereby achieving high-precision ET0 estimation under public weather forecast data.

[0079] Wherein: the net radiation value R of the crop surface n The calculation formula is:

[0080] ;

[0081] In the formula: R n It is the net radiation of crop surface, MJ·m -2 ·d -1 ;R ns Net solar radiation, MJ·m -2 ·d -1 ;R nl For net output longwave radiation, MJ·m -2 ·d -1 ;

[0082] Wherein: the net radiation value R of the crop surface n The determination is made using the following two methods, resulting in the value denoted as ET0- PWF The estimated value of ET0 is denoted as ET0- PM Estimated ET0 value:

[0083] I. Under the conditions of meteorological observation data:

[0084] ;

[0085] ;

[0086] ;

[0087] ;

[0088] In the formula: R s This is the measured value of solar radiation, in MJ·m. -2 ·d -1 ;R a Zenithal radiation, MJ·m -2 ·d -1 ;R so For clear-sky solar radiation, MJ·m -2 ·d -1 ; This refers to reflectivity or canopy reflectivity. =0.23; The Stefan-Boltzmann constant is taken as 4.903 × 10⁻⁶. -9 MJK -4 ·m -2 ·d -1 ;T max,k The highest absolute temperature value within 24 hours (K = ℃ + 273.6); T min,k The lowest absolute temperature value within 24 hours (K = ℃ + 273.6); e a The actual water vapor pressure is expressed in kPa; a s a is the regression constant; s +b s The zenith radiation reaching the ground on a clear day; n is the actual duration of sunshine; N is the maximum possible duration or number of hours of sunshine; G sc G is the solar constant. sc =0.0820 MJ·m -2 ·min -1 ;d r This refers to the relative distance between the Earth and the Sun. The angle of the sun; The latitude is local, expressed in radians. Angle at sunset;

[0089] II. Under the condition of public weather forecast data:

[0090]

[0091] ;

[0092] ;

[0093] In the formula: R s ’ This is a calculated value for solar radiation, in MJ·m. -2 ·d -1 ;R so For clear-sky solar radiation, MJ·m -2 ·d -1 ;R sc For total negative solar radiation, MJ·m -2 ·d -1 ; For weather type analysis coefficients;

[0094] The ET0 estimation formula is:

[0095] ;

[0096] In the formula: R n G is the net radiation value of the crop surface, and G is the soil heat flux density, in MJ·m. -2 ·d -1 T is the air temperature at a height of 2m, in °C; u2 is the wind speed at a height of 2m, in m·s. -1 ;e s It is the saturated vapor pressure, kPa; e a This is the actual water vapor pressure, kPa; e s -e a It is the saturated water vapor pressure difference, in kPa; It is the slope of the temperature-vapor pressure curve, in kPa·℃. -1 ; It is the thermometer constant, kPa·℃ -1 ; The reflectance or canopy reflectance coefficient is taken as 0.23; The Stefan-Boltzmann constant is taken as 4.903 × 10⁻⁶. -9 MJK -4 ·m -2 ·d -1 For ET0 calculated on a daily basis, soil heat flux can be ignored.

[0097] Specifically, the acquisition of daily meteorological observation data over many years is obtained through the National Meteorological Center platform, and the acquisition of concurrent public weather forecast data from multiple meteorological stations distributed across multiple agricultural areas is obtained through the Weather Network platform.

[0098] Wherein: the solar tilt angle It relates to the ordinal number J of each day in the year, which is determined by the month number M and the day number d. If the month is less than 3, J = J + 2. If it is a leap year and the month is greater than 2, J = J + 1.

[0099] In the embodiment: the method further includes: adjusting parameter a s and b s Regional calibration was performed: based on daily observation data from multiple meteorological stations, using sunshine duration n and solar radiation R. s Measured values, obtained through the formula Using R s / R a The least squares regression method with n / N was used to analyze the solar radiation parameter a in multiple agricultural areas. s and b s Regional calibration was performed to obtain the a values ​​for multiple agricultural regions at monthly and annual scales. s and b s Parameter values. See Table 1 for examples of China's nine major agricultural regions.

[0100] Table 1. Angstrom parameters for China's nine major agricultural regions s and b s Monthly scale calibration value

[0101]

[0102] In this embodiment, there are three methods for obtaining the optimal analytical coefficients for the weather type and wind force. The optimal analytical coefficients for the weather type and wind force can be obtained using any one of these three methods. These three methods are:

[0103] I. Under fixed wind speed conditions, two methods, Monte Carlo iteration and differential evolution, are used to achieve quantitative analysis of solar radiation under weather conditions and optimize the weather type analysis coefficients.

[0104] II. Under the premise of fixed weather type resolution coefficients, two methods, Monte Carlo iteration and differential evolution, are used to achieve quantitative analysis of wind force level to wind speed and optimize wind force resolution coefficients.

[0105] Third, under the condition of coordinated change of two factors, weather type and wind force level, the analytical coefficients of weather type and wind force are optimized by introducing particle swarm optimization algorithm and adaptive particle swarm optimization algorithm.

[0106] The method for quantitatively analyzing solar radiation under fixed wind speed conditions using Monte Carlo iteration and optimizing weather type analysis coefficients includes the following steps:

[0107] The first step is to categorize the different weather types into ten categories and set the initial weather type resolution coefficients. The intervals are shown in Table 2; and with reference to the new version of the "Ground Meteorological Observation Specifications" of the China Meteorological Administration, the wind speed is fixed as the median of the wind speed range at a height of 2m, as shown in Table 3. The initial value range of the wind speed is determined based on the empirical relationship between weather type and weather type resolution coefficient, and the weather type resolution coefficient is set within the corresponding interval according to the weather type classification.

[0108] Table 2 Weather Type Classification and Corresponding Weather Type Resolution Coefficients interval

[0109]

[0110] Table 3 Wind force and corresponding wind speed at 10m and 2m above the ground.

[0111]

[0112] The second step is to randomly select a dataset in a uniform distribution within the given corresponding interval, conduct 5000 Monte Carlo simulations, and obtain 5000 candidate values ​​for the resolving coefficients of weather types.

[0113] The third step is to substitute each candidate value into the ET0 estimation formula to calculate the estimated ET0 value under the public weather forecast data conditions, compare it with the estimated ET0 value of the actual observation data at the station, and select the value with the smallest mean absolute error as the optimal resolution coefficient for this weather type.

[0114] The method for quantitatively analyzing solar radiation under fixed wind speed conditions using differential evolution and optimizing weather type analysis coefficients includes the following steps:

[0115] Step 1: Fix the wind speed to the median value of the wind speed range at a height of 2m, and determine the weather type resolution coefficient based on the physical meaning of the weather type resolution coefficient. The global search range is set to cover a reasonable physical range;

[0116] The second step is to initialize the parameter population within the defined global search interval, generate candidate parameters through differential mutation, reorganize parameter fragments through crossover operation, and then substitute the generated experimental vector into the ET0 estimation formula to calculate the objective function value, that is, the average error between the ET0 estimated value and the measured value. The better parameters are retained through selection operation, and the above evolution operation is repeated until the iteration termination condition is met.

[0117] The third step is to substitute the final optimal weather type resolution coefficients into the ET0 estimation formula to calculate the ET0 value, and compare it with the estimated ET0 value of the actual observation data at the station to verify its fitting effect. This value is the global optimal parameter under the corresponding weather type.

[0118] The method for quantitatively analyzing wind speed at a given wind force level using Monte Carlo iteration under a fixed weather type resolution coefficient, and optimizing the wind force resolution coefficient, includes the following steps:

[0119] The first step is to fix the weather type resolution coefficient to the median of its initial interval, and then determine the wind force resolution coefficient based on the wind speed range at a height of 2m in the "Ground Meteorological Observation Specification". The initial interval is shown in Table 3;

[0120] The second step involves randomly selecting a dataset within a given initial range of wind resolving coefficients using a uniform distribution, and conducting 5000 Monte Carlo simulations to obtain 5000 wind resolving coefficients. Candidate values;

[0121] The third step is to substitute each candidate value into the ET0 estimation formula to calculate the estimated ET0 value under the public weather forecast data conditions, compare it with the estimated ET0 value of the actual observation data of the station, and select the value with the smallest mean absolute error as the optimal resolution coefficient for the wind force level.

[0122] The method for optimizing wind force resolving coefficients by using differential evolution to achieve quantitative analysis of wind force level to wind speed under fixed weather type resolving coefficients includes the following steps:

[0123] The first step is to fix the weather type resolution coefficient to the median of its initial interval. Based on the physical meaning of the wind resolution coefficient, under different wind force levels, the global search interval of the wind resolution coefficient is determined based on the wind speed range at a height of 2m given in the "Ground Meteorological Observation Specification" to cover a reasonable physical interval.

[0124] The second step is to initialize the parameter population within the defined search interval, generate candidate parameters through differential mutation, reorganize the parameter fragments through crossover operation, and then substitute the generated experimental vector into the ET0 estimation formula to calculate the objective function value, that is, the average error between the ET0 estimated value and the measured value. The better parameters are retained through selection operation, and the above evolution operation is repeated until the iteration termination condition is met.

[0125] The third step is to substitute the final optimal wind force resolution coefficient into the ET0 estimation formula and compare it with the estimated ET0 value of the actual observation data at the station to verify its fitting effect. This value is the global optimal parameter under the corresponding wind force level.

[0126] The steps of the particle swarm optimization algorithm under the condition of coordinated change of two factors: weather type and wind force level include:

[0127] The first step is to determine the search range of the two parameters based on the empirical constraints between weather type and weather type resolution coefficient, and between wind force level and wind force resolution coefficient. The range of values ​​for weather type resolution coefficient is the range corresponding to 10 weather types, and the range of values ​​for wind force resolution coefficient is the range of wind speed at a height of 2m given in the "Ground Meteorological Observation Specification".

[0128] The second step is to initialize the particle swarm within the range of values ​​mentioned above and the wind speed at a height of 2m, set fixed inertial weights and learning factors, and update the velocity and position of the particles by combining the "individual historical best position + the global best position of the population"; substitute the parameter combination of each particle into the ET0 estimation formula, use the mean absolute error (MAE) as the fitness function, and iterate continuously for 50 rounds to complete the optimization.

[0129] The third step, after the iteration is completed, select the parameter combination with the smallest fitness MAE in the population as the optimal weather type and wind force resolution coefficient under the condition of "weather type + wind force level".

[0130] The steps of the adaptive particle swarm optimization algorithm under the condition of coordinated change of two factors: weather type and wind force level include:

[0131] The first step is to determine the search range of the two parameters based on the physical constraints between weather type and weather type resolution coefficient, and between wind force level and wind force resolution coefficient. The range of values ​​for weather type resolution coefficient is the range corresponding to 10 weather types, and the range of values ​​for wind force resolution coefficient is the range of wind speed at a height of 2m given in the "Ground Meteorological Observation Specification".

[0132] The second step involves initializing the particle swarm within a defined range of values ​​and wind speeds at a height of 2 meters, and adaptively adjusting key parameters during the iteration process: the inertia weight is linearly reduced from 0.9 to 0.4, while the learning factor is dynamically adjusted based on particle fitness. Particles update their velocity and position by combining "individual optimality + global optimality". The corresponding parameter combinations are substituted into the ET0 estimation formula, with MAE as the fitness function, and the optimization is completed by iterating 50 times.

[0133] The third step, after the iteration, is to select the parameter combination with the smallest fitness MAE in the population, substitute it into the ET0 estimation formula, and perform a fitting test with the ET0 estimated value of the actual observation data of the station, which is used as the optimal weather type and wind force resolution coefficient under the condition of "weather type + wind force level".

[0134] The above method is an improvement on the existing method mentioned in the background art for converting public weather forecasts into the FAO-56 PM equation to calculate the weather variables needed for daily ETO. In the traditional and well-known ETO estimation formula, the measured value of solar radiation R is usually calculated using the Angstrom formula. sFurthermore, the net radiation value R on the crop surface was estimated. n This embodiment improves upon the previous one, proposing a calculated value of solar radiation R. s ’ The calculation formula was developed, and the weather type and wind force resolution coefficient were optimized during the conversion of weather conditions into solar radiation and wind force level into wind speed, thus realizing ET0 estimation under weather forecast data conditions.

[0135] The following is in accordance with Figure 1 The technical route shown, taking China as an example, will be further explained in eight steps:

[0136] Step 1: Agricultural Area Division and Site Layout Plan:

[0137] The ETO forecasting study covers the entire territory of China. my country is a vast country with significant differences in climate and geographical characteristics across different regions. Based on the "Comprehensive Agricultural Zoning of China" published by the National Agricultural Zoning Committee, China is divided into nine agricultural regions: Gansu-Xinjiang Region, Inner Mongolia and Great Wall Area, Northeast Region, Qinghai-Tibet Region, Loess Plateau Region, Huang-Huai-Hai Region, Southwest Region, Middle and Lower Yangtze River Region, and South China Region. This zoning system matches agricultural management practices and is conducive to its application in integrated agricultural management. Thirty meteorological stations were selected for each agricultural region for subsequent research, totaling 270 stations nationwide. These stations cover a range from 75.98°E to 131.15°E and 18.23°N to 53.47°N, with altitudes ranging from 1.1m to 4507m.

[0138] Step 2, Data Acquisition:

[0139] Through the National Meteorological Center platform ( Download data from 1990 to 2023 from http: / / data.cma.cn / The daily meteorological observation data, spanning 34 years, generally includes maximum and minimum temperatures, relative humidity, solar radiation, sunshine duration, and wind speed. Weather forecast data for the same period can be downloaded from the Weather Network platform (https: / / www.tianqi.com / ), which typically includes maximum and minimum temperatures, weather conditions, and wind direction and force. Weather conditions and wind direction and force need to be interpreted as corresponding solar radiation and wind speed values ​​at a height of 2 meters.

[0140] Step 3: Solar radiation parameter a s and b s Regionalized calibration, see appendix. Figure 2 :

[0141] Without actual measurements of solar radiation R s Under the condition of available data, solar radiation is usually calculated using the Angstrom formula (Equation 2), where parameter a s and b s The default value (a) is usually used.s =0.25, b s =0.5). But parameter a s and b s It varies with different meteorological conditions and solar magnetic declination, therefore, in order to improve R s The accuracy of the calculation is determined when using Angstrom's formula for R. s When estimating, it is necessary to combine measured data with the parameter a. s and b s Perform correction;

[0142] Formula 1

[0143] Formula 2

[0144] In the formula: R a Zenithal radiation, MJ·m -2 ·d -1 G sc The solar constant is 0.0820 MJ·m. -2 ·min -1 ;d r This refers to the relative distance between the Earth and the Sun. The solar tilt angle is related to the ordinal number J of each day in the year. J can be determined by the month number M and the day number d. If the month is less than 3, J = J + 2. If it is a leap year and the month is greater than 2, J = J + 1. The latitude is local, expressed in radians. The angle at sunset. s The regression constant is the portion of zenith radiation reaching the ground during cloudy weather (n=0); a s +b s The portion of the zenith radiation reaching the ground on a clear day (n=N); n is the actual duration of sunshine; N is the maximum possible duration or number of hours of sunshine.

[0145] To increase solar radiation R s For calculation accuracy, this embodiment is based on daily observation data from 270 meteorological stations in nine major agricultural regions, utilizing sunshine duration n and solar radiation R. s Measured values, using R s / R a The least squares regression method with n / N was used to apply the Angstrom formula parameter a to the nine major agricultural regions of China. s and b s Regional calibration was performed to obtain the a values ​​of China's nine major agricultural regions at monthly and annual scales. s and b s Parameter values ​​are shown in Table 1;

[0146] The fourth step involves determining the optimal weather type analytical coefficients using Monte Carlo iteration and differential evolution methods: see appendix. Figure 3 Technical roadmap (only optimizing weather type resolution coefficients);

[0147] 1. First, different weather types are categorized into ten major types. Initial weather type analytical coefficient ω intervals are set (see Table 2). Referring to the new version of the China Meteorological Administration's "Ground Meteorological Observation Specifications," wind speed is fixed at the median of the 2m height wind speed range (see Table 3). The optimal weather type analytical coefficients for China's nine major agricultural regions are solved using both Monte Carlo iteration and differential evolution methods, defined as follows: MC-opt and DE-opt ;

[0148] Monte Carlo method for optimizing weather type analytical coefficients The optimization process involves obtaining numerical simulation results through repeated random sampling within a given initial range, and then randomly selecting parameters that best fit the measured values. The optimization steps include: ① Analyzing the weather type and its resolving coefficients. The initial value range is determined by empirical relationships, and the weather type classification table (see Table 2) is used to further classify the values. Set within the corresponding range, such as when the weather type is "sunny". The interval is (0.8, 1.0); ② In the given... A dataset was randomly selected from the range using a uniform distribution, and 5000 Monte Carlo simulations were performed to obtain 5000 datasets. Candidate values; ③ Each candidate Substitute the values ​​into the FAO-56 PM formula to calculate ET0, and compare them with the measured ET0 values ​​at the stations. Select the value with the smallest mean absolute error. The value is used as the optimal analytical coefficient for this weather type.

[0149] Differential evolution method for optimizing weather type analytical coefficients It is a global search mechanism based on swarm intelligence. Through evolutionary operations such as mutation, crossover, and selection on candidate solutions within the parameter space, the population is iteratively updated generation by generation to select and obtain the globally optimal parameter values. The optimization steps include: ① Based on the weather type resolvable coefficients... The physical meaning is determined The global search range is within the interval (0.0, 2.0) to cover a reasonable physical range; ② In the determined Initialize the parameter population within the range, generate candidate parameters through differential mutation, and then recombine parameter fragments through crossover. Subsequently, substitute the generated experimental vectors into the FAO-56 PM formula to calculate the objective function value, i.e., the average error between the estimated and measured ET0 values. Select the best parameters through a selection operation, and repeat the above evolutionary operation until the iteration termination condition is met; ③ The final optimal... Substituting the values ​​into the FAO-56 PM formula, the ET0 value was calculated, and compared with the measured ET0 values ​​at the site to verify its fitting effect. The value is the globally optimal parameter for the corresponding weather type;

[0150] 2. Assume the above 10 weather forecast scenarios are distributed across clear-sky radiation R. so and total anodizing radiation R sc Within this, a new solar radiation R was created. s ’ The estimation formula (Equation 5) will MC-opt and DE-opt Substitute the values ​​into the formulas to solve;

[0151] Formula 3

[0152] Formula 4

[0153] Formula 5

[0154] 3. The weather forecast data does not include humidity data, therefore, when calculating the actual water vapor pressure e... a At that time, using the lowest temperature T min Instead of dew point temperature T dew Perform calculations;

[0155] 4. Substitute the numerical values ​​of the above forecast information into the FAO-56 PM formula (Equation 6) to solve for ET0, denoted as ET. 0-PWF Substitute the observation values ​​from the meteorological station during the same period into the FAO-56 PM formula to solve for ET0, denoted as ET. 0-PM ;

[0156] Formula 6

[0157] Formula 7

[0158] Formula 8

[0159] Formula 9

[0160] In the formula: R n It is the net radiation value of the crop surface, MJ·m -2·d -1 ;R ns Net solar radiation, MJ·m -2 ·d -1 ;R nl For net output longwave radiation, MJ·m -2 ·d -1 G is the soil heat flux density, in MJ·m³. -2 ·d -1 T is the air temperature at a height of 2m, in °C; u2 is the wind speed at a height of 2m, in m·s. -1 ;e s It is the saturated vapor pressure, kPa; e a This is the actual water vapor pressure, kPa; e s -e a It is the saturated water vapor pressure difference, in kPa; It is the slope of the temperature-vapor pressure curve, in kPa·℃. -1 ; It is the thermometer constant, kPa·℃ -1 ; The reflectance or canopy reflectance coefficient is taken as 0.23; The Stefan-Boltzmann constant is taken as 4.903 × 10⁻⁶. -9 MJK -4 ·m -2 ·d -1 ;T max,k The highest absolute temperature value within 24 hours (K = ℃ + 273.6); T min,k This represents the lowest absolute temperature value within 24 hours; for ET0 calculated on a daily basis, soil heat flux can be ignored.

[0161] 5. ETO- under both Monte Carlo iteration and differential evolution conditions PWF With ET0- PM By comparing and evaluating the accuracy of ET0 estimation, the optimal weather type analytical coefficients for China's nine major agricultural regions under this condition were obtained.

[0162] The fifth step involves solving for the optimal wind force analytical coefficients using Monte Carlo iteration and differential evolution methods: see appendix. Figure 4 The technical roadmap for optimizing only the wind force resolution coefficient;

[0163] When the wind force level is converted to the wind speed at 2m height, the weather type resolution coefficient is first fixed to the median of its initial interval (see Table 2). Then, based on the wind speed range at 2m height in the "Ground Meteorological Observation Specification", the wind force resolution coefficient is optimized by Monte Carlo iteration and differential evolution methods. The accuracy of ET0 estimation under the two conditions is evaluated, and the specific values ​​of wind speed at 2m height under different wind force levels in the nine major agricultural areas of China under this condition are solved.

[0164] The steps for solving the wind force resolution coefficient using the Monte Carlo method include: ① Determining the initial interval of the wind force resolution coefficient based on the wind speed range at a height of 2m in the "Ground Meteorological Observation Specification" (see Table 3); ② Randomly sampling the dataset within the given wind speed range according to a uniform distribution, conducting 5000 Monte Carlo simulations to obtain 5000 candidate values; ③ Substituting each candidate value into the FAO-56PM formula to calculate ET0, and comparing it with the measured ET0 values ​​at the station, selecting the value with the smallest mean absolute error as the optimal resolution coefficient for that wind force level.

[0165] The steps of solving the wind force resolution coefficient using the differential evolution method include: ① Based on the physical meaning of the wind force resolution coefficient, under different wind force levels, determine the global search interval of the wind force resolution coefficient based on the wind speed range at a height of 2m given in the "Ground Meteorological Observation Specification" to cover a reasonable physical interval; ② Initialize the parameter population within the determined range, generate candidate parameters through differential mutation, and then reorganize the parameter fragments through crossover operations. Subsequently, substitute the generated experimental vector into the FAO-56 PM formula to calculate the objective function value, i.e., the average error between the estimated ET0 value and the measured value. Select the better parameters through selection operations and repeat the above evolutionary operations until the iteration termination condition is met; ③ Substitute the finally obtained optimal wind force resolution coefficient into the FAO-56PM formula to calculate the ET0 value, and compare it with the measured ET0 value at the station to verify its fitting effect. This value is the globally optimal parameter under the corresponding wind force level.

[0166] Step 6: Solve for the optimal weather type and wind force analytical coefficients using Particle Swarm Optimization (PSO) and Adaptive Particle Swarm Optimization (APSO) algorithms: See Appendix Figure 5 Simultaneously optimize the technical roadmap for weather type and wind force resolution coefficient;

[0167] Considering both weather type and wind force level, the optimal weather type and wind force analytical coefficients are solved using both PSO and APSO algorithms under the combined effect of the two factors. The ET0 estimation accuracy under the two algorithm conditions is compared.

[0168] When Particle Swarm Optimization (PSO) solves for optimal weather type and wind resolution coefficients, it simulates the collaborative behavior of birds foraging, combining individual and group experience to traverse the two-dimensional parameter space and select the optimal two-parameter combination that best fits the measured values. The optimization steps include: ① Based on weather type and weather type resolution coefficients... The empirical constraint relationship between wind force level and wind force analysis coefficient is used to determine the search range of the two parameters; The value range of is the range corresponding to 10 weather types, and the range of wind force resolution coefficient is the range of wind speed at a height of 2m given in the "Ground Meteorological Observation Specification". ② Initialize the particle swarm within the above dual parameter range, set fixed inertia weights and learning factors, and update the velocity and position of the particles by combining "individual historical best position + global best position of the population"; substitute the parameter combination of each particle into the FAO-56 PM formula to calculate the estimated value of ET0, and use the mean absolute error (MAE) as the fitness function, and continuously iterate for 50 rounds to complete the optimization. ③ After the iteration, select the parameter combination with the best fitness (i.e., the smallest MAE) in the population as the optimal weather type and wind force resolution coefficient under the condition of "weather type + wind force level".

[0169] When solving for the optimal weather type and wind force analytical coefficients, the Adaptive Particle Swarm Optimization (APSO) algorithm introduces a dynamic adjustment mechanism for key control parameters based on the standard Particle Swarm Optimization (PSO) framework. This achieves an adaptive balance between global exploration and local convergence, thereby efficiently searching and selecting the globally optimal combination of two parameters in a two-dimensional parameter space. The optimization steps include: ① Based on the weather type and weather type analytical coefficients... The physical constraint relationship between wind force level and wind force analysis coefficient is used to determine the search range of the two parameters; The value range of is the range corresponding to 10 weather types, and the range of wind speed resolution coefficient is the range of wind speed at a height of 2m given in the "Ground Meteorological Observation Specification". ② Initialize the particle swarm within the determined dual-parameter range, and adaptively adjust the key parameters during the iteration process: the inertia weight is linearly reduced from 0.9 to 0.4, and the learning factor is dynamically corrected according to the particle fitness. The particles update their velocity and position by combining "individual optimal + global optimal"; substitute the corresponding parameter combination into the FAO-56 PM formula to calculate the estimated value of ET0, use MAE as the fitness function, and repeat the iteration for 50 rounds to complete the optimization. ③ After the iteration, select the parameter combination with the best fitness (i.e., the smallest MAE) in the population, substitute it into the FAO-56 PM formula to calculate ET0, and perform a fitting test on the measured value, which is taken as the optimal weather type and wind speed resolution coefficient under the condition of "weather type + wind speed level".

[0170] Step 7: Optimal Weather Forecast Information Analysis Strategy

[0171] Compare the ET0 estimation accuracy under the above three analytical conditions, and output the optimal weather type and wind force analytical coefficient for China's nine major agricultural regions.

[0172] Step 8, ET0 calculation component accuracy evaluation:

[0173] In calculating ET0 using public weather forecast numerical values, the accuracy and applicability of each meteorological component are evaluated, such as actual water vapor pressure e.a wait.

[0174] The above-described embodiment of an optimization method for estimating weather type and wind force analytical coefficients for ET0 is based on agricultural zone division. Meteorological stations are evenly selected within each agricultural zone to collect historical meteorological observation data over many years. The least squares regression method is used to optimize the solar radiation parameter a. s and b s Regional calibration was performed to obtain the solar radiation calculation parameter a for different agricultural regions. s and b s Monthly and annual calibration values; relying on public resources, three public weather forecast information analysis strategies are proposed to optimize weather type and wind force analysis coefficients, achieve accurate quantitative analysis of weather conditions into solar radiation and wind force level into wind speed, and complete low-cost, high-precision ETO estimation.

Claims

1. An optimization method for estimating weather type and wind force resolution coefficients for ETO, used to achieve ETO estimation under public weather forecast data conditions, including obtaining daily meteorological observation data over many years and obtaining weather forecast data from multiple meteorological stations distributed in multiple agricultural areas concurrent with the daily meteorological observation data; The meteorological observation data includes maximum temperature, minimum temperature, relative humidity, solar radiation, sunshine duration, and wind speed; The public weather forecast data includes maximum temperature, minimum temperature, weather conditions, and wind speed and direction; its characteristic is that... The optimization method is as follows: Based on daily meteorological observation data and the measured value of solar radiation R s The net radiation value R of the crop surface in the ET0 estimation formula was calculated. n Then, the estimated value of ET0 under the observed data conditions is calculated using the ET0 estimation formula, and denoted as ET0- PM ; Under the condition of public weather forecast data contemporaneous with daily meteorological observation data, the calculated value of solar radiation R is determined by the information analysis value of public weather forecast data. s ’ The net radiation value R of the crop surface in the ET0 estimation formula was calculated. n Then, the estimated value of ET0 under the condition of public weather forecast data is calculated by the ET0 estimation formula, and denoted as ET0- PWF ; Comparison with ET0- PWF With ET0- PM The accuracy of ET0 estimation is evaluated, and the optimal analytical coefficients for weather type and wind force in multiple agricultural areas under public weather forecast data are obtained, thereby achieving high-precision ET0 estimation under public weather forecast data.

2. The optimization method according to claim 1, characterized in that, The net radiation value R of the crop surface n The calculation formula is: ； In the formula: R n It is the net radiation from the crop surface; R ns Net solar radiation; R nl Net output longwave radiation; Wherein: the net radiation value R of the crop surface n The determination is made using the following two methods, resulting in the value denoted as ET0- PWF The estimated value of ET0 is denoted as ET0- PM Estimated ET0 value: I. Under the conditions of meteorological observation data: ； ； ； ； In the formula: R s R represents the measured value of solar radiation. a For zenith radiation; R so Solar radiation under clear skies; This refers to reflectivity or canopy reflectance coefficient. T is the Stefan-Boltzmann constant; max,k The highest absolute temperature value within 24 hours; T min,k The lowest absolute temperature value within 24 hours; e a This is the actual water vapor pressure; a s a is the regression constant; s +b s The zenith radiation reaching the ground on a clear day; n is the actual duration of sunshine; N is the maximum possible duration or number of hours of sunshine; G sc d is the solar constant; r This refers to the relative distance between the Earth and the Sun. The angle of the sun; The latitude is the local latitude. Angle at sunset; II. Under the condition of public weather forecast data: ； ； ； In the formula: R s ’ This is the calculated value for solar radiation; R so Solar radiation under clear skies; R sc It is a completely negative solar radiation; For weather type analysis coefficients; The ET0 estimation formula is: ； In the formula: R n It is the net radiation value of the crop surface; G is the soil heat flux density; T is the air temperature at a height of 2m; u2 is the wind speed at a height of 2m; e s It is the saturated vapor pressure; e a It is the actual water vapor pressure; e s -e a It is the saturated water vapor pressure difference; It is the slope of the temperature-vapor pressure curve; It is the thermometer constant; This refers to reflectivity or canopy reflectance coefficient. This is the Stefan-Boltzmann constant.

3. The optimization method according to claim 2, characterized in that, The method further includes: adjusting parameter a s and b s Regional calibration was performed: based on daily observation data from multiple meteorological stations, using sunshine duration n and solar radiation R. s Measured values, obtained through the formula Using R s / R a The least squares regression method with n / N was used to analyze the solar radiation parameter a in multiple agricultural areas. s and b s Regional calibration was performed to obtain the a values ​​for multiple agricultural regions at monthly and annual scales. s and b s Parameter value.

4. The optimization method according to claim 1, characterized in that, The methods and strategies for obtaining the optimal analytical coefficients for weather type and wind force include three types, which are: I. Under fixed wind speed conditions, two methods, Monte Carlo iteration and differential evolution, are used to achieve quantitative analysis of solar radiation under weather conditions and optimize the weather type analysis coefficients. II. Under the premise of fixed weather type resolution coefficients, two methods, Monte Carlo iteration and differential evolution, are used to achieve quantitative analysis of wind force level to wind speed and optimize wind force resolution coefficients. Third, under the condition of coordinated change of two factors, weather type and wind force level, particle swarm optimization algorithm and adaptive particle swarm optimization algorithm are introduced to optimize the analytical coefficients of weather type and wind force.

5. The optimization method according to claim 4, characterized in that, The method for quantitatively analyzing solar radiation under fixed wind speed conditions using Monte Carlo iteration and optimizing weather type analysis coefficients includes the following steps: The first step is to fix the wind speed to the median of the wind speed range at a height of 2m, determine its initial value range based on the empirical relationship between weather type and weather type resolution coefficient, and set the weather type resolution coefficient within the corresponding interval according to the weather type classification. The second step is to randomly select a dataset in a uniform distribution within the given corresponding interval, conduct 5000 Monte Carlo simulations, and obtain 5000 candidate values ​​for the resolving coefficients of weather types. The third step is to substitute each candidate value into the ET0 estimation formula to calculate the estimated ET0 value under the public weather forecast data conditions, compare it with the estimated ET0 value of the actual observation data at the station, and select the value with the smallest mean absolute error as the optimal resolution coefficient for this weather type.

6. The optimization method according to claim 4, characterized in that, The method for quantitatively analyzing solar radiation under fixed wind speed conditions using differential evolution and optimizing weather type analysis coefficients includes the following steps: The first step is to fix the wind speed as the median of the wind speed range at a height of 2m, and determine the global search interval of the weather type resolution coefficient based on the physical meaning of the weather type resolution coefficient. The second step is to initialize the parameter population within the defined global search interval, generate candidate parameters through differential mutation, reorganize parameter fragments through crossover operation, and then substitute the generated experimental vector into the ET0 estimation formula to calculate the objective function value, that is, the average error between the ET0 estimated value and the measured value. The better parameters are retained through selection operation, and the above evolution operation is repeated until the iteration termination condition is met. The third step is to substitute the final obtained weather type and wind force optimal analytical coefficients into the ET0 estimation formula to calculate the ET0 value, and compare it with the ET0 estimated value of the actual observation data of the station to verify its fitting effect. This value is the global optimal parameter under the corresponding weather type.

7. The optimization method according to claim 4, characterized in that, The method for quantitatively analyzing wind speed at a given wind force level using Monte Carlo iteration under a fixed weather type resolution coefficient, and optimizing the wind force resolution coefficient, includes the following steps: The first step is to fix the weather type resolution coefficient to the median of its initial interval, and determine the initial interval of the wind force resolution coefficient based on the wind speed range at a height of 2m in the "Ground Meteorological Observation Specification"; The second step is to randomly select a dataset within the given initial range of wind resolution coefficients according to a uniform distribution, and conduct 5000 Monte Carlo simulations to obtain 5000 candidate values ​​for wind resolution coefficients. The third step is to substitute each candidate value into the ET0 estimation formula to calculate the estimated ET0 value under the public weather forecast data conditions, compare it with the estimated ET0 value of the actual observation data of the station, and select the value with the smallest mean absolute error as the optimal resolution coefficient for the wind force level.

8. The optimization method according to claim 4, characterized in that, The method for optimizing wind force resolving coefficients by using differential evolution to achieve quantitative analysis of wind force level to wind speed under fixed weather type resolving coefficients includes the following steps: The first step is to fix the weather type resolution coefficient to the median of its initial interval. Based on the physical meaning of the wind resolution coefficient, under different wind force levels, the global search interval of the wind resolution coefficient is determined based on the wind speed range at a height of 2m given in the "Ground Meteorological Observation Specification" to cover a reasonable physical interval. The second step is to initialize the parameter population within the defined search interval, generate candidate parameters through differential mutation, reorganize the parameter fragments through crossover operation, and then substitute the generated experimental vector into the ET0 estimation formula to calculate the objective function value, that is, the average error between the ET0 estimated value and the measured value. The better parameters are retained through selection operation, and the above evolution operation is repeated until the iteration termination condition is met. The third step is to substitute the final optimal wind force resolution coefficient into the ET0 estimation formula and compare it with the estimated ET0 value of the actual observation data at the station to verify its fitting effect. This value is the global optimal parameter under the corresponding wind force level.

9. The optimization method according to claim 4, characterized in that, The steps of the particle swarm optimization algorithm under the condition of coordinated change of two factors: weather type and wind force level include: The first step is to determine the search range of the two parameters based on the empirical constraints between weather type and weather type resolution coefficient, and between wind force level and wind force resolution coefficient. The range of values ​​for weather type resolution coefficient is the range corresponding to 10 weather types, and the range of values ​​for wind force resolution coefficient is the range of wind speed at a height of 2m given in the "Ground Meteorological Observation Specification". The second step is to initialize the particle swarm within the range of values ​​mentioned above and the wind speed at a height of 2m, set fixed inertial weights and learning factors, and update the velocity and position of the particles by combining the "individual historical best position + the global best position of the population"; substitute the parameter combination of each particle into the ET0 estimation formula, use the mean absolute error (MAE) as the fitness function, and iterate continuously for 50 rounds to complete the optimization. The third step is to select the parameter combination with the smallest fitness MAE in the population after the iteration, as the optimal weather type and wind force resolution coefficient under the condition of "weather type + wind force level".

10. The optimization method according to claim 4, characterized in that, The steps of the adaptive particle swarm optimization algorithm under the condition of coordinated change of two factors: weather type and wind force level include: The first step is to determine the search range of the two parameters based on the physical constraints between weather type and weather type resolution coefficient, and between wind force level and wind force resolution coefficient. The range of values ​​for weather type resolution coefficient is the range corresponding to 10 weather types, and the range of values ​​for wind force resolution coefficient is the range of wind speed at a height of 2m given in the "Ground Meteorological Observation Specification". The second step involves initializing the particle swarm within a defined range of values ​​and wind speeds at a height of 2 meters, and adaptively adjusting key parameters during the iteration process: the inertia weight is linearly reduced from 0.9 to 0.4, while the learning factor is dynamically adjusted based on particle fitness. Particles update their velocity and position by combining "individual optimality + global optimality". The corresponding parameter combinations are substituted into the ET0 estimation formula, with MAE as the fitness function, and the optimization is completed by iterating 50 times. The third step, after the iteration is completed, selects the parameter combination with the smallest fitness MAE in the population, substitutes it into the ET0 estimation formula, and performs a fitting test with the ET0 estimated value of the actual observation data of the station, which is taken as the optimal weather type and wind force resolution coefficient under the condition of "weather type + wind force level".

Images