Weather service digitization system

Abstract

The invention belongs to the technical field of weather prediction, and particularly relates to a weather service digitization system, which adopts an internet of things type weather sensor array to collect atmospheric temperature, air pressure, water vapor flux, wind speed and wind direction data as original data through a data acquisition module; a grid generation unit of a multi-scale grid processing module is used for generating a multi-scale basic grid used for processing different-scale atmospheric motion, and a grid adaptation unit calculates an atmospheric motion real-time feature weight, a numerical prediction error feedback weight and a corrected topographic feature weight based on the data processed in the step S1; according to the method, a grid size adjustment formula is introduced to adjust a grid, an adaptive target grid is determined, and then the grid is fitted through a database, so that the accuracy of grid fitting is improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of weather prediction, and particularly relates to a weather service digitization system. BACKGROUND

[0002] In the field of weather prediction, numerical prediction technology is one of the core technologies, which simulates and predicts atmospheric movement through the construction of a mathematical model to provide important data support for weather services. In the prior art, a single fixed scale grid is usually used for atmospheric movement data processing. For atmospheric movement with a larger or smaller spatial range, the mapping of data in the grid will be distorted, resulting in inaccurate basic data for subsequent prediction calculations. Secondly, in the mathematical model, the parameters are determined by experience or simple statistical methods, which lack deep mining and scientific verification of historical data, resulting in deviations between the parameter values and the actual atmospheric movement, and further causing large deviations between the prediction results and the actual situation. Therefore, the weather service digitization system is improved and designed. SUMMARY

[0003] In view of the above problems in the prior art, the weather service digitization system is provided to solve the problems in the background.

[0004] In order to solve the above technical problems, the application adopts the following technical solutions:

[0005] The weather service digitization system comprises a data acquisition module, a multi-scale grid processing module, a parameter optimization module, an equation set correction module, a multi-disciplinary data fusion module and a result output module.

[0006] The data acquisition module adopts an Internet of Things type meteorological sensor array to acquire original data related to atmospheric movement, and the original data comprises atmospheric temperature, air pressure, water vapor flux, wind speed and wind direction data.

[0007] The multi-scale grid processing module comprises a grid generation unit and a grid adaptation unit.

[0008] The parameter optimization module comprises a parameter database unit, a parameter screening unit, a parameter fitting unit and a parameter verification unit.

[0009] The equation set correction module is used to construct a correction factor of the differential process of atmospheric change, and the correction factor comprises a parameter correction factor and a historical feature correction factor.

[0010] The multi-disciplinary data fusion module is used to acquire meteorology and geography related data, establish a topographic feature database, and fuse and process the multi-disciplinary data and the original atmospheric movement data by using a data algorithm.

[0011] The result output module is configured to output the historical prediction deviation to the grid adaptation unit and output the atmospheric motion prediction results processed by the multi-scale grid processing module, the parameter optimization module, the equation set correction module and the multi-disciplinary data fusion module.

[0012] Preferably, the grid generation unit is configured to generate a plurality of basic grids of different scales, including a first scale for processing large-scale atmospheric motion of typhoon and large-scale rainfall, a second scale for processing mesoscale atmospheric motion of ordinary rainfall cloud system and cold front, and a third scale for processing small-scale atmospheric motion of local thunderstorm and urban heat island effect.

[0013] The grid adaptation unit is configured to calculate real-time atmospheric motion feature weights by normalization based on the real-time wind speed, water vapor flux and calculated pressure gradient obtained by the data acquisition module, and determine terrain feature weights and correction by calculating terrain undulation using DEM data in the terrain feature database of the multi-disciplinary data fusion module, and input the real-time atmospheric motion feature weights, numerical prediction error feedback weights and corrected terrain feature weights into a grid size adjustment formula to adjust the grid.

[0014] Preferably, the parameter database unit is configured to establish a SQL database and store atmospheric motion parameter numerical data of different regions, including dynamic parameters, thermodynamic parameters and water vapor parameters, and is equipped with an index design.

[0015] The parameter screening unit is configured to screen the contents in the SQL database through two levels of screening, the first level of screening is terrain screening, and the target historical grid data with terrain feature weights and grid deviation ≤0.1 in the SQL database is screened; the second level of screening is real-time atmospheric motion weight screening, and historical data with a deviation ≤0.15 from the real-time atmospheric motion weight of the target grid is screened.

[0016] The parameter fitting unit is configured to construct a differentiated fitting function by combining the three types of parameters, i.e., thermodynamic parameters, dynamic parameters and water vapor parameters, with grid features, and iteratively optimize the fitting function by using the LM algorithm.

[0017] The parameter verification unit is configured to verify the optimized parameters by using the historical backtracking of the WRF model.

[0018] Preferably, the equation set correction module is configured to mark available parameters, extract key features of medium and long-term atmospheric changes based on the ARIMA model and machine learning, construct dynamic correction factors according to grid scene classification, and fuse the correction factors with the atmospheric change differential equation set to obtain corrected atmospheric dynamic equations and atmospheric heat conduction equations.

[0019] Preferably, the atmospheric motion real-time feature weight calculation formula is: Wherein v max is the maximum wind speed of the region, v is the current wind speed collected, p max is the extreme pressure gradient of the region, p is the calculated pressure gradient, q is the water vapor flux collected, q max is the saturated water vapor flux of the region; the numerical prediction error feedback weight calculation formula is w2=1-e -k×ε , wherein ε is the historical prediction deviation, and the calculation formula is ε 原始 is the original deviation data of the previous 3 times of the same type of prediction extracted by the result output module every 24 hours, ε max is based on the meteorological business standard design, k is the calibration coefficient, when ε 原始 is greater than ε max , 1 is taken; the terrain relief degree calculation formula is Wherein h i is the elevation value of the i-th DEM unit in the grid, is the average elevation of all DEM units in the grid, and n is the number of DEM units in the grid:

[0020] Preferably, the parameter correction factor includes a kinetic parameter correction factor, a thermodynamic parameter correction factor, and a water vapor parameter correction factor.

[0021] Preferably, the multi-disciplinary data fusion module divides the terrain complexity into extremely flat, relatively flat, medium complex, relatively complex, and extremely complex based on the terrain relief degree, and outputs the terrain feature weight based on the terrain complexity, which is 0.1, 0.2, 0.3, 0.4, and 0.5 in turn.

[0022] Preferably, the atmospheric dynamic equation is The atmospheric heat conduction equation is Wherein f is the correction factor.

[0023] Preferably, it further comprises a system monitoring module, which is used to monitor the running state of the data acquisition module, the multi-scale grid processing module, the parameter optimization module, the equation set correction module, the multi-disciplinary data fusion module, and the result output module in real time.

[0024] A weather service processing method based on the weather service digital system of any one of claims 1-9, comprising the following steps:

[0025] S1: Collecting atmospheric temperature, pressure, water vapor flux, wind speed and wind direction data as original data by using a networked meteorological sensor array through the data acquisition module, identifying abnormal data by the 3σ principle, and supplementing missing data by the linear interpolation method;

[0026] S2: A grid generation unit of a multi-scale grid processing module is used to generate a multi-scale basic grid for processing different scale atmospheric movements, a grid fitting unit calculates atmospheric movement real-time feature weights, numerical prediction error feedback weights and corrected terrain feature weights based on the data processed in step S1, adjusts the grid by bringing in a grid size adjustment formula to determine the target grid fitting;

[0027] S3: A parameter screening unit screens historical parameter data similar to the current scene from the SQL database of the parameter database unit, a parameter fitting unit constructs a differentiated fitting function in combination with three types of parameters, i.e., thermodynamic parameters, kinetic parameters and water vapor parameters, and grid features, and obtains preliminary optimized parameters through iterative optimization by an LM algorithm, and a parameter verification unit adopts historical backtracking verification of the WRF mode, and if the deviation is less than 6%, the final optimized parameters are determined;

[0028] S4: An equation set correction module is used to attach grid numbers, parameter types and deviation value tags to each available parameter, extract key features of medium and long-term atmospheric changes based on an ARIMA model and machine learning, construct dynamic correction factors according to grid scene classification, and integrate the correction factors into the equation set of atmospheric prediction to generate corrected atmospheric dynamic equations and atmospheric heat conduction equations;

[0029] S5: A multi-disciplinary data fusion module is used to obtain meteorology and geography related data, establish a terrain feature database, calculate terrain relief degree and determine terrain feature weights and corrections, and fuse the multi-disciplinary data with the original data of step S1;

[0030] S6: After completing the atmospheric movement prediction calculation based on the fitting grid of step S2, the optimized parameters of step S3, the corrected equation set of step S4 and the fused data of step S5, a result output module is used to output the prediction results and store the results.

[0031] In the whole process of steps S1-S6, a system monitoring module is used to monitor the running state of each module in real time.

[0032] Compared with the prior art, the present application has the following beneficial effects:

[0033] Through the array of Internet of Things type meteorological sensors, five core data of atmospheric temperature, air pressure, water vapor flux, wind speed and wind direction are synchronously collected, the grid fitting unit calculates the atmospheric movement real-time feature weight w1 through normalization based on the real-time wind speed, water vapor flux and calculated air pressure gradient obtained by the data collection module, and the calculation formula is: Then based on the historical prediction deviation ε of the result output module, the numerical prediction error feedback weight w2 is calculated by using an exponential function, and the calculation formula is w2=1-e -k×ε Then the terrain feature weight w02 , combine the atmospheric motion real-time feature weight w1 and the numerical prediction error feedback weight w2, and substitute into the grid size adjustment formula: L=L0x(1-λ0w 02 -λ1w1-λ2w2), wherein λ1 is the atmospheric motion real-time feature weight coefficient, the more intense the real-time motion, the finer the grid, and the greater the value; λ0 and λ2 are the same, and finally the last grid size is obtained, so as to ensure that the grid size matches the actual geographical situation, avoid the motion simulation distortion caused by the terrain, then the parameter database unit is screened in two levels, the target historical grid data with the terrain feature weight w0 and w 02 The grid deviation is less than or equal to 0.1, and the historical data with the target grid real-time atmospheric motion weight w1 deviation less than or equal to 0.15 are screened, so that the historical model sample that can be fitted can be determined, then the parameter fitting unit is used to fit the historical model sample, avoid the deviation of single function fitting all scenes, at the same time, the LM algorithm is used to iteratively optimize the fitting parameters, prevent the abnormal physical meaning, ensure that the parameters meet the physical law of atmospheric motion, then the terrain fluctuation and atmospheric motion data are integrated through the multidisciplinary data fusion module, the influence of terrain on meteorological data is added to the model fitting, and the simulation accuracy is further improved, so as to ensure the stability and accuracy of the prediction model and improve the prediction accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0034] Figure 1 It is a framework diagram of the weather service digital system of the application;

[0035] Figure 2 It is a step flow chart of the weather service digital system of the application;

[0036] Figure 3 It is a parameter optimization module structure diagram in the embodiment of the weather service digital system of the application;

[0037] Figure 4 It is a three-dimensional structure diagram of the device of the multiscale grid processing module of the weather service digital system of the application;

[0038] Figure 5 It is a three-dimensional structure diagram of the equation correction module in the embodiment of the weather service digital system of the application. DETAILED DESCRIPTION

[0039] In order to make those skilled in the art better understand the application, the technical solutions of the application are further described below with reference to the drawings and embodiments.

[0040] The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual images. They should not be construed as limiting the scope of this patent. To better illustrate the embodiments of the present invention, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual dimensions of the product. It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0041] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components. In the description of the present invention, it should be understood that if terms such as "upper," "lower," "left," "right," "inner," and "outer" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting the present patent. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0042] In the description of this invention, unless otherwise explicitly specified and limited, the term "connection" or similar designation indicating a connection between components should be interpreted broadly. For example, it can refer to a fixed connection, a detachable connection, or an integral part; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can refer to the internal communication between two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0043] Example 1:

[0044] like Figures 1-5 The weather service digital system shown includes a data acquisition module, a multi-scale grid processing module, a parameter optimization module, an equation correction module, a multi-disciplinary data fusion module, and a result output module.

[0045] The data acquisition module uses an IoT-based meteorological sensor array to collect raw data related to atmospheric motion, including atmospheric temperature, air pressure, water vapor flux, wind speed, and wind direction.

[0046] The multi-scale mesh processing module includes a mesh generation unit and a mesh adaptation unit;

[0047] The parameter optimization module includes a parameter database unit, a parameter filtering unit, a parameter fitting unit, and a parameter verification unit.

[0048] The equation system correction module is configured to construct a correction factor of a differential process of atmospheric changes, and the correction factor includes a parameter correction factor and a historical feature correction factor.

[0049] The multidisciplinary data fusion module is configured to acquire meteorological and geographical data, establish a terrain feature database, and fuse the multidisciplinary data and the original atmospheric movement data by using a data algorithm.

[0050] The result output module is configured to output the historical prediction deviation to the grid adaptation unit, and output the atmospheric movement prediction results processed by the multi-scale grid processing module, the parameter optimization module, the equation system correction module and the multidisciplinary data fusion module.

[0051] The grid generation unit is configured to generate a plurality of basic grids of different scales, including a first scale, a second scale and a third scale, the first scale is used to process large-scale atmospheric movement of typhoon and large-scale rainfall, the second scale is used to process mesoscale atmospheric movement of ordinary rainfall cloud system and cold front, and the third scale is used to process small-scale atmospheric movement of local thunderstorm and urban heat island effect.

[0052] The grid adaptation unit is configured to calculate the real-time feature weight of atmospheric movement by normalization based on the real-time wind speed, water vapor flux and the calculated pressure gradient acquired by the data acquisition module, and determine the terrain feature weight and correction by calculating the terrain relief degree by using the DEM data in the terrain feature database of the multidisciplinary data fusion module, and input the real-time feature weight of atmospheric movement, the numerical prediction error feedback weight and the corrected terrain feature weight into the grid size adjustment formula to adjust the grid.

[0053] The parameter database unit is configured to establish an SQL database, store numerical data of atmospheric movement parameters of different regions, including dynamic parameters, thermodynamic parameters and water vapor parameters, and be equipped with an index design.

[0054] The parameter screening unit is configured to screen the contents in the SQL database by two-level screening, the first-level screening is terrain screening, and the target historical grid data with a terrain feature weight and a grid deviation ≤0.1 in the SQL database is screened; the second-level screening is real-time atmospheric movement weight screening, and the historical data with a deviation ≤0.15 from the real-time atmospheric movement weight of the target grid is screened.

[0055] The parameter fitting unit is configured to construct a differentiated fitting function by combining the three types of parameters, i.e., thermodynamic parameters, dynamic parameters and water vapor parameters, with the grid features, and iteratively optimize the fitting function by using the LM algorithm.

[0056] The parameter verification unit is configured to verify the optimized parameters by using the historical backtracking of the WRF mode.

[0057] The equation system correction module marks the available parameters, extracts key features of medium and long-term atmospheric changes based on ARIMA model and machine learning, constructs dynamic correction factors according to grid scene classification, fuses the correction factors with the atmospheric change differential equation system, and obtains the corrected atmospheric dynamic equation and atmospheric heat conduction equation.

[0058] The real-time atmospheric movement feature weight calculation formula is: Wherein v max is the regional historical maximum wind speed, v is the current wind speed collected, p max is the regional extreme air pressure gradient, p is the calculated air pressure gradient, q is the collected water vapor flux, q max is the regional saturated water vapor flux; the numerical prediction error feedback weight calculation formula is w2=1-e -k×ε , wherein ε is the historical prediction deviation, and its calculation formula is ε 原始 is the original deviation data of the previous three times of the same type of prediction in the current region extracted by the result output module every 24 hours, ε max is the deviation data of the previous three times of the same type of prediction in the current region extracted by the result output module every 24 hours, and k is the calibration coefficient, which is 1 when ε 原始 is greater than ε max ; the terrain relief degree calculation formula is Wherein h i is the elevation value of the i-th DEM unit in the grid, is the average elevation of all DEM units in the grid, and n is the number of DEM units in the grid:

[0059] The parameter correction factor includes a dynamic parameter correction factor, a thermodynamic parameter correction factor, and a water vapor parameter correction factor.

[0060] The multi-disciplinary data fusion module divides the terrain complexity into extremely flat, relatively flat, medium complexity, relatively complex, and extremely complex based on the terrain relief degree, and outputs the terrain feature weight based on the terrain complexity, which is 0.1, 0.2, 0.3, 0.4, and 0.5 in turn.

[0061] The atmospheric dynamic equation is The atmospheric heat conduction equation is Wherein f is the correction factor.

[0062] It also includes a system monitoring module, which is used to monitor the running state of the data acquisition module, the multi-scale grid processing module, the parameter optimization module, the equation system correction module, the multi-disciplinary data fusion module, and the result output module in real time. When a module fails, an alarm signal is sent and fault information is recorded, and a standby module is started to replace the failed module.

[0063] A weather service processing method based on the weather service digitization system of any one of claims 1-9, comprising the following steps:

[0064] S1: Collecting atmospheric temperature, pressure, water vapor flux, wind speed and direction data as raw data by using an array of Internet of Things type meteorological sensors through a data acquisition module, identifying abnormal data through the 3σ principle, and supplementing missing data using linear interpolation method;

[0065] S2: Generating a multi-scale base grid for processing different scale atmospheric movements by a grid generation unit of a multi-scale grid processing module, calculating atmospheric movement real-time feature weight, numerical prediction error feedback weight and corrected terrain feature weight based on the data processed in step S1 by a grid adaptation unit, adjusting the grid by a grid size adjustment formula, and determining the target grid adapted;

[0066] S3: Selecting similar historical parameter data from the SQL database of the parameter database unit by a parameter screening unit, constructing a differentiated fitting function combining thermodynamic parameters, kinetic parameters and water vapor parameters with grid features by a parameter fitting unit, obtaining preliminary optimized parameters by iterative optimization through the LM algorithm, and determining the final optimized parameters if the deviation is less than 6% by a parameter verification unit using historical backtracking verification of the WRF model;

[0067] S4: Attaching grid number, parameter type and deviation value tags to each available parameter by an equation system correction module, extracting key features of medium and long-term atmospheric changes based on ARIMA model and machine learning, constructing dynamic correction factors according to grid scene classification, and integrating the correction factors into the equation system of atmospheric prediction to generate corrected atmospheric dynamic equation and atmospheric heat conduction equation;

[0068] S5: Obtaining meteorology and geography related data by a multi-disciplinary data fusion module, establishing a terrain feature database, calculating terrain undulation and determining terrain feature weight and correction, and fusing multi-disciplinary data with raw data of step S1;

[0069] S6: Based on the adapted grid of step S2, the optimized parameters of step S3, the corrected equation system of step S4 and the fused data of step S5, completing atmospheric movement prediction calculation, outputting the prediction results by a result output module, and storing the results;

[0070] In the whole process of steps S1-S6, the system monitoring module monitors the running state of each module in real time.

[0071] The present application specifically collects various original data such as atmospheric temperature, air pressure, water vapor flux, wind speed and wind direction data by adopting an Internet of Things type weather sensor array, identifies abnormal data in the original data through the 3 sigma principle, supplements missing data through the linear interpolation method, calculates the air pressure gradient through the edge design through the atmospheric motion characteristic classification model, inputs the cleaned data into the classification model, and the grid generation unit classifies the atmospheric motion type by adopting the random forest algorithm. When the classification model outputs large-scale motion, the first scale grid, i.e., the 14km*14km-18km*18km grid, is selected to process atmospheric motion such as typhoon and large-scale rainfall. When the model outputs mesoscale motion, the second scale grid is selected according to the national standard of 10km*10km to process atmospheric motion such as ordinary rainfall cloud system and cold front passage. When the model outputs small-scale motion, the third scale, i.e., 6km*6km-8km*8km, is selected to process atmospheric motion such as local thunderstorm and urban heat island effect.

[0072] The grid adaptation unit more accurately adjusts the grid according to the adaptive weight, calculates the real-time characteristic weight w1 of the atmospheric motion based on the real-time wind speed, water vapor flux and the calculated air pressure gradient obtained by the data acquisition module, and the calculation formula is: wherein v max is the regional historical maximum wind speed, v is the current wind speed collected, p max is the regional extreme air pressure gradient, p is the calculated air pressure gradient, q is the collected water vapor flux, q max is the regional saturated water vapor flux, and the higher the calculated value is, the more intense the atmospheric motion is, indicating that a finer grid is needed. Then, based on the historical prediction deviation ε of the result output module, the exponential function is adopted to calculate the numerical prediction error feedback weight w2, and the calculation formula is w2=1-e -k×ε , wherein ε is the historical prediction deviation, and the calculation formula is ε 原始 is the original deviation data of the previous three times of the same type of prediction of the current region extracted by the result output module every 24 hours, ε max is the historical prediction deviation of the result output module, k is the calibration coefficient and is taken as 5, when ε 原始 is greater than ε max , 1 is taken, and the multi-disciplinary data fusion module: the topographic feature database established in the multi-disciplinary data fusion module takes the grid in the grid generation unit as the unit, calculates the terrain undulation degree through the DEM data, and the calculation formula is wherein h i is the elevation value of the i th DEM unit in the grid, is the average elevation of all DEM units in the grid, n is the number of DEM units in the grid, the terrain complexity is divided into extremely flat σ≤10, relatively flat 10<σ≤30, medium complex 30<σ≤50, relatively complex 50<σ≤80 and extremely complex σ>80 according to the terrain relief σ, the terrain feature weight w0 is output based on the terrain complexity, and the output is 0.1, 0.2, 0.3, 0.4, 0.5 in turn, then the terrain feature weight w0 of the current terrain is corrected based on the terrain type and auxiliary data, if more than 90% of the area in the grid is a unified terrain type, the basic weight of the terrain type is used for correction, and the formula is: 01 =w0+w 类型 , wherein w 类型 is a terrain type correction coefficient; the area weighting method is used for calculation in the mixed terrain, and the formula is: , wherein S j is the area ratio of the jth terrain, w j is the correction coefficient of the jth terrain, and m is the number of terrain types in the grid. 01 The terrain feature weight w 01 is further adjusted in combination with the mountain trend and the slope change and the like auxiliary data, when the mountain trend is perpendicular to the dominant airflow direction of the forecast period, 0.1-0.2 is added to the basis of w 02 , when it is parallel, it is not corrected, the slope difference value is calculated according to the DEM in the adjacent grid, so as to obtain the slope change rate, if the change rate is greater than 15%, it indicates that the terrain is steep and the airflow has a lifting trend, 0.05-0.1 is added to the basis of w 02 , if the change rate is less than or equal to 15%, it is not corrected, and finally the corrected terrain feature weight w 02 is obtained. 02 The terrain feature weight w 02 obtained is transmitted to the grid adaptation unit, then the terrain feature weight w 02 , the atmospheric motion real-time feature weight w1 and the numerical prediction error feedback weight w2 are substituted into the grid size adjustment formula: L=L0×(1-λ0w 02 -λ1w1-λ2w2), wherein λ1 is an atmospheric motion real-time feature weight coefficient, the more intense the real-time motion is, the finer the grid is, and the value is larger; λ0 and λ2 are the same, and finally the final grid size is obtained.

[0073] The parameter database unit in the parameter optimization module is established with an SQL database, and stores atmospheric motion parameter numerical data in different regions in nearly 20 years, including dynamic parameters, thermodynamic parameters and water vapor parameter data, and is equipped with an index design, the parameter screening unit screens the contents in the SQL database through two-stage screening, the first-stage screening is terrain screening, and the terrain feature weights w0 and w 02Target historical grid data with grid deviation ≤0.1; the second screening is real-time atmospheric motion weight w1, screening historical data with deviation ≤0.15 from the real-time atmospheric motion weight w1 of the target grid, and finally screening out historical model samples that can be fitted;

[0074] Parameter fitting unit: then, combined with the thermodynamic parameters, kinetic parameters and water vapor parameters, the three types of parameters are differentially fitted with the grid characteristics, when w1≤0.3 and the terrain has no requirement, a quadratic function is used for fitting, C x =a+bx+cx 2 , wherein x is the grid average temperature, the calculation formula of a is reflecting the base specific heat capacity at a temperature of 0℃, wherein T w is the temperature weight provided by the parameter database unit, v i is the specific heat capacity value of the i-th historical data, b reflects the linear change rate of specific heat capacity with temperature, which is weighted based on the linear fitting result of “temperature-specific heat capacity” of historical data, and the formula is: reflecting the linear change rate of specific heat capacity with temperature, wherein k i is the specific heat capacity-temperature linear regression slope of the i-th set of historical data, c reflects the nonlinear influence of temperature on heat capacity, and the formula is reflecting the nonlinear influence of temperature on heat capacity, k q is the quadratic regression coefficient of the i-th set of historical data; when w1>0.3 and w2>0.5 for complex terrain, an exponential function is used for fitting, k (x) =ae bx +c, wherein x is the grid wind speed, reflecting the turbulent base value when the wind speed is 0, k i is the turbulent coefficient of the i-th historical data, w 1i is the real-time atmospheric motion weight corresponding to the i-th data, w 0i is the terrain weight corresponding to the i-th data, and n is the total amount of screened data, wherein k ei indicates the “turbulence-wind speed” exponential regression coefficient of the i-th set of historical data, and indicates the sensitivity of the increase amplitude of wind speed to the turbulent coefficient, ensuring that the turbulent coefficient is not 0 when the wind speed is very small, wherein S is the set of historical data with wind speed less than or equal to 2 m / s; when w0≤0.3 and in the plain terrain, a power function is used for fitting, e s =ax b +c, x is the grid relative humidity, wherein reflecting the water vapor pressure base value when the relative humidity is 100%, wherein w hi is the humidity weight, e si is the saturated water vapor pressure of the i-th historical data, reflecting the degree of non-linear growth of water vapor pressure with humidity, where k pi is the power regression coefficient of the i-th group of historical data, w hi squares the impact of high humidity data, is the water vapor pressure compensation when the humidity is 0, T is the historical data set when the humidity is less than or equal to 10%, m is the data set size, and after fitting, the a, b, or c is iteratively optimized by the LM algorithm to prevent physical meaning abnormalities;

[0075] Parameter verification unit: the optimized parameters are verified by historical backtracking of WRF model, DEM data of target grid is obtained from multi-disciplinary data fusion module for WRF model terrain following coordinate setting to ensure the consistency of model terrain, YSU boundary layer scheme or WSM6 microphysical scheme is selected according to the terrain and motion characteristics of target grid to preliminarily optimize the parameters, the preliminarily optimized parameters are used to replace the default values of corresponding parameters in WRF model to obtain further fitting function, and then the bias is calculated where m is the number of verification samples, F j is the WRF forecast value of the j-th sample, O j is the radar observation value of the j-th sample, if the bias ε2 is less than 6%, the parameter accuracy meets the requirements, the parameter is marked as available, and is transmitted to the equation correction module for parameter assignment of atmospheric change differential equation;

[0076] Equation correction module: each available parameter is attached with a grid number-parameter type-bias value label to ensure that the correction factor is accurately bound with the grid scene, then based on ARIMA model and machine learning, the key features of medium and long-term atmospheric changes are extracted, then according to the data transmitted by the parameter verification unit, the dynamic correction factor f is constructed according to the grid scene classification, the dynamic correction factor f includes parameter correction factor f1 and historical feature correction factor f2, the parameter correction factor f1 is calculated according to the current grid scene classification; the formula of the dynamic correction factor f is where a is the turbulent coefficient of the further fitting function output by the parameter verification unit, a 默 is the default value in YSU boundary layer scheme or WSM6 microphysical scheme, and κ is the calibration coefficient; the thermodynamic parameter correction factor where κ2 is the calibration coefficient, c s , c 默 are the specific heat capacity of the further fitting function output by the parameter verification unit and the default value in YSU boundary layer scheme or WSM6 microphysical scheme, respectively; the water vapor parameter correction factor

[0077] The historical feature correction factor includes: seasonal feature correction factor f 2s and terrain cumulative correction factor f 2t, seasonal feature correction factor f 2s The calculation formula is: Wherein f s is the quarterly basis coefficient, SST 当前 is the sea surface temperature of the current forecast period, SST 历史平均 is the average sea surface temperature of the same period; terrain cumulative correction factor f 2t = 1 + μ × w0 × t, wherein t is the forecast length, μ is the terrain error cumulative coefficient, then the parameter correction factor f1 and the historical feature correction factor f2 are weighted and fused in different scenarios, f = 0.6f1 + 0.4f2 in the scenario of w0> 0.3 and w1> 0.5, f = 0.4f1 + 0.6f2 in the scenario of w0≤ 0.3 and w1≤ 0.5, then the correction factor f is fused with the atmospheric change differential equation set to obtain the corrected atmospheric dynamic equation: The corrected atmospheric heat conduction equation is obtained: Through optimization, the final output is the dynamic element data, thermodynamic element data and water vapor element data accurately bound with the grid scene.

[0078] The result output module generates GRIB2 format and JOSN format to adapt to different push scenarios, which are respectively used for pushing by the forecaster business system and pushing by the public platform.

[0079] The multi-disciplinary data fusion module integrates the raw data of atmospheric temperature, air pressure, water vapor flux, wind speed and wind direction collected by the Internet of Things meteorological sensor array, then calculates the terrain undulation based on DEM data in units of grid, divides the terrain undulation into five categories: extremely flat, relatively flat, medium complex, relatively complex and extremely complex, and outputs the initial terrain feature weight corresponding to each category, then generates the corrected terrain feature weight through unified terrain correction, mixed terrain correction, auxiliary feature fine tuning, and atmospheric motion real-time feature weight and numerical prediction error feedback weight, which provides data support for grid size adjustment and parameter fitting. The raw data is directly input into the atmospheric motion feature classification model as the basis for preliminary classification of grid size, and is associated with historical meteorological data in the parameter database unit to provide data support for subsequent grid size adjustment, and transmits the DEM data of the target grid to the parameter verification unit to ensure that the terrain simulated by the WRF model is consistent with the actual terrain, which provides terrain basis for the subsequent parameter accuracy judgment deviation ε2.

[0080] The above is only an embodiment of the present application, and relates to circuits and electronic components and modules, which are all prior art. Those skilled in the art can implement the present application without further description. The present application does not involve improvement of software and methods. Commonly known specific structures and characteristics in the scheme are not described in detail herein. Those skilled in the art know all common technical knowledge in the field of the present application before the filing date or the priority date, can know all prior art in the field, and have the ability to apply conventional experimental means before the date. Those skilled in the art can perfect and implement the present scheme based on their own ability under the guidance of the present application. Some typical known structures or known methods should not be an obstacle for those skilled in the art to implement the present application. It should be pointed out that, for those skilled in the art, a number of modifications and improvements can be made without departing from the structure of the present application. These should also be considered as the protection scope of the present application, and these will not affect the implementation effect and practicality of the patent.

Claims

1. A weather business digitization system characterized by: It includes a data acquisition module, a multi-scale grid processing module, a parameter optimization module, an equation system correction module, a multidisciplinary data fusion module, and a result output module; The data acquisition module uses an IoT-based meteorological sensor array to collect raw data related to atmospheric motion, including atmospheric temperature, air pressure, water vapor flux, wind speed, and wind direction. The multi-scale mesh processing module includes a mesh generation unit and a mesh adaptation unit; The parameter optimization module includes a parameter database unit, a parameter filtering unit, a parameter fitting unit, and a parameter verification unit. The equation correction module is used to construct correction factors for the differential process of atmospheric change, including parameter correction factors and historical feature correction factors. The multidisciplinary data fusion module is used to acquire meteorological and geographical data, establish a topographic feature database, and use data algorithms to fuse multidisciplinary data with raw atmospheric motion data. The result output module is used to output historical prediction bias to the grid adaptation unit and output atmospheric motion prediction results processed by the multi-scale grid processing module, parameter optimization module, equation correction module and multi-disciplinary data fusion module.

2. The weather operations digitization system of claim 1, wherein: The grid generation unit is used to generate base grids of various different scales, including a first scale, a second scale, and a third scale. The first scale is used to handle large-scale atmospheric motions such as typhoons and large-scale rainfall. The second scale is used to handle mesoscale atmospheric motions such as ordinary rain clouds and cold fronts. The third scale is used to handle small-scale atmospheric motions such as local thunderstorms and urban heat island effects. The grid adaptation unit is used to calculate the real-time atmospheric motion feature weights by normalizing the real-time wind speed, water vapor flux and calculated pressure gradient obtained by the data acquisition module, and at the same time, to calculate the terrain relief using DEM data in the terrain feature database of the multi-disciplinary data fusion module, determine the terrain feature weights and correct them, and then substitute the real-time atmospheric motion feature weights, numerical forecast error feedback weights and corrected terrain feature weights into the grid size adjustment formula to adjust the grid.

3. The weather operations digitization system of claim 1, wherein: The parameter database unit is equipped with an SQL database that stores numerical data of atmospheric motion parameters from different regions, including dynamic parameters, thermodynamic parameters, and water vapor parameters, and is also equipped with an index design. The parameter filtering unit filters the contents of the SQL database through two levels of filtering. The first level of filtering is terrain filtering, which filters the target historical grid data in the SQL database whose terrain feature weights and grid deviations are ≤0.

1. The second level of filtering is real-time atmospheric motion weight filtering, which filters historical data whose real-time atmospheric motion weights and deviations from the target grids are ≤0.

15. The parameter fitting unit constructs a differentiated fitting function by combining three types of parameters—thermodynamic parameters, kinetic parameters, and water vapor parameters—with grid features, and iteratively optimizes the fitting function using the LM algorithm. The parameter verification unit uses historical backtracking in WRF mode to verify the optimized parameters.

4. The weather operations digitization system of claim 1, wherein: The equation correction module marks the available parameters and extracts key features of medium- and long-term atmospheric changes based on the ARIMA model and machine learning. It constructs dynamic correction factors according to grid scene classification, and then fuses the correction factors with the differential equations of atmospheric changes to obtain the corrected atmospheric dynamic equations and atmospheric heat conduction equations.

5. The weather service digitalization system as described in claim 2, characterized in that: The atmospheric movement real-time feature weight calculation formula is: Wherein v max is the regional historical maximum wind speed, v is the current wind speed collected, p max is the regional extreme pressure gradient, p is the calculated pressure gradient, q is the collected water vapor flux, q max is the regional saturated water vapor flux; the numerical prediction error feedback weight calculation formula is w2=1-e -k×ε , wherein ε is the historical prediction deviation, and the calculation formula is ε 原始 is the original deviation data of the previous 3 times of the same type of prediction of the current region extracted by the result output module every 24 hours, ε max is the average deviation of the previous 3 times of the same type of prediction of the current region, k is a calibration coefficient, and when ε 原始 is greater than ε max , 1 is taken; the terrain relief degree calculation formula is Wherein h i is the elevation value of the i-th DEM unit in the grid, is the average elevation of all DEM units in the grid, and n is the number of DEM units in the grid.

6. The weather business digitalization system of claim 4, wherein: The parameter correction factors include kinetic parameter correction factors, thermodynamic parameter correction factors, and water vapor parameter correction factors.

7. The weather business digitalization system of claim 1, wherein: The multidisciplinary data fusion module classifies terrain complexity into extremely flat, relatively flat, moderately complex, relatively complex, and extremely complex based on terrain undulation, and outputs terrain feature weights based on terrain complexity, which are 0.1, 0.2, 0.3, 0.4, and 0.5 respectively.

8. The weather service digitalization system as described in claim 4, characterized in that: The atmospheric dynamic equation is The atmospheric heat conduction equation is where f is a correction factor.

9. The weather operations digitization system of claim 1, wherein: It also includes a system monitoring module, which is used to monitor the operation status of the data acquisition module, multi-scale grid processing module, parameter optimization module, equation correction module, multi-disciplinary data fusion module, and result output module in real time.

10. A weather service processing method based on the weather service digitization system according to any one of claims 1-9, characterized in that: Includes the following steps: S1: The data acquisition module uses an IoT-based meteorological sensor array to collect atmospheric temperature, air pressure, water vapor flux, wind speed and wind direction data as raw data. Abnormal data is identified by the 3σ principle, and missing data is supplemented by linear interpolation. S2: The grid generation unit of the multi-scale grid processing module generates a multi-scale basic grid for processing atmospheric motion at different scales. The grid adaptation unit calculates the real-time feature weights of atmospheric motion, the numerical prediction error feedback weights, and the corrected terrain feature weights based on the data processed in step S1. The grid size adjustment formula is then used to adjust the grid and determine the target grid for adaptation. S3: The parameter filtering unit filters historical parameter data similar to the current scene from the SQL database of the parameter database unit. The parameter fitting unit combines three types of parameters—thermodynamic parameters, kinetic parameters, and water vapor parameters—with grid features to construct a differentiated fitting function. The preliminary optimized parameters are obtained through iterative optimization using the LM algorithm. The parameter verification unit uses historical backtracking verification in the WRF mode. If the deviation is less than 6%, it is determined as the final optimized parameter. S4: The equation correction module attaches grid number, parameter type and deviation value label to each available parameter, and extracts key features of medium and long-term atmospheric changes based on ARIMA model and machine learning. Dynamic correction factors are constructed according to grid scene classification. The correction factors are integrated into the equation set of atmospheric prediction to generate the corrected atmospheric dynamic equation and atmospheric heat conduction equation. S5: Obtain meteorological and geographical data through the multidisciplinary data fusion module, establish a topographic feature database, calculate the topographic relief and determine and correct the topographic feature weights, and fuse the multidisciplinary data with the original data from step S1. S6: Based on the adapted grid from step S2, the optimized parameters from step S3, the corrected equations from step S4, and the fused data from step S5, after completing the atmospheric motion prediction calculation, the prediction results are output by the result output module and stored. Throughout the entire process of steps S1-S6, the system monitoring module monitors the operating status of each module in real time.

Images