GRAPES-REPS regional ensemble forecast method and system

Abstract

The present invention belongs to the field of numerical forecasting technology, and discloses a GRAPES-REPS regional ensemble forecasting method and system. The GRAPES-REPS regional ensemble forecasting method includes: 0.1 degree horizontal resolution GRAEPS-REPS ensemble forecasting test parameters and model determination; GRAPES-REPS lateral boundary perturbation construction; China GRAPES-REPS regional ensemble forecasting business system construction. The present invention improves the ETKF initial value perturbation scheme and optimizes the boundary perturbation uncertainty scheme, improves the forecasting system discreteness and regional ensemble forecasting system performance, thereby meeting the business application requirements of China's regional ensemble forecasting system, and targetedly produces and provides weather forecasting services. The present invention provides 1h interval ensemble forecasting products of different levels and different forecasting time effectiveness for meteorological stations at all levels in China, provides guidance and reference for weather forecasting, and facilitates model developers to improve the performance of ensemble forecasting systems.

Description

Technical Field

[0001] The invention belongs to the technical field of numerical forecasting, and in particular relates to a GRAPES-REPS regional ensemble forecasting method and system. Background Art

[0002] At present, numerical forecasting is an effective tool for weather forecasting in my country, especially for forecasting severe weather. However, due to the uncertainty of single numerical forecasting, ensemble forecasting has become an important development direction in the field of numerical forecasting. However, due to the incomplete physical perturbation scheme and lateral boundary perturbation scheme, the inaccurate forecast of model radar echo diagnostic quantity, and other reasons, the existing ensemble forecasting system in my country has problems such as insufficient system divergence and insufficient probabilistic forecasting skills, which makes it unable to fully meet the business application needs of the system, and thus it is difficult to provide strong technical support for meteorological forecasting services for disaster prevention and mitigation.

[0003] GRAPES (Global and Regional Assimilation and Prediction Enhanced System) is a new generation of numerical models independently developed in my country, which includes global and regional versions. In recent years, although high-resolution numerical models have a certain forecasting ability for mesoscale weather systems, due to the forecast uncertainty caused by the model initial field error and the discretization model error, the availability of numerical models is relatively low when it comes to fixed points, timing, and quantity. Severe weather in small and medium scales is sudden and has a turning point, and its forecast uncertainty factors are difficult to grasp. How to construct an effective regional ensemble forecast based on regional models and through appropriate perturbation methods to describe the uncertainty of mesoscale disastrous weather is a current research hotspot and difficulty.

[0004] Since the dynamic mechanism of small and medium-scale and tropical cyclones is different from that of large-scale weather systems in mid- and high-latitudes, the corresponding ensemble forecasting theories and methods are still incomplete. Abroad, some scholars have applied methods such as BGM and SVs developed from medium-term ensemble forecasting to mesoscale ensemble forecasting experiments, and some scholars have developed initial value perturbation methods specifically for the characteristics of mesoscale forecasting. In my country, the Numerical Forecast Research Center of the Chinese Academy of Meteorological Sciences has established a multi-initial value multi-physics regional ensemble forecasting system using the growth mode propagation method. At the same time, it has also studied and developed the GRAPES-MESO regional ensemble forecasting technology and experimental system coupled with the global ensemble forecasting system data, developed the ensemble transform Kalman filter (ETKF, Ensemble Transform Kalman Filter) initial value perturbation technology and GRAPES model random perturbation technology, and established the GRAPES-regional ensemble forecasting experimental system.

[0005] The prior art mostly uses the ETKF initial value perturbation method and the multi-physical process combined model perturbation method to construct the regional ensemble forecasting test method, but the simulated observation data are used in the calculation, and the real observation data are not used; in addition, the forecasting system constructed by only combining the initial value perturbation with the physical process perturbation is insufficient in discreteness. The prior art also uses the random physical process tendency term perturbation scheme SPPT to express the model uncertainty by the random perturbation net tendency, thereby constructing an ensemble forecasting system, but the system divergence is not enough to fully meet the system's business application requirements. The prior art also uses the SLAF method (using the scaling factor σ to scale the difference between the forecast field and the analysis field at the past moment and superimpose it on the measurement boundary) to construct the measurement boundary perturbation scheme of the regional ensemble forecasting system. However, in the SLAF method, the forecast errors of the two moments with a large time distance are too large, resulting in an unreasonable discreteness structure, so that in the forecasting process, the unreasonable measurement boundary perturbation is quickly filtered by the model, resulting in a large deviation in the forecasting skill of the scheme. Therefore, a new regional ensemble forecasting method is urgently needed to make up for the defects of the prior art.

[0006] Through the above analysis, the problems and defects of the prior art are as follows:

[0007] (1) The existing ensemble forecasting system has problems such as insufficient system divergence and insufficient probabilistic forecasting skills due to the incomplete physical perturbation scheme and lateral boundary perturbation scheme, and the inaccurate forecast of model radar echo diagnostic quantity. As a result, it is unable to fully meet the system's business application needs, making it difficult to provide strong technical support for meteorological forecasting services for disaster prevention and mitigation.

[0008] (2) Due to the forecast uncertainty caused by the model initial field error and the discretization model error, the availability of high-resolution numerical models is low when it comes to fixed points, timing, and quantity. The forecast uncertainty factors of small and medium-scale severe weather are difficult to grasp. How to construct an effective regional ensemble forecast based on the regional model and through appropriate perturbation methods to describe the uncertainty of mesoscale disastrous weather is currently a research hotspot and difficulty.

[0009] (3) The existing regional ensemble forecasting test method uses the ETKF initial value perturbation method and the multi-physical process combined model perturbation method. The calculation uses simulated observation data instead of real observation data. Only the initial value perturbation is combined with the physical process perturbation, and the constructed forecasting system has insufficient discreteness.

[0010] (4) The existing perturbation scheme SPPT uses the random physical process tendency term to represent the model uncertainty through the random perturbation net tendency, thereby constructing an ensemble forecasting system. However, the system divergence is not enough to fully meet the system's business application requirements.

[0011] (5) The existing technology also uses the SLAF method (using the scaling factor σ to scale the difference between the forecast field and the analysis field at the past moment and superimpose it on the measurement boundary) to construct the measurement boundary perturbation scheme of the regional ensemble forecast system. The forecast error between two moments with a large time distance is too large, resulting in an unreasonable discrete structure. Therefore, in the forecast process, the unreasonable measurement boundary perturbation is quickly filtered by the model, resulting in a large deviation in the forecast skill of the scheme.

[0012] The difficulty of solving the above problems and defects is:

[0013] Research requires a lot of time and financial investment, and experiments require a lot of real data support. It is difficult for general enterprises or universities to meet the system construction and development needs.

[0014] The commercialization and promotion of the forecast system requires the support of authoritative meteorological forecast service agencies. Local meteorological stations lack financial and material resources and it is difficult to carry out nationwide promotion.

[0015] The significance of solving the above problems and defects is:

[0016] Ensure the system's divergence, improve probabilistic forecasting skills, and reduce forecast deviations so that it can meet the system's business application needs, and promote the business system to meteorological stations at all levels across the country, thereby providing strong technical support for meteorological forecast services for disaster prevention and reduction. Summary of the invention

[0017] In view of the problems existing in the prior art, the present invention provides a GRAPES-REPS regional ensemble forecasting method and system.

[0018] The present invention is achieved by: a GRAPES-REPS regional ensemble forecasting method comprising:

[0019] Establish the ETKF calculation scheme of the initial value disturbance of GRAPES-REPS with a horizontal resolution of 0.1 degrees, improve the calculation variables of the weight modulus of the ensemble transformation Kalman filter transformation matrix, optimize the calculation scheme of the observation error covariance matrix and the amplification factor, and realize the technology representing the uncertainty of the initial value is updated from ETKF+MSB to ETKF technology; the model uncertainty is updated from multi-physical process+SPPT to single physical process+SPPT, and the precipitation equivalence and precipitation probability forecasting skills of the ensemble forecast members are improved;

[0020] The uncertainty of lateral boundary perturbations was updated from T639 to GRAPES-GEPS global ensemble forecast, and a lateral boundary perturbation calculation scheme suitable for the characteristics of GRAPES-GEPS variables was designed; the ensemble forecast members added a typhoon vortex condition repositioning scheme, and the ensemble forecast members added cloud analysis assimilation technology; the GRAPES-REPS regional ensemble forecast real-time monitoring and scheduling process of ECFLOW was established to realize the GRAPES-REPS regional ensemble forecast operation scheduling; the unified post-processing system was optimized and improved, and the probabilistic forecast products were improved.

[0021] Furthermore, the probabilistic forecast products include wind, temperature, humidity, potential height, precipitation, multi-factors, multi-level ensemble average, dispersion, noodle charts, severe convection diagnosis products, and typhoon emergency service products.

[0022] Further, the GRAPES-REPS regional ensemble forecast method comprises the following steps:

[0023] Step 1: Determine the parameters and model of the GRAEPS-REPS ensemble forecast experiment with a horizontal resolution of 0.1 degrees;

[0024] Step 2: Construction of GRAPES-REPS lateral boundary perturbation;

[0025] Step three: Construction of the GRAPES-REPS regional ensemble forecast operational system.

[0026] Furthermore, in step 1, the 0.1 degree horizontal resolution GRAEPS-REPS ensemble forecast test parameters and model determination include:

[0027] The experimental model adopts GRAPES-MESO V4.0, the vertical coordinates adopt terrain-following height coordinates, the vertical level is 51, the number of ensemble forecast members is 14 ensemble perturbation members, plus the control forecast, there are 15 ensemble members in total, the physical processes of the ensemble forecast members are slightly different, the initial value of the control forecast adopts the analysis field generated by the GRAPES-3DVAR system, and the background field and lateral boundary conditions are provided by the T639 global ensemble forecast system of the Numerical Forecast Center; the ETKF initial value perturbation method adopts a 12h ensemble perturbation cycle calculation scheme, that is, the forecast perturbation composed of the 12h short-term forecast field of the ensemble forecast system is updated to the analysis perturbation at the current moment, the variables used in calculating the projection matrix E are the zonal wind U and the meridional wind V of the GRAPES-MESO model surface, and the vertical level adopted is 11 layers; the observation data adopts simulated observation data, that is, the T639 global model analysis field is interpolated to the observation site, and the simulated observation site selection refers to the distribution of real observation stations.

[0028] The number of observation stations is 1100, and the disturbance analysis field is obtained by adding the 14 disturbance fields calculated by the ETKF method to the initial value of the control forecast; the experimental forecast time is 72h, and the forecast range covers the entire Chinese region 15°~55°N, 70°~140°E; the experimental period is from 12:00 on June 1, 2015 to 12:00 on June 15, 2015, Coordinated Universal Time, of which June 1 to 10 is the adaptation and adjustment period for the calculation of the ETKF analysis disturbance α parameter.

[0029] Further, in step 2, the GRAPES-REPS lateral boundary perturbation construction includes:

[0030] The large-scale background driving data based on the GRAPES-GEPS global ensemble forecast system is designed to extract the perturbation values ​​of the lateral boundary conditions required by the GRAPES-REPS regional ensemble forecast system, and add them to the control forecast lateral boundary conditions to form the perturbation lateral boundary conditions required for the perturbation member forecast, including:

[0031] (1) Based on the GRAPES-MESO-3km global model background field data, the dynamical downscaling method is used to generate the initial values ​​and boundary conditions of the GRAPES-REPS control forecast model BDY_CTL REPS .

[0032] (2) The control forecast and the background field data of the 1st to 14th ensemble members are selected from the GRAPES-GEPS global ensemble forecast. The perturbation members with GRAPES-GEPS ensemble forecast business numbers 01 to 14 and the control forecast background field data with number 00 are used to generate the lateral boundary conditions of ensemble members 01 to 14 using the dynamical downscaling method. The lateral boundary generated by the control forecast with number 00 is subtracted to obtain the lateral boundary perturbation value BDY_Perti of the i-th ensemble forecast member:

[0033]

[0034] Where i is the ensemble forecast member number.

[0035] (3) Control the forecast side boundary condition BDY_CTL REPS Adding it to the lateral boundary disturbance calculated by the above formula, the GRAPES-REPS lateral boundary disturbance is generated, as shown in the following formula:

[0036]

[0037] Furthermore, in step 3, the GRAPES-REPS regional ensemble forecasting operational system is constructed, including:

[0038] On the Sugon PI-SUGON high-performance computer of the China Meteorological Administration, a regional ensemble forecast system was developed and applied. The scheduling of the system operation process was built based on the ECFLOW technology. Through concurrent scheduling and concurrent number control mechanisms, the GRAPES_REPS regional ensemble forecast system was established.

[0039] The GRAPES-REPS regional ensemble forecast system parameters include:

[0040] ① When the control forecast mode is horizontal resolution / vertical layer, GRAPES-MESO4.3 is 0.1 degree and 50 layers;

[0041] ② When the control forecast mode is the control forecast initial value, the initial value generation method of GRAPES-MESO4.3 is the same as that of GRAPES-3km;

[0042] ③ When the control forecast model is physical process, GRAPES-MESO4.3 is the same as GRAPES-3km+KF cumulus convection parameterization process;

[0043] ④ When the control forecast mode is the integration step, GRAPES-MESO4.3 is 60 seconds;

[0044] ⑤ When the control forecast mode is initial value uncertainty, GRAPES-MESO4.3 is the ensemble transformation Kalman filter method;

[0045] ⑥ When the control forecast model is model uncertainty, GRAPES-MESO4.3 is a random physical process tendency term perturbation method;

[0046] ⑦ When the control forecast mode is to measure boundary uncertainty, GRAPES-MESO4.3 is the GRAPES-GEPS global ensemble forecast;

[0047] ⑧When the control forecast mode is typhoon positioning uncertainty, GRAPES-MESO4.3 is a typhoon vortex condition relocation method;

[0048] ⑨When the control forecast model is the number of ensemble members, GRAPES-MESO4.3 is 15;

[0049] ⑩ When the control forecast mode is forecast time, GRAPES-MESO4.3 is 84 hours, 00 and 12 hours; 6 hours, 06 and 18 hours;

[0050] When the control forecast mode is the model product output frequency, GRAPES-MESO4.3 is 1 hour; When the control forecast mode is ensemble forecast postprocessing, GRAPES-MESO4.3 is GRIB2 data, conventional weather forecast products, and emergency service data.

[0051] Another object of the present invention is to provide a GRAPES-REPS regional ensemble forecast system using the GRAPES-REPS regional ensemble forecast method, the GRAPES-REPS regional ensemble forecast system comprising:

[0052] The background data preprocessing module is used for processing the GRAPES-REPS background field data; the background field data processing of GRAPES-REPS is divided into control forecast background field data processing and disturbance member background field data processing based on GRAPES-GEPS; wherein, the control forecast background field data processing uses the same global model data as GRAPES-3km, that is, the global 0.5°GRIB2 format data, to obtain preliminary control forecast initial fields and lateral boundary conditions, providing input data for the control forecast typhoon vortex repositioning and lateral boundary disturbance construction; the disturbance member background field data processing is based on the ensemble background field data provided by GRAPES-GEPS, that is, the global ensemble forecast 0.5°GRIB2 format data, to obtain preliminary 15 groups of disturbance member forecast initial fields and lateral boundary conditions, providing input data for the lateral boundary disturbance construction; wherein, the disturbance members include the GEPS control forecast.

[0053] The lateral boundary perturbation construction module is used to construct the lateral boundary perturbations of the control forecast and ensemble forecast members; the GRAPES-REPS control forecast lateral boundary conditions are generated based on the GRAPES-MESO-3km global model background field data, and the lateral boundary perturbations of the ensemble perturbation members are generated using the GRAPES-GEPS global ensemble forecast background field data, and superimposed on the control forecast lateral boundary conditions to obtain the lateral boundaries of the ensemble forecast perturbation members.

[0054] The initial value perturbation module is used to perform initial value perturbation on the ensemble Kalman filter transform ETKF. It uses the 6-hour forecast field of the previous GRAPES-REPS regional ensemble forecast, the simulated observation data of the current forecast time, the simulated observation site information, and the model initial value of the GRAPES-REPS control forecast of the current forecast time to construct and analyze the disturbance field, generate the initial disturbance field of 14 disturbance members, and provide input data for the repositioning of the typhoon vortex of the disturbance members.

[0055] The typhoon vortex relocation module is used to relocate the typhoon vortex conditions; the typhoon vortex relocation scheme of GRAPES-REPS corrects the forecast initial field by retrieving typhoon messages, calculating whether the distance between the typhoon vortex center of the model analysis field and the observed typhoon vortex center is greater than the 0.15 degree threshold; the typhoon vortex relocation scheme of the control forecast is executed after the control forecast background field data is processed, providing an initial field for the control forecast model integration, and providing a more reasonable control forecast model initial value to the ensemble Kalman filter initial value perturbation module; the typhoon vortex relocation scheme of the disturbance member is designed after the ensemble Kalman filter initial value perturbation, and on the basis of the reasonable disturbance initial field provided by ETKF, the forecast initial field is corrected to provide a more reasonable disturbance initial field for the disturbance member model integration.

[0056] The cloud analysis module is used to analyze the clouds of ensemble forecast members and generate cloud water data, providing cloud analysis data for the control forecast and the model integration of 14 disturbance members; it integrates the three-dimensional grid mosaic data of Doppler weather radars of various countries, the brightness temperature and model background field data of Fengyun geostationary meteorological satellites, and analyzes the cloud information according to the principles of cloud thermodynamics and empirical relationships of observational tests, and applies it to the early stage of model forecast to improve the model precipitation forecast effect.

[0057] The forecast integration control module is used to perform 84-hour integration using the 0.1° resolution GRAPES-MESO v4.3 model, the control forecast initial field and lateral boundary conditions obtained in the previous steps to obtain the control forecast integration results; the input data is the 0.1° control forecast initial value and lateral boundary conditions, and the output is the 1-hour forecast result within the 84-hour forecast period. The output results are stored in the file system of Sugon high-performance computers.

[0058] The perturbation member mode integration module is used to generate mode perturbation parameters and integrate 14 perturbation member modes; generate the random seeds, perturbation magnitudes, random perturbation cutoff wavenumbers, and loss of correlation time coefficients required for the SPPT mode perturbation schemes of the 14 perturbation members, and use the mode perturbation SPPT mode running program added to the GRAPES-MESO mode to complete the 84-hour integration of the 14 perturbation members and obtain the forecast results of the 14 perturbation members; the input data are the parameters including the initial fields of the 14 perturbation members, random seeds, perturbation magnitudes, random perturbation cutoff wavenumbers, and loss of correlation time coefficients; the output is the 1-hour forecast results within the 84-hour forecast period, and the output results are stored in the file system of the Sugon high-performance computer.

[0059] The integral data processing module is used for unified post-processing of ensemble forecast integral data. The elements and levels to be output are specified by the control file. The pattern variable output of the model surface is processed by the unified post-processing module to generate the output results in GRIB2 format, which includes isobaric surface variables, pattern surface variables and all diagnostic variables. The processing process is started immediately after the ensemble forecast result is generated at a certain time, providing timely GRIB2 data for product post-processing.

[0060] The ensemble forecast product post-processing module is used to generate business ensemble forecast products using the GRIB2 format ensemble forecast results obtained through unified post-processing; the conventional ensemble forecast product module will generate ensemble forecast box plot data, typhoon path ensemble forecast data and MICAPS format data; design a reasonable system startup time node to achieve automated business operation; design the ensemble forecast product transmission process to achieve the transmission of ensemble forecast products to the Central Meteorological Observatory; wherein the product content includes: ensemble mean, ensemble dispersion, ensemble maximum / minimum, ensemble quantile, mode and probability.

[0061] Another object of the present invention is to provide a computer device, the computer device comprising a memory and a processor, the memory storing a computer program, and when the computer program is executed by the processor, the processor performs the following steps:

[0062] Establish the ETKF calculation scheme of the initial value disturbance of GRAPES-REPS with a horizontal resolution of 0.1 degrees, improve the calculation variables of the weight modulus of the ensemble transformation Kalman filter transformation matrix, optimize the calculation scheme of the observation error covariance matrix and the amplification factor, and realize the technology representing the uncertainty of the initial value is updated from ETKF+MSB to ETKF technology; the model uncertainty is updated from multi-physical process+SPPT to single physical process+SPPT, and the precipitation equivalence and precipitation probability forecasting skills of the ensemble forecast members are improved;

[0063] The uncertainty of lateral boundary perturbations was updated from T639 to GRAPES-GEPS global ensemble forecast, and a lateral boundary perturbation calculation scheme suitable for the characteristics of GRAPES-GEPS variables was designed; the ensemble forecast members added a typhoon vortex condition repositioning scheme, and the ensemble forecast members added cloud analysis assimilation technology; the GRAPES-REPS regional ensemble forecast real-time monitoring and scheduling process of ECFLOW was established to realize the GRAPES-REPS regional ensemble forecast operation scheduling; the unified post-processing system was optimized and improved, and the probabilistic forecast products were improved.

[0064] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the following steps:

[0065] Establish the ETKF calculation scheme of the initial value disturbance of GRAPES-REPS with a horizontal resolution of 0.1 degrees, improve the calculation variables of the weight modulus of the ensemble transformation Kalman filter transformation matrix, optimize the calculation scheme of the observation error covariance matrix and the amplification factor, and realize the technology representing the uncertainty of the initial value is updated from ETKF+MSB to ETKF technology; the model uncertainty is updated from multi-physical process+SPPT to single physical process+SPPT, and the precipitation equivalence and precipitation probability forecasting skills of the ensemble forecast members are improved;

[0066] The uncertainty of lateral boundary perturbations was updated from T639 to GRAPES-GEPS global ensemble forecast, and a lateral boundary perturbation calculation scheme suitable for the characteristics of GRAPES-GEPS variables was designed; the ensemble forecast members added a typhoon vortex condition repositioning scheme, and the ensemble forecast members added cloud analysis assimilation technology; the GRAPES-REPS regional ensemble forecast real-time monitoring and scheduling process of ECFLOW was established to realize the GRAPES-REPS regional ensemble forecast operation scheduling; the unified post-processing system was optimized and improved, and the probabilistic forecast products were improved.

[0067] Another object of the present invention is to provide an information data processing terminal, which is used to implement the GRAPES-REPS regional ensemble forecast system.

[0068] Combined with all the above technical solutions, the advantages and positive effects of the present invention are as follows: the GRAPES-REPS regional ensemble forecast system provided by the present invention improves the ETKF initial value perturbation scheme and optimizes the measurement boundary perturbation uncertainty scheme, improves the forecast system discreteness and regional ensemble forecast system performance, thereby meeting the business application requirements of the regional ensemble forecast system, and can produce and provide weather forecast services in a targeted manner. The present invention can provide meteorological stations at all levels with 1h interval ensemble forecast products of different levels and different forecast time effectiveness, providing guidance and reference for weather forecasts; it can also provide ensemble forecast developers with products that reflect model performance and ensemble member characteristics, so that model developers can improve the performance of ensemble forecast systems; it can also provide China regional ensemble forecast products for my country's participation in WMO's FDP and RDP projects; it provides more targeted products for major meteorological support work: such as ensemble forecast products at large sports event venues and grid points, and fine and humanized ensemble forecast products for large social activities, etc., to provide technical guarantees for the normal development of events. The technical solution of the present invention also has the following advantages:

[0069] (1) Establish the ETKF calculation scheme for the initial value perturbation of GRAPES-REPS with a horizontal resolution of 0.1 degrees, improve the calculation variables of the weight modulus of the ensemble transform Kalman filter transformation matrix, optimize the calculation scheme of the observation error covariance matrix and the amplification factor, increase the calculation stability, and realize the update of the technology representing the initial value uncertainty from ETKF+MSB to ETKF technology;

[0070] (2) The model uncertainty is updated from multi-physics process + SPPT to single physical process + SPPT, which improves the precipitation equivalence and precipitation probability forecasting skills of ensemble forecast members.

[0071] (3) The uncertainty of the lateral boundary perturbation is updated from T639 to the GRAPES-GEPS global ensemble forecast, and a lateral boundary perturbation calculation scheme suitable for the characteristics of GRAPES-GEPS variables is designed.

[0072] (4) The ensemble forecast members add a typhoon vortex condition relocation scheme to improve the rationality of the initial uncertainty of the typhoon center positioning.

[0073] (5) Ensemble forecast members will add cloud analysis and assimilation technology to improve the quality of initial water vapor analysis and precipitation forecast.

[0074] (6) Establish a real-time monitoring and scheduling process for the GRAPES-REPS regional ensemble forecast of ECFLOW to realize the scheduling of GRAPES-REPS regional ensemble forecast operations.

[0075] (7) Optimize and improve the unified post-processing system and improve the probabilistic forecast products, including multi-factor and multi-level ensemble averages of wind, temperature, humidity, geopotential height, precipitation, etc., discreteness, noodle diagrams, severe convection diagnosis products, and typhoon emergency service products. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0077] Figure 1 It is a flow chart of the GRAPES-REPS regional ensemble forecast method provided by an embodiment of the present invention.

[0078] Figure 2 It is a structural block diagram of the GRAPES-REPS regional ensemble forecast system provided by an embodiment of the present invention;

[0079] In the figure: 1. Background data preprocessing module; 2. Side boundary disturbance construction module; 3. Typhoon vortex repositioning module; 4. Cloud analysis module; 5. Forecast integration control module; 6. Disturbance member model integration module; 7. Integration data processing module; 8. Ensemble forecast product post-processing module.

[0080] Figure 3 It is a flow chart of the GRAPES-REPS lateral boundary perturbation scheme provided in an embodiment of the present invention.

[0081] Figure 4 It is a flow chart of the operation of the GRAPES-REPS regional ensemble forecast system provided by an embodiment of the present invention.

[0082] Figure 5 It is a schematic diagram of the distribution of eigenvalues ​​of the forecast and analysis disturbance covariance matrix averaged from 00:00 on June 11, 2015 to 12:00 on June 15, 2015 (twice a day, 10 times in total) provided by an embodiment of the present invention.

[0083] Figure 5 (a) is a schematic diagram of a 10 km horizontal resolution test (referred to as 10 km test) provided in an embodiment of the present invention; dark color: predicted disturbance; light color: analyzed disturbance.

[0084] Figure 5 (b) is a schematic diagram of a 15 km horizontal resolution test (referred to as 15 km test) provided in an embodiment of the present invention; dark color: predicted disturbance; light color: analyzed disturbance.

[0085] Figure 6 It is a schematic diagram of the model initial value disturbance quantity of the 0.1 degree test and the 0.15 degree test from 12:00 on June 1 to 12:00 on June 15, 2015 (twice a day, a total of 10 times) provided by an embodiment of the present invention.

[0086] Figure 6 (a) is a schematic diagram of the daily evolution of the parameter value α provided by an embodiment of the present invention.

[0087] Figure 6 (b) is a schematic diagram of the daily evolution of the amplification factor Π provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0088] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0089] In view of the problems existing in the prior art, the present invention provides a GRAPES-REPS regional ensemble forecasting method and system, which will be described in detail below in conjunction with the accompanying drawings.

[0090] like Figure 1 As shown, the GRAPES-REPS regional ensemble forecast method provided by the embodiment of the present invention includes the following steps:

[0091] S101, 0.1 degree horizontal resolution GRAEPS-REPS ensemble forecast test parameters and model determination;

[0092] S102, GRAPES-REPS lateral boundary perturbation construction;

[0093] S103, Construction of GRAPES-REPS regional ensemble forecasting operational system.

[0094] like Figure 2 As shown, the GRAPES-REPS regional ensemble forecast system provided by the embodiment of the present invention includes:

[0095] Background data preprocessing module 1, used for GRAPES-REPS background field data processing;

[0096] The lateral boundary disturbance construction module 2 is used for constructing the lateral boundary disturbance of the control forecast and the ensemble forecast member;

[0097] A typhoon vortex relocation module 3 is used to relocate typhoon vortex conditions;

[0098] Cloud analysis module 4 is used to perform cloud analysis on ensemble forecast members, generate cloud water data, and provide cloud analysis data for control forecast and model integration of 14 disturbance members;

[0099] The forecast integration control module 5 is used to perform 84-hr integration using the 0.1° resolution GRAPES-MESO v4.3 model, the control forecast initial field obtained in the previous step, and the lateral boundary conditions;

[0100] The disturbance member mode integration module 6 is used for generating mode disturbance parameters and integrating 14 disturbance member modes;

[0101] An integral data processing module 7, used for unified post-processing of ensemble forecast integral data;

[0102] The ensemble forecast product post-processing module 8 is used to generate business ensemble forecast products using the GRIB2 format ensemble forecast results obtained through unified post-processing.

[0103] The technical solution of the present invention is further described below in conjunction with specific embodiments.

[0104] The technical solution of the present invention is:

[0105] (1) Establish the ETKF calculation scheme for the initial value perturbation of GRAPES-REPS with a horizontal resolution of 0.1 degrees, improve the calculation variables of the weight modulus of the ensemble transform Kalman filter transformation matrix, optimize the calculation scheme of the observation error covariance matrix and the amplification factor, increase the calculation stability, and realize the update of the technology representing the initial value uncertainty from ETKF+MSB to ETKF technology;

[0106] (2) The model uncertainty is updated from multi-physics process + SPPT to single physical process + SPPT, which improves the precipitation equivalence and precipitation probability forecasting skills of ensemble forecast members.

[0107] (3) The uncertainty of the lateral boundary perturbation is updated from T639 to the GRAPES-GEPS global ensemble forecast, and a lateral boundary perturbation calculation scheme suitable for the characteristics of GRAPES-GEPS variables is designed.

[0108] (4) The ensemble forecast members add a typhoon vortex condition relocation scheme to improve the rationality of the initial uncertainty of the typhoon center positioning.

[0109] (5) Ensemble forecast members will add cloud analysis and assimilation technology to improve the quality of initial water vapor analysis and precipitation forecast.

[0110] (6) Establish a real-time monitoring and scheduling process for the GRAPES-REPS regional ensemble forecast of ECFLOW to realize the scheduling of GRAPES-REPS regional ensemble forecast operations.

[0111] (7) Optimize and improve the unified post-processing system and improve the probabilistic forecast products, including multi-factor and multi-level ensemble averages of wind, temperature, humidity, geopotential height, precipitation, etc., discreteness, noodle diagrams, severe convection diagnosis products, and typhoon emergency service products.

[0112] The principle of the technical solution is described in detail below and illustrated with the accompanying drawings.

[0113] 1.0.1 degree horizontal resolution GRAEPS-REPS ensemble forecast test parameters and model

[0114] The experimental model uses GRAPES-MESO V4.0, the vertical coordinates use terrain-following height coordinates, the vertical level is 51, the number of ensemble forecast members is 14 ensemble perturbation members, plus the control forecast, there are 15 ensemble members in total, the physical processes of the ensemble forecast members are slightly different, the initial value of the control forecast uses the analysis field generated by the GRAPES-3DVAR system, and the background field and lateral boundary conditions are provided by the T639 global ensemble forecast system of the numerical forecast center. The ETKF initial value perturbation method uses a 12h ensemble perturbation cycle calculation scheme (that is, the forecast perturbation composed of the 12h short-term forecast field of the ensemble forecast system is updated to the analysis perturbation at the current moment), the variables used to calculate the projection matrix E are the GRAPES-MESO model surface zonal wind U and meridional wind V, and the vertical level used is 11. The observation data uses simulated observation data, that is, the T639 global model analysis field is interpolated to the observation site, and the simulated observation site selection refers to the distribution of real observation stations.

[0115] The number of observation stations is about 1100, and the disturbance analysis field is obtained by adding the 14 disturbance fields calculated by the ETKF method to the initial value of the control forecast. The experimental forecast time is 72h, and the forecast range covers the entire Chinese region (15°~55°N, 70°~140°E). The test period is from 12:00 on June 1, 2015 (Coordinated Universal Time, the same below) to 12:00 on June 15, of which June 1 to 10 is the adaptation adjustment period for the calculation of the α parameter of the ETKF analysis disturbance.

[0116] 2. GRAPES-REPS lateral boundary perturbation construction scheme

[0117] The large-scale background driving data based on the GRAPES-GEPS global ensemble forecast system is designed to extract the perturbation value of the lateral boundary conditions required by the GRAPES-REPS regional ensemble forecast system, and add it to the control forecast lateral boundary conditions to form the perturbed lateral boundary conditions required for the perturbation member forecast. The lateral boundary perturbation generation scheme and process are as follows: Figure 3 shown.

[0118] Firstly, based on the GRAPES-MESO-3km global model background data, the dynamical downscaling method is used to generate the initial values ​​and boundary conditions of the GRAPES-REPS control forecast model BDY_CTL. REPS Secondly, the control forecast and the background field data of the 1st to 14th ensemble members are selected from the GRAPES-GEPS global ensemble forecast. Specifically, the perturbation members with the GRAPES-GEPS ensemble forecast business numbers 01-14 and the control forecast background field data with the number 00 are used. The lateral boundary conditions of the 01-14 ensemble members are generated by the dynamic downscaling method, and the lateral boundary generated by the control forecast with the number 00 is subtracted to obtain the lateral boundary perturbation value BDY-Perti of the i-th ensemble forecast member:

[0119]

[0120] where i is the ensemble forecast member number. Finally, the boundary condition BDY_CTL on the control forecast side is REPS Adding it to the lateral boundary disturbance calculated by the above formula, the GRAPES-REPS lateral boundary disturbance is generated, as shown in the following formula:

[0121]

[0122] 3. Construction of GRAPES-REPS Regional Ensemble Forecasting Operational System

[0123] On the Sugon PI-SUGON high-performance computer of the China Meteorological Administration, a regional ensemble forecast system was developed and applied. The scheduling of the system operation process was built based on the ECFLOW technology. Through concurrent scheduling and concurrent number control mechanisms, the GRAPES_REPS regional ensemble forecast system was established. The system technical parameters are shown in Table 1.

[0124] Table 1 GRAPES-REPS regional ensemble forecast system parameters

[0125]

[0126]

[0127] The operation process of the GRAPES-REPS system is as follows: Figure 4 shown.

[0128] The system integrates multiple modules such as background data preprocessing, lateral boundary disturbance construction, typhoon vortex relocation, ensemble Kalman filter initial value disturbance, cloud analysis, ensemble forecast member integration mode and product post-processing technology. The main functions of each module are briefly described as follows:

[0129] (1) GRAPES-REPS background data processing. The background data processing of GRAPES-REPS is divided into control forecast background data processing and disturbance member background data processing based on GRAPES-GEPS. The former uses the same global model data as GRAPES-3km (global 0.5° GRIB2 format data) to obtain preliminary control forecast initial fields and lateral boundary conditions, providing input data for control forecast typhoon vortex repositioning and lateral boundary disturbance construction. The latter is based on the ensemble background data provided by GRAPES-GEPS (global ensemble forecast 0.5° GRIB2 format data) to obtain preliminary 15 groups of disturbance member (including GEPS control forecast) forecast initial fields and lateral boundary conditions, providing input data for lateral boundary disturbance construction.

[0130] (2) Construction of control forecast and ensemble forecast member lateral boundary perturbations. The GRAPES-REPS control forecast lateral boundary conditions are generated based on the GRAPES-MESO-3km global model background field data. The ensemble perturbation member lateral boundary perturbations are generated using the GRAPES-GEPS global ensemble forecast background field data and superimposed on the control forecast lateral boundary conditions to obtain the ensemble forecast perturbation member lateral boundaries.

[0131] (3) Ensemble Kalman Filter Transformation ETKF Initial Value Perturbation: Using the 6-hour forecast field of the previous time period of the GRAPES-REPS regional ensemble forecast, the simulated observation data of the current forecast time period, the simulated observation site information, and the model initial value of the GRAPES-REPS control forecast of the current forecast time period, the analysis disturbance field is constructed and the initial disturbance fields of 14 disturbance members are generated, providing input data for the repositioning of the typhoon vortex of the disturbance member.

[0132] (4) Typhoon vortex condition relocation scheme. The typhoon vortex relocation scheme of GRAPES-REPS corrects the forecast initial field by retrieving typhoon messages, calculating the distance between the typhoon vortex center of the model analysis field and the observed typhoon vortex center (whether it is greater than the 0.15 degree threshold). The typhoon vortex relocation scheme of the control forecast is executed after the control forecast background field data is processed, providing an initial field for the control forecast model integration, and providing a more reasonable control forecast model initial value to the ensemble Kalman filter initial value perturbation module. The typhoon vortex relocation scheme of the disturbance member is designed after the ensemble Kalman filter initial value perturbation, and on the basis of the reasonable disturbance initial field provided by ETKF, the forecast initial field is corrected to provide a more reasonable disturbance initial field for the disturbance member model integration.

[0133] (5) Cloud analysis of ensemble forecast members. Based on the three-dimensional grid mosaic data of Doppler weather radars in my country, the brightness temperature of Fengyun geostationary meteorological satellites, and the model background field data, the cloud information is analyzed according to the cloud thermodynamics principle and the empirical relationship of observational experiments, and applied to the early stage of model forecast to improve the model precipitation forecast effect. This module generates cloud water data and provides cloud analysis data for the control forecast and the model integration of 14 disturbance members.

[0134] (6) Control forecast integration. The 0.1° resolution GRAPES-MESO v4.3 model, the control forecast initial field and lateral boundary conditions obtained in the previous step are used for 84-hour integration to obtain the control forecast integration results. The input data are the 0.1° control forecast initial value and lateral boundary conditions, and the output is the 1-hour forecast result within the 84-hour forecast period. The output results are stored in the file system of the Sugon high-performance computer.

[0135] (7) Model perturbation parameter generation and 14 perturbation member model integration. Generate the random seeds, perturbation magnitude, random perturbation cutoff wavenumber, and de-correlation time coefficient required for the SPPT model perturbation scheme for the 14 perturbation members. Use the model perturbation SPPT model running program added to the GRAPES-MESO model to complete the 84-hour integration of the 14 perturbation members and obtain the forecast results of the 14 perturbation members. The input data are the initial fields of the 14 perturbation members, random seeds, perturbation magnitude, random perturbation cutoff wavenumber, de-correlation time coefficient and other parameters. The output is the 1-hour forecast result within the 84-hour forecast period. The output results are stored in the file system of the Sugon high-performance computer.

[0136] (8) Unified post-processing of ensemble forecast integral data. The elements and levels to be output are specified by the control file. The pattern variable output of the model surface is processed by the unified post-processing module to generate the output results in GRIB2 format, which includes isobaric surface variables, model surface variables and all diagnostic variables. This processing process is started immediately after the ensemble forecast result is generated at a certain time, providing timely GRIB2 data for product post-processing.

[0137] (9) Ensemble forecast product post-processing. Generate operational ensemble forecast products using the GRIB2 format ensemble forecast results obtained through unified post-processing. The conventional ensemble forecast product module will generate ensemble forecast box plot data, typhoon path ensemble forecast data, and MICAPS format data. Product content includes: ensemble mean, ensemble dispersion, ensemble maximum / minimum, ensemble quantile, mode, and probability, etc. Design a reasonable system startup time node to achieve automated business operation. Design the ensemble forecast product transmission process to achieve the transmission of ensemble forecast products to the Central Meteorological Observatory.

[0138] The technical solution of the present invention is further described below in conjunction with simulation experiments.

[0139] 1.0.1 degree horizontal resolution GRAEPS-REPS ensemble forecast test results

[0140] (1) Error variance distribution along orthogonal directions

[0141] Figure 5 The distribution of eigenvalues ​​of the covariance matrix of the forecast and analysis perturbations averaged from 00:00 on June 11 to 12:00 on June 15, 2015 (twice a day, 10 times in total): (a) 10 km horizontal resolution experiment (referred to as 10 km experiment); (b) 15 km horizontal resolution experiment (referred to as 15 km experiment). Dark red and dark blue: forecast perturbations; light red and light blue: analysis perturbations.

[0142] Figure 5The eigenvalue spectrum distribution of the covariance matrix of the average 12h forecast disturbance and the analysis disturbance from 00:00 on June 11, 2015 to 12:00 on June 15, 2015 (twice a day, a total of 10 times) is given (obtained by normalizing the first eigenvalue). The flatter the normalized eigenvalue spectrum distribution, the more eigenvectors are involved in describing the forecast uncertainty. On the contrary, the steeper the eigenvalue spectrum distribution, the less eigenvectors are involved in describing the forecast uncertainty. It can be seen from the figure that the eigenvalue spectrum distribution of the 12h forecast disturbance of the model of the 0.1 degree and 0.15 degree experiments is steeper, indicating that the eigenvalues ​​are quite different, while the eigenvalue distribution of the covariance matrix of the analysis disturbance is relatively flat, and the eigenvalues ​​are basically the same. Therefore, after the ensemble Kalman filter, the ensemble variance can be evenly distributed to all the eigenvectors of the ensemble subspace, ensuring that each perturbation member participates in describing the error variance. The analytical perturbations generated by the ETKF initial value perturbation scheme are mutually orthogonal and uncorrelated in the standardized observation space. The ensemble forecasts in the 0.15 and 0.1 degree experiments have this characteristic.

[0143] (2) Characteristics of the change of the parameter value of the collective perturbation α and the amplification factor Π

[0144] Figure 6 The daily evolution of the model initial value perturbation parameter α and the amplification factor Π for the 0.1 degree (red) test and the 0.15 degree (blue) test from 12:00 on June 1 to 12:00 on June 15, 2015 (twice a day, a total of 10 times).

[0145] Amplification factor Π n is determined by different cycle times α n The parameters are determined to ensure that the ensemble discreteness is comparable to the ensemble average root mean square error. In the early stage of ETKF ensemble perturbation calculation, the initial ensemble discreteness is inconsistent with the forecast root mean square error, resulting in α n The parameter value is very unstable; after about 3 to 4 days of perturbation cycle calculation, the initial ensemble discreteness and the magnitude of the forecast root mean square error gradually tend to be coordinated, α n The parameters gradually stabilized around 0.8σ1.2, and the distribution of the amplification factor also tended to be stable. Figure 6 In the 0.1 degree and 0.15 km experiments, α n Parameter value (see Figure 6 (a)) and the amplification factor Π n (See Figure 6 (b)) Evolution over time, α at the initial moment n Parameter value and amplification factor Π n are all 1. Figure 6 It can be seen that during the test period from 12:00 on June 1, 2015 to 12:00 on June 4, 2015 (i.e., the initial disturbance adaptation adjustment period), α nThe parameter values ​​are very unstable, and the parameter values ​​of the 10km test are larger than those of the 15km test. This is because during this process, the ensemble mean square error of the 10km test increases significantly compared with the 15km test, while the ensemble dispersion increases less. n The parameter values ​​are maintained between 0.8 and 1.2, the average value is close to 1, and the amplification factor Π n The amplification factor of the 15 km test is maintained between 30 and 40, and that of the 10 km test is maintained between 40 and 50. Therefore, the α parameter and amplification factor of the ETKF initial value schemes with different resolutions have good stability.

[0146] In summary, the GRAPES-REPS regional ensemble forecast system provided by the present invention improves the ETKF initial value perturbation scheme and optimizes the uncertainty scheme of the measured boundary perturbation, improves the forecast system discreteness and regional ensemble forecast system performance, thereby meeting the business application requirements of the regional ensemble forecast system, and can produce and provide weather forecast services in a targeted manner. The present invention can provide 1h interval ensemble forecast products of different levels and different forecast time effectiveness for meteorological stations at all levels, provide guidance and reference for weather forecast; can also provide ensemble forecast developers with products that reflect model performance and ensemble member characteristics, so that model developers can improve the performance of ensemble forecast systems; can also provide China's regional ensemble forecast products for my country's participation in WMO's FDP and RDP projects; provide more targeted products for major meteorological support work: such as ensemble forecast products at large sports event venues and grid points, and fine and humanized ensemble forecast products for large social activities, etc., to provide technical support for the normal development of events.

[0147] In the description of the present invention, unless otherwise specified, "plurality" means two or more than two; the orientations or positional relationships indicated by the terms "upper", "lower", "left", "right", "inner", "outer", "front end", "rear end", "head", "tail", etc. are based on the orientations or positional relationships shown in the drawings, and 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, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", "third", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0148] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When the use is implemented in whole or in part in the form of a computer program product, the computer program product includes one or more computer instructions. When the computer program instructions are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL) or wireless (e.g., infrared, wireless, microwave, etc.) mode) to another website site, computer, server or data center. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk SolidState Disk (SSD)), etc.

[0149] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A GRAPES-REPS regional ensemble forecast system, characterized in that: The GRAPES-REPS regional ensemble forecast system includes: Background data preprocessing module, used for GRAPES-REPS background field data processing; GRAPES-REPS background field data processing is divided into control forecast background field data processing and disturbance member background field data processing based on GRAPES-GEPS; wherein, the control forecast background field data processing uses the same global model data of GRAPES-3km, that is, global 0.5°GRIB2 format data, to obtain preliminary control forecast initial field and lateral boundary conditions, providing input data for control forecast typhoon vortex repositioning and lateral boundary disturbance construction; the disturbance member background field data processing, based on the ensemble background field data provided by GRAPES-GEPS, that is, global ensemble forecast 0.5°GRIB2 format data, obtains preliminary 15 groups of disturbance member forecast initial fields and lateral boundary conditions, providing input data for lateral boundary disturbance construction; wherein, the disturbance member includes GEPS control forecast; The lateral boundary perturbation construction module is used to construct the lateral boundary perturbations of the control forecast and ensemble forecast members. The GRAPES-REPS control forecast lateral boundary conditions are generated based on the GRAPES-MESO-3km global model background field data. The lateral boundary perturbation of the ensemble perturbation members is generated using the GRAPES-GEPS global ensemble forecast background field data and superimposed on the control forecast lateral boundary conditions to obtain the lateral boundary of the ensemble forecast perturbation members. The initial value perturbation module is used to perform the initial value perturbation of the ensemble Kalman filter transform ETKF. It uses the 6-hour forecast field of the previous time period of the GRAPES-REPS regional ensemble forecast, the simulated observation data of the current forecast time period, the simulated observation site information and the model initial value of the GRAPES-REPS control forecast of the current forecast time period to construct and analyze the disturbance field, generate the initial disturbance field of 14 disturbance members, and provide input data for the repositioning of the typhoon vortex of the disturbance member; The typhoon vortex relocation module is used to relocate the typhoon vortex conditions. The typhoon vortex relocation scheme of GRAPES-REPS corrects the forecast initial field by retrieving typhoon messages and calculating whether the distance between the typhoon vortex center of the model analysis field and the observed typhoon vortex center is greater than the 0.15 degree threshold. The typhoon vortex relocation scheme of the control forecast is executed after the control forecast background field data is processed to provide an initial field for the control forecast model integration and provide a more reasonable control forecast model initial value to the ensemble Kalman filter initial value perturbation module. The typhoon vortex relocation scheme of the disturbance member is designed after the ensemble Kalman filter initial value perturbation, and on the basis of the reasonable disturbance initial field provided by ETKF, the forecast initial field is corrected to provide a more reasonable disturbance initial field for the disturbance member model integration. The cloud analysis module is used to analyze the clouds of the ensemble forecast members and generate cloud water data, providing cloud analysis data for the control forecast and the model integration of the 14 disturbance members; it integrates the three-dimensional grid mosaic data of Doppler weather radars from various countries, the brightness temperature of wind and cloud geostationary meteorological satellites, and the model background field data information, analyzes the cloud information according to the cloud thermodynamics principle and the empirical relationship of observation tests, and applies it to the early stage of model forecast to improve the model precipitation forecast effect; The forecast integration control module is used to use the 0.1° resolution GRAPES-MESO v4.3 model, the control forecast initial field and the lateral boundary conditions obtained in the previous steps to perform 84-hour integration to obtain the control forecast integration results; the input data is the 0.1° control forecast initial value and the lateral boundary conditions, and the output is the 1-hour forecast result within the 84-hour forecast time limit, and the output results are stored in the file system of the Sugon high-performance computer; The perturbation member mode integration module is used to generate mode perturbation parameters and integrate 14 perturbation member modes; generate the random seeds, perturbation magnitude, random perturbation cutoff wavenumber, and loss of correlation time coefficient required for the SPPT mode perturbation scheme of 14 perturbation members, and use the mode perturbation SPPT mode running program added to the GRAPES-MESO mode to complete the 84-hour integration of 14 perturbation members and obtain the forecast results of 14 perturbation members; the input data are the parameters including the initial field of 14 perturbation members, random seeds, perturbation magnitude, random perturbation cutoff wavenumber, and loss of correlation time coefficient; the output is the 1-hour forecast result within the 84-hour forecast period, and the output results are stored in the file system of Sugon high-performance computers; The integral data processing module is used for unified post-processing of ensemble forecast integral data. The elements and levels to be output are specified by the control file. The pattern variable output of the pattern surface is processed by the unified post-processing module to generate the output results in GRIB2 format, which includes isobaric surface variables, pattern surface variables and all diagnostic variables. The processing process is started immediately after the ensemble forecast result is generated at a certain time, providing timely GRIB2 data for product post-processing. The ensemble forecast product post-processing module is used to generate business ensemble forecast products using the GRIB2 format ensemble forecast results obtained through unified post-processing; the conventional ensemble forecast product module will generate ensemble forecast box plot data, typhoon path ensemble forecast data and MICAPS format data; design a reasonable system startup time node to achieve automated business operation; design the ensemble forecast product transmission process to achieve the transmission of ensemble forecast products to the Central Meteorological Observatory; wherein the product content includes: ensemble mean, ensemble dispersion, ensemble maximum / minimum, ensemble quantile, mode and probability.

2. The GRAPES-REPS regional ensemble prediction system according to claim 1, characterized in that: Probabilistic forecast products include wind, temperature, humidity, geopotential height, precipitation, multi-factor ensemble average, dispersion, noodle charts, severe convection diagnosis products, and typhoon emergency service products.

3. The GRAPES-REPS regional ensemble prediction system according to claim 1, characterized in that: The GRAPES-REPS regional ensemble forecast method includes the following steps: Step 1: Determine the parameters and model of the GRAEPS-REPS ensemble forecast experiment with a horizontal resolution of 0.1 degrees; Step 2: Construction of GRAPES-REPS lateral boundary perturbation; Step three: Construction of the GRAPES-REPS regional ensemble forecast operational system.

4. The GRAPES-REPS regional ensemble prediction system according to claim 3, characterized in that: In step 1, the 0.1 degree horizontal resolution GRAEPS-REPS ensemble forecast test parameters and model determination include: The experimental model adopts GRAPES-MESO V4.0, the vertical coordinates adopt terrain-following height coordinates, the vertical level is 51, the number of ensemble forecast members is 14 ensemble perturbation members, plus the control forecast, there are 15 ensemble members in total, the physical processes of the ensemble forecast members are slightly different, the initial value of the control forecast adopts the analysis field generated by the GRAPES-3DVAR system, and the background field and lateral boundary conditions are provided by the T639 global ensemble forecast system of the Numerical Forecast Center; the ETKF initial value perturbation method adopts a 12h ensemble perturbation cycle calculation scheme, that is, the forecast perturbation composed of the 12h short-term forecast field of the ensemble forecast system is updated to the analysis perturbation at the current moment, the variables used to calculate the projection matrix E are the GRAPES-MESO model surface zonal wind U and meridional wind V, and the vertical level used is 11; the observation data adopts simulated observation data, that is, the T639 global model analysis field is interpolated to the observation site, and the simulated observation site selection refers to the distribution of real observation stations; The number of observation stations is 1100, and the disturbance analysis field is obtained by adding the 14 disturbance fields calculated by the ETKF method to the initial value of the control forecast; the experimental forecast time is 72h, and the forecast range covers the entire Chinese region 15°~55°N, 70°~140°E; the experimental period is from 12:00 on June 1, 2015 to 12:00 on June 15, 2015, Coordinated Universal Time, of which June 1 to 10 is the adaptation and adjustment period for the calculation of the ETKF analysis disturbance α parameter.

5. The GRAPES-REPS regional ensemble prediction system according to claim 3, characterized in that: In step 2, the GRAPES-REPS lateral boundary perturbation is constructed, including: The large-scale background driving data based on the GRAPES-GEPS global ensemble forecast system is designed to extract the perturbation values ​​of the lateral boundary conditions required by the GRAPES-REPS regional ensemble forecast system, and add them to the control forecast lateral boundary conditions to form the perturbation lateral boundary conditions required for the perturbation member forecast, including: (1) Based on the GRAPES-MESO-3km global model background field data, the dynamical downscaling method is used to generate the initial values ​​and boundary conditions of the GRAPES-REPS control forecast model BDY_CTL REPS ; (2) The control forecast and the background field data of the 1st to 14th ensemble members are selected from the GRAPES-GEPS global ensemble forecast. The perturbation members with GRAPES-GEPS ensemble forecast business numbers 01 to 14 and the control forecast background field data with number 00 are used to generate the lateral boundary conditions of ensemble members 01 to 14 using the dynamical downscaling method. The lateral boundary generated by the control forecast with number 00 is subtracted to obtain the lateral boundary perturbation value BDY_Perti of the i-th ensemble forecast member: Where i is the ensemble forecast member number; (3) Control the forecast side boundary condition BDY_CTL REPS Adding it to the lateral boundary disturbance calculated by the above formula, the GRAPES-REPS lateral boundary disturbance is generated, as shown in the following formula:

6. The GRAPES-REPS regional ensemble prediction system according to claim 3, characterized in that: In step 3, the GRAPES-REPS regional ensemble forecasting operational system is constructed, including: On the China Meteorological Administration's Sugon PI-SUGON high-performance computer, we developed and applied a regional ensemble forecast system, built a scheduling system for the system's operating process based on the ECFLOW technology, and established the GRAPES_REPS regional ensemble forecast system through concurrent scheduling and concurrent number control mechanisms; Among them, the GRAPES-REPS regional ensemble forecast system parameters include: ① When the control forecast mode is horizontal resolution / vertical layer, GRAPES-MESO4.3 is 0.1 degree and 50 layers; ② When the control forecast mode is the control forecast initial value, the initial value generation method of GRAPES-MESO4.3 is the same as that of GRAPES-3km; ③ When the control forecast model is physical process, GRAPES-MESO4.3 is the same as GRAPES-3km+KF cumulus convection parameterization process; ④ When the control forecast mode is the integration step, GRAPES-MESO4.3 is 60 seconds; ⑤ When the control forecast mode is initial value uncertainty, GRAPES-MESO4.3 is the ensemble transformation Kalman filter method; ⑥ When the control forecast model is model uncertainty, GRAPES-MESO4.3 is a random physical process tendency term perturbation method; ⑦ When the control forecast mode is to measure boundary uncertainty, GRAPES-MESO4.3 is the GRAPES-GEPS global ensemble forecast; ⑧When the control forecast mode is typhoon positioning uncertainty, GRAPES-MESO4.3 is a typhoon vortex condition relocation method; ⑨When the control forecast model is the number of ensemble members, GRAPES-MESO4.3 is 15; ⑩ When the control forecast mode is forecast time, GRAPES-MESO4.3 is 84 hours, 00 and 12 hours; 6 hours, 06 and 18 hours; When the control forecast mode is the model product output frequency, GRAPES-MESO4.3 is 1 hour; When the control forecast mode is ensemble forecast postprocessing, GRAPES-MESO4.3 is GRIB2 data, conventional weather forecast products, and emergency service data.

7. An information data processing terminal, characterized in that: The information data processing terminal is used to implement the GRAPES-REPS regional ensemble forecast system as described in any one of claims 1 to 6.

Images