CN-121995539-A - Cloud analysis combined random disturbance method for regional convection scale set prediction

CN121995539ACN 121995539 ACN121995539 ACN 121995539ACN-121995539-A

Abstract

The invention discloses a cloud analysis combined random disturbance method for regional convection scale set prediction in the technical field of meteorological data processing, which comprises the steps of acquiring cloud analysis data such as black body brightness temperature, cloud total quantity, radar three-dimensional networking reflectivity and the like of a monitored region and other data required by analysis and assimilation based on a regional WRF mode. Aiming at uncertainty sources of a cloud analysis system, the method carries out random disturbance on multisource observation data input in the formation process of a cloud initial field of each set forecasting member in set forecasting, carries out three-dimensional disturbance on key dynamics and thermodynamic sensitive parameters of cloud analysis aiming at uncertainty of a micro-physical parameterization scheme of the cloud analysis system, forms combined random disturbance of an initial value and the two processes of the parameterization scheme, further reflects the initial value of the cloud analysis system and the uncertainty of the micro-physical parameterization of the cloud, and the interaction of the initial value and the cloud micro-physical parameterization of the cloud analysis system, can change the convection triggering and precipitation evolution process, and improves accuracy of a mesoscale service set forecasting system on cloud and precipitation forecasting.

Inventors

ZHOU YUSHU
XU ZHIZHEN
ZHUANG ZHAORONG
WU ZHUOHENG
ZHANG HONGCHI
LAI ZIYANG
LI YINGLIN
ZHOU FEIFEI
WU KAIYUAN
LI HEYUAN
CHEN FEIFAN
ZHU LIJUAN
LIU MANQI
YU SHIYUAN
GUO NANNAN
CHEN JING
DENG GUO
YANG YILIN
YUE JIAN
LI HONGQI
XU ZHIFANG
Chen Fajing
WANG JINGZHUO

Assignees

中国科学院大气物理研究所
中国气象局地球系统数值预报中心
国家气象中心(中央气象台、中国气象局气象导航中心)

Dates

Publication Date: 20260508
Application Date: 20251204

Claims (10)

1. The cloud analysis random combined disturbance method for regional convection scale set prediction is characterized by being applied to a convection scale mode with horizontal resolution of 2 km-4 km, wherein the convection scale mode comprises a WRF mode and a CMA-MESO mode, and the method comprises the following steps: Based on a WRF mode, analyzing and assimilating observation data, hour average black body bright temperature data, cloud total data, radar three-dimensional networking reflectivity data sets and/or analyzing and assimilating required data of a monitoring area are obtained, and aggregate forecast data are obtained, wherein the analyzing and assimilating required data comprise aircraft report data, exploring data and GPS water vapor inversion data; Extracting background field information from the set forecast data by adopting a background extraction technology to obtain isobaric surface coarse grid file data, and converting the isobaric surface coarse grid file data into an initial field and side boundary conditions on mode grid points of the mode to obtain initial field data and side boundary condition field data; Based on a normal distribution random number generator and a cloud analysis system, inputting total product data of domestic wind cloud satellite clouds, average black body bright temperature data in hours and a three-dimensional radar networking reflectivity data set into the cloud analysis system; based on the statistical characteristics of the observation errors, generating a group of observation data with the same possibility in the statistical sense, and forming two expression methods of uncertainty of the observation data of the cloud analysis system: Collecting the minimum radar reflectivity of a cloud analysis system, and carrying out a standard deviation observation data random disturbance module based on a default mean value to generate a minimum radar reflectivity random threshold value which accords with the default mean value and standard deviation; Estimating error characteristics of the blackbody bright temperature, sparsely extracting the hour average blackbody bright temperature data by adopting a random sampling method, and obtaining uncertain state data of the hour average blackbody bright temperature data; Based on the error characteristics of the total cloud product, sparsely extracting the data of the total satellite cloud product data of the domestic wind cloud by adopting a random sampling method to obtain the uncertain data of the total satellite cloud; based on different altitudes, three-dimensional key sensitive parameter data of a cloud analysis system are mutually converted, on the basis of default values of original parameters of the system, the following key parameters showing thermodynamic and dynamic uncertainties in the cloud analysis process are subjected to random disturbance of different thresholds through a normal distribution random disturbance generator, and a sensitive parameter random disturbance module is constructed, wherein the threshold value of converting relative humidity into cloud quantity, the lowest relative humidity threshold value of converting cloud quantity into relative humidity, the highest relative humidity threshold value of converting cloud quantity into relative humidity, the lowest cloud quantity threshold value of converting cloud quantity into relative humidity and the cloud quantity threshold value of converting cloud quantity into relative humidity reach a saturated state; The conversion result is obtained through calculation of key sensitive parameters of thermodynamic and dynamic uncertainty in the cloud analysis process, and based on domestic wind cloud satellite hour average blackbody bright temperature data, cloud total data and radar three-dimensional networking reflectivity data set corresponding to the conversion result, joint random disturbance is carried out, and a method of controlling forecast integration is adopted to obtain a set forecast product corresponding to the set forecast data.
2. The method for cloud analysis and random disturbance combined prediction of regional flow scale set according to claim 1, wherein the method for cloud analysis and random disturbance combined prediction of regional flow scale set is characterized by obtaining analysis and assimilation observation data, hour average black body bright temperature data, cloud total data, radar three-dimensional networking reflectivity data set and/or analysis and assimilation required data of a monitoring region based on a WRF mode, and comprises the following steps: acquiring blackbody bright temperature original data from a wind cloud satellite data receiving station, sequentially decoding, decompressing and time averaging the blackbody bright temperature original data to obtain hour average blackbody bright temperature data; acquiring cloud total original data from a wind-cloud satellite data receiving station, and sequentially decoding and decompressing the cloud total original data to obtain cloud total data; Acquiring original weather radar jigsaw data, inquiring the legend reflectivity RGB and the annotation RGB of the original weather radar jigsaw data to obtain a reflectivity and annotation RGB mapping table, performing reflectivity conversion on the annotation RGB of all grid points of the original weather radar jigsaw data to obtain converted weather radar jigsaw data, and complementing missing grid points of the converted weather radar jigsaw data by adopting an interpolation method to obtain a radar three-dimensional networking reflectivity data set; And/or collecting the original data of the data required by analysis and assimilation, performing format conversion on the original data of the data required by analysis and assimilation to obtain the data required by analysis and assimilation, and improving the quality of the assimilation data through a quality control scheme.
3. The method for cloud analysis combined random disturbance of regional streaming scale set prediction according to claim 2, wherein the step of extracting background field information from the set prediction data by using a background extraction technology to obtain isobaric coarse grid file data, and converting the isobaric coarse grid file data into an initial field and a side boundary condition on a mode lattice point of the mode to obtain initial field data and side boundary condition field data comprises the steps of: Sequentially adopting a horizontal interpolation method, a vertical interpolation method and a variable transformation method to process the isophase coarse grid file data to obtain initial field and side edge conditions; And respectively endowing a corresponding mode initial value set and a side boundary condition to analysis and assimilation observation data, exploration data, ship data, satellite cloud wind-guiding and GPS water vapor inversion data and the like, hour average blackbody brightness temperature data, cloud total data, radar three-dimensional networking reflectivity data set and/or analysis and assimilation required data to obtain mode initial value set data and side boundary condition set data, wherein the mode initial value set data is initial field data, and the side boundary condition set data is side boundary condition field data.
4. The method for expressing uncertainty of input data of a cloud analysis system for regional scale set forecasting according to claim 3, wherein the method for expressing uncertainty of input data of a cloud analysis system for regional scale set forecasting is characterized in that the method for expressing uncertainty of input data of a regional scale set forecasting based on a normal distribution random number generator and a cloud analysis system inputs total product data of domestic wind cloud satellite clouds, hour average black body bright temperature data and radar three-dimensional networking reflectivity data sets into the cloud analysis system, and comprises the following steps: When the cloud analysis system receives radar reflectivity data, a normal distribution random number generator is utilized to randomly disturb the standard deviation of the radar reflectivity around a default mean value, a lowest radar reflectivity random result conforming to the default mean value and the standard deviation is generated, and uncertainty random distribution data of the radar reflectivity is reflected; When the cloud analysis system receives the hour average blackbody bright temperature data, according to error characteristics of the blackbody bright temperature, adopting a random sampling method to sparsely extract the hour average blackbody bright temperature data, and obtaining uncertain state data of the hour average blackbody bright temperature data; When the cloud analysis system receives the total cloud data of the domestic wind cloud satellites, the total cloud data is sparsely extracted by adopting a random sampling method according to the error characteristics of the total cloud product, and the extracted samples reflect the space-time correlation of the product errors as much as possible, so that the uncertainty data of the total cloud is obtained.
5. The cloud analysis stochastic combined disturbance method for regional convection scale set forecasting according to claim 4, further comprising the step of randomly disturbing key dynamics and thermodynamic sensitive parameters of the micro-physical parameterization of the cloud analysis system to form a key sensitive parameter random disturbance module reflecting uncertainty of the cloud analysis micro-physical parameterization scheme.
6. The method for cloud analysis stochastic united disturbance of regional convection scale set forecast of claim 5, wherein the method for cloud analysis stochastic united disturbance of regional convection scale set forecast is characterized by performing inter-conversion on relative humidity and cloud total data of a cloud analysis system based on different altitudes to obtain conversion results, and comprises the following steps: Based on the threshold value of cloud total data corresponding to different altitudes, the relative humidity with the standard deviation of 0.05 is converted into random cloud total data.
7. The method for cloud analysis stochastic united disturbance of regional convection scale set forecast of claim 5, wherein the method for cloud analysis stochastic united disturbance of regional convection scale set forecast is characterized by performing inter-conversion on relative humidity and cloud total data of a cloud analysis system based on different altitudes to obtain conversion results, and comprises the following steps: Setting a threshold value for converting relative humidity with height into cloud quantity based on a cloud analysis system, wherein when the height is less than 600m, the average value is 0.925, the standard deviation is 0.025, when the height is 600m-1500m, the average value is 0.9, the standard deviation is 0.025, when the height is 1500m-2500m, the average value is 0.85, the standard deviation is 0.025, and when the height is higher than 2500m, the average value is 0.8, and the standard deviation is 0.075; setting a random disturbance value with a default mean value of 0.5 and a standard deviation of 0.1, and generating a lowest random threshold value of relative humidity corresponding to cloud total data; Setting a random disturbance value with a default mean value of 1 and a standard deviation of 0.05, and generating a highest random threshold value of relative humidity corresponding to cloud total data; setting a random disturbance value with a default mean value of 0.2 and a standard deviation of 0.1, and generating a minimum random threshold value of cloud total data corresponding to the relative humidity; And setting a random disturbance value with a default mean value of 0.7 and a standard deviation of 0.1, and generating corresponding cloud total data under the condition that the relative humidity reaches a saturated state.
8. The method for cloud analysis random combined disturbance of regional convection dimension set forecast of claim 7, wherein the method for carrying out combined random disturbance based on domestic wind cloud satellite hour average blackbody bright temperature data, cloud total data and radar three-dimensional networking reflectivity data set corresponding to conversion results and adopting a method for controlling forecast integration to obtain set forecast products corresponding to set forecast data comprises the following steps: And adopting a random generator to randomly perturb analysis and assimilation observation data outside the cloud analysis system, and generating an area set forecast perturbation initial value based on the three-dimensional variation system and the set assimilation method.
9. The method for cloud analysis stochastic combined perturbation of regional convection dimension set forecast of claim 8, wherein the cloud analysis input data comprises analysis assimilation observation data, radar three-dimensional networking reflectivity and TBB data sets, and key sensitive parameters reflecting uncertainty of cloud analysis microphysics parameterization process, the key sensitive parameters comprise a threshold value of converting relative humidity into cloud quantity, a lowest relative humidity threshold value of converting cloud quantity into relative humidity, a highest relative humidity threshold value of converting cloud quantity into relative humidity, a lowest cloud quantity threshold value of converting cloud quantity into relative humidity and a cloud quantity threshold value of converting cloud quantity into relative humidity reaching saturation state, the method for controlling forecast integration is adopted to obtain a set forecast product corresponding to the set forecast data, and the method further comprises: Generating an initial field file and a side condition file of the set forecast data based on a stream scale set forecast initial value disturbance method and a side disturbance method; Integrating the initial field file and the side edge condition file by adopting a WRF mode with 3km resolution and a dynamic downscaling method to obtain a control forecast integration result; and respectively adopting a disturbance mode in the WRF mode to obtain disturbance member integration for the set forecast data to obtain a set forecast product.
10. A system, comprising: The data acquisition module (1) is configured to acquire analysis and assimilation observation data, hour average black body bright temperature data, cloud total data, radar three-dimensional networking reflectivity data sets and/or analysis and assimilation and cloud analysis required data of a monitoring area based on a WRF mode to obtain set forecast data; The information extraction module (2) is configured to extract background field information from the set forecast data by adopting a background extraction technology to obtain isobaric plane coarse grid file data, and convert the isobaric plane coarse grid file data into an initial field and side boundary conditions on a mode grid point of the mode to obtain initial field data and side boundary condition field data; The cloud analysis system observation data module (3) inputs the total amount data of domestic wind cloud satellite clouds, the average black body bright temperature data in hours, the radar three-dimensional networking reflectivity data set and the initial field lattice point field data after analysis and assimilation into the cloud analysis system; The cloud analysis observation data random disturbance module (4) collects the minimum radar reflectivity of the cloud analysis system, adopts a normal distribution random number generator, carries out standard deviation random disturbance based on a default mean value to generate a minimum radar reflectivity random threshold value which accords with the default mean value and the standard deviation, adopts a random sampling method to respectively sparsify and extract domestic wind cloud satellite hour average blackbody bright temperature data and cloud total data to obtain the uncertainty data of the hour average blackbody bright temperature data and the cloud total, thereby controlling radar and satellite observation data entering the cloud analysis system, reflecting the uncertainty of the observation data, and providing a plurality of cloud analysis input observation data sets with different observation initial values for the collection forecasting system; The cloud analysis system key sensitive parameter three-dimensional random disturbance module (5) is used for carrying out random disturbance of different thresholds on key parameters of a cloud analysis system by a normal distribution random disturbance generator according to key dynamic and thermal sensitive parameters in the following cloud analysis system on the basis of default values of original parameters of the system, and constructing a sensitive parameter random disturbance module, wherein the threshold value (Rh 0) for converting relative humidity into cloud quantity) +the lowest relative humidity threshold value (rh_thr1) for converting the cloud quantity into relative humidity) +the highest relative humidity threshold value (rh_thr2) for converting the cloud quantity into relative humidity) +the lowest cloud quantity threshold value (cvr rh_thr1) for converting the cloud quantity into the relative humidity into the cloud quantity threshold value (cvr rh_thr1) for converting the cloud quantity into the relative humidity into a saturated state; The cloud analysis system combines random disturbance (6), combines the modules, and simultaneously performs the random disturbance on the observation data of the cloud analysis system and the key sensitive parameters of the dynamics and the thermodynamics of the microphysics parameterization scheme, namely, the uncertainty of the input data of the cloud analysis system and the uncertainty of the dynamic thermodynamic process of the cloud analysis system are reflected; The product obtaining module (7) is configured to mutually convert the relative humidity and the cloud total data of the cloud analysis system based on different altitudes to obtain conversion results, perform joint random disturbance based on analysis and assimilation observation data corresponding to the conversion results and a radar three-dimensional networking reflectivity data set, and obtain an aggregate forecast product corresponding to the aggregate forecast data by adopting a control forecast method and a method of integrating different aggregate members.

Description

Cloud analysis combined random disturbance method for regional convection scale set prediction Technical Field The invention relates to the technical field of meteorological data processing, in particular to a cloud analysis random combined disturbance method for regional convection scale set prediction. Background Strong convection weather is one of the main disastrous weather. The weather with strong convection is weather with sudden, rapid movement, severe weather and extremely strong destructive power, and mainly comprises thunderstorm, strong wind, hail, tornado, local heavy rainfall, and line. The strong convection weather has the characteristics of small space scale, short life cycle, strong burst, rapid development and evolution and the like, and is a difficulty in weather forecast business. However, because of its great destructiveness, its effective forecast is an urgent need for disaster prevention and reduction, major social activities, and refined weather services. Since the introduction of meteorology, the idea of aggregate forecasting breaks through the idea of raising forecasting as a deterministic initial value problem in the traditional numerical weather forecasting field, and realizes that the forecasting skill of the atmospheric mode depends not only on initial conditions, forecasting modes and the degree of coincidence with the actual condition of the atmosphere, but also on the stability of the atmospheric ring itself. The integrated forecasting can provide a forecasting skill higher than that of single deterministic forecasting and also can provide a possibility forecasting of sudden and disastrous weather process by absorbing the existing achievements of the traditional deterministic weather forecasting and taking the uncertainty of the initial value of a numerical forecasting mode, the forecasting mode error and the chaotic characteristic of the atmospheric motion development into consideration, and becomes an important component of a weather forecasting system in actual business forecasting. The resolution of the existing main stream area set forecasting system is usually 2-10 km, mainly focusing on short-term weather forecasting for 1-3 days, and can forecast some local details of the development process of the weather system and judge predictability of the strong weather process. The initial value disturbance method is one of the core contents of a regional set prediction system, at present, the method for constructing the regional set prediction initial disturbance field is mainly three, the first method is a dynamic downscaling method (Frogner et al, 2006; li et al, 2008;Bowler and Mylne,2009), the global set prediction initial value disturbance is downscaling to the regional set initial disturbance, and the method is simple and easy to implement, but the generated initial disturbance is difficult to accurately describe the medium-small scale information of the atmospheric motion. The second method is a method of generating a disturbance initial value by using a pattern self-circulation, such as a growth mode propagation method (the breeding method, abbreviated as BGM), a set transform kalman filter method (Ensemble Transform KALMAN FILTER, abbreviated as ETKF), a set transform method (Ensemble Transform, abbreviated as ET) (Bishop ET al.,2001,2009;Bowler and Mylne,2009;Du ET al, 2003;Majumdar ET al, 2002; ma Xulin, etc., 2006,2008,2021; zhang Han, etc., 2014a,2014b,2017; liu Kan, etc., 2023). Although the disturbance method of the mode self-circulation can describe uncertainty of the development process of the small-scale weather system, false gravitational waves are easy to excite to influence the forecasting effect because of the matching problem of the disturbance scale of the side boundary (Caron, 2013). The third method is a Multi-Scale Blending initial perturbation method (MSB for short) combining the advantages of the dynamic downscaling method and the mode self-circulation generating initial perturbation method. The hybrid disturbance field generated by the spectral decomposition and filtering method has advantages in reflecting large-scale and medium-scale motion development information (Zhang et al 2015b; zhang Han et al, 2016). However, it is difficult to objectively and accurately determine information of large scale and medium scale and information of small scale, and excessive or insufficient fusion of disturbance information may be caused when multi-scale disturbance information is fused, so that a new scheme of a mixed scale initial disturbance method based on a data assimilation idea is further developed by Wastl et al (2021) and Ma Xulin and the like (2018). The numerical forecast is used as a technical core of weather forecast, and is required to have quick response capability to rainfall and the like, and is required to have the capability of proximity forecast. Generally, the proximity forecast refers to a weather forecast of 0-2 hours, and the proximity f