A method, device, equipment and medium for determining a sparsification parameter of meteorological observation

Abstract

The application discloses a meteorological observation sparsification parameter determination method, device, equipment and medium, and relates to the technical field of meteorological numerical prediction, to solve the problem that the existing meteorological numerical prediction selects a set sparsification parameter value for sparsification processing, which causes a model to be unable to accurately describe the atmospheric state displayed by high-resolution observation data. The method comprises the following steps: determining an upper limit value of a sparsification parameter based on a background field water vapor correlation scale; performing horizontal grid division on satellite observation data to obtain a plurality of sparsification grid units; calculating a local standard deviation corresponding to each sparsification grid unit, and determining a local standard deviation probability density distribution corresponding to each sparsification parameter value; and determining a first target sparsification parameter value based on the local standard deviation probability density distribution. The meteorological observation sparsification parameter determination method provided by the application is used to determine a sparsification parameter value that matches the resolution of a regional numerical prediction assimilation model and sparsified observation data, and improves the meteorological prediction accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of meteorological numerical prediction, and in particular to a meteorological observation sparsification parameter determination method, device, equipment and medium. BACKGROUND

[0002] Meteorological satellite data needs to be sparsified before entering the regional numerical prediction assimilation system. However, in the prior art, when sparsifying the satellite data before assimilation, the sparsification is usually performed by using the parameters preset in the assimilation system. However, with the increase of satellite observation information, when the satellite data is assimilated, the resolution of the observation data after sparsification may not match the resolution of the regional numerical prediction assimilation mode if the preset sparsification parameter value is too low, and the observation resolution after sparsification is still high, and the correlation and redundant information processing between observations cannot be reduced. Conversely, if the preset sparsification parameter value is too high, the resolution of the observation after sparsification is low, and a lot of key observation information is lost. SUMMARY

[0003] The present application aims to provide a meteorological observation sparsification parameter determination method, device, equipment and medium, which can ensure that the sparsification parameter value used by the regional numerical prediction assimilation system matches the resolution of the observation data, improve the calculation efficiency, retain sufficient observation information, obtain the best analysis field, and make the meteorological prediction more accurate.

[0004] To achieve the above-mentioned purpose, the present application provides the following technical scheme:

[0005] In a first aspect, the present application provides a meteorological observation sparsification parameter determination method, comprising:

[0006] acquiring satellite observation data of a to-be-observed region and system information of a regional numerical prediction assimilation system; the system information includes a background field water vapor correlation scale and a plurality of preset sparsification parameter values;

[0007] determining an upper limit value of the sparsification parameter based on the background field water vapor correlation scale;

[0008] performing horizontal grid division on the satellite observation data by using the sparsification parameter values, to obtain a plurality of sparsification grid units;

[0009] calculating a local standard deviation corresponding to each sparsification grid unit, and determining a local standard deviation probability density distribution corresponding to each sparsification parameter value according to the local standard deviation;

[0010] determining a first target sparsification parameter value based on the upper limit value and the local standard deviation probability density distribution corresponding to each sparsification parameter value.

[0011] Optionally, the calculating the local standard deviation corresponding to each of the sparsification grid cells and determining the local standard deviation probability density distribution corresponding to each sparsification parameter value according to the local standard deviation comprises:

[0012] According to the formula:

[0013]

[0014] The local standard deviation corresponding to each sparsification grid cell is calculated.

[0015] wherein, is the local standard deviation, N is the total number of observations in the sparsification grid cell, x i is the value of the i-th observation, and μ is the average value of the N observations in the sparsification grid cell.

[0016] The square root of the local standard deviation is calculated to obtain the local standard deviation corresponding to the sparsification grid cell, and the local standard deviation probability density distribution is generated according to the local standard deviation.

[0017] Optionally, the determining the first target sparsification parameter value based on the upper limit value and the local standard deviation probability density distribution corresponding to each sparsification parameter value further comprises:

[0018] The satellite observation data is sparsified using the first target sparsification parameter value to obtain sparsified data within the model range.

[0019] According to the formula:

[0020]

[0021] The global variance corresponding to the model range is calculated to obtain the global variance.

[0022] wherein, is the global variance, N1 is the total number of observations of the sparsified data within the model range, x i1 is the value of the i1-th observation, and μ1 is the average value of the N1 observations within the model range.

[0023] The first target sparsification parameter value corresponding to the global variance less than the preset threshold value is determined as the second target sparsification parameter value.

[0024] Optionally, the determining the first target sparsification parameter value based on the upper limit value and the local standard deviation probability density distribution corresponding to each sparsification parameter value further comprises:

[0025] The observation and background field difference value probability density distribution corresponding to each first target sparsification parameter value is determined based on the regional numerical prediction assimilation system.

[0026] The first target sparsification parameter value corresponding to the observation and background field difference value probability density distribution closest to the normal distribution is determined as the third target sparsification parameter value.

[0027] Optionally, the regional numerical prediction assimilation system determines the observation and background field difference value probability density distribution corresponding to each first target sparsification parameter value, comprising:

[0028] The formula is:

[0029]

[0030] The observation and background field difference value data set corresponding to each first target sparsification parameter value is determined.

[0031] Wherein, J(x) is the weighted sum of the background field error and the observation error, x is the analysis variable, that is, the analysis field generated after assimilation, x b is the background field; H is the observation operator, y o is the observation field; B is the background error covariance matrix, O is the observation covariance matrix, and T represents the matrix transpose;

[0032] The observation and background field difference value probability density distribution corresponding to the first target sparsification parameter value is determined according to the observation and background field difference value data set.

[0033] Optionally, the upper limit value is half of the water vapor related scale of the background field.

[0034] Optionally, the horizontal scale relationship of the model variable and the observation variable corresponding to the observation operator is 1:1.

[0035] Compared with the prior art, the present application provides a meteorological observation sparsification parameter determination method, which determines the upper limit value of the sparsification parameter based on the background field water vapor correlation scale; adopts the sparsification parameter value to perform horizontal grid division on satellite observation data to obtain a plurality of sparsification grid units; calculates the local standard deviation corresponding to each sparsification grid unit, and determines the local standard deviation probability density distribution corresponding to each sparsification parameter value according to the local standard deviation; and determines the first target sparsification parameter value based on the upper limit value and the local standard deviation probability density distribution corresponding to each sparsification parameter value. The present application can quickly and accurately determine the sparsification parameter value according to the requirement, and the sparsification performed by using the sparsification parameter value determined by the method can not only reduce redundant information processing and improve the calculation efficiency, but also match the background field to retain key observation information. Moreover, the upper limit value of the sparsification parameter is determined according to the background field water vapor correlation scale of the regional numerical prediction assimilation system, which reduces the calculation amount of solving the target function value in the assimilation process, and also makes the observation resolution after sparsification match the assimilation system resolution, so that the regional numerical prediction assimilation system can better absorb the observation information, reduce the representative error, the background field absorbs the observation information, and generates a more actual initial field, thereby improving the prediction effect of the regional numerical prediction assimilation system.

[0036] In a second aspect, the present application provides a meteorological observation sparsification parameter determination device, comprising:

[0037] A data acquisition module is configured to acquire satellite observation data of a to-be-observed region and system information of a regional numerical prediction assimilation system; the system information comprises a background field water vapor correlation scale and a plurality of preset sparsification parameter values;

[0038] A sparsification parameter upper limit value determination module is configured to determine an upper limit value of the sparsification parameter based on the background field water vapor correlation scale;

[0039] A horizontal sparsification grid division module is configured to perform horizontal grid division on the satellite observation data by using the sparsification parameter value to obtain a plurality of sparsification grid units;

[0040] A local standard deviation probability density distribution result determination module is configured to calculate the local standard deviation corresponding to each sparsification grid unit, and determine the local standard deviation probability density distribution corresponding to each sparsification parameter value according to the local standard deviation;

[0041] A first target sparsification parameter value determination module is configured to determine a first target sparsification parameter value based on the upper limit value and the local standard deviation probability density distribution corresponding to each sparsification parameter value.

[0042] In a third aspect, the present application provides a meteorological observation sparsification parameter determination device, comprising:

[0043] A communication unit / communication interface is configured to acquire satellite observation data of a region to be observed and system information of a regional numerical prediction assimilation system; the system information comprises a background field water vapor related scale and a plurality of preset sparse parameter values;

[0044] A processing unit / processor is configured to determine an upper limit value of the sparse parameter based on the background field water vapor related scale;

[0045] The satellite observation data is horizontally grid-divided by using the sparse parameter value, to obtain a plurality of sparse grid cells;

[0046] A local standard deviation corresponding to each sparse grid cell is calculated, and a local standard deviation probability density distribution corresponding to each sparse parameter value is determined according to the local standard deviation;

[0047] A first target sparse parameter value is determined based on the upper limit value and the local standard deviation probability density distribution corresponding to each sparse parameter value.

[0048] In a fourth aspect, the present application provides a computer readable storage medium, wherein instructions are stored in the computer readable storage medium, and when the instructions are executed, the weather observation sparse parameter determination method is implemented.

[0049] Compared with the prior art, the beneficial effects of the second aspect device scheme, the third aspect equipment scheme and the fourth aspect computer readable storage medium scheme of the present application are the same as those of the weather observation sparse parameter determination method described above, and will not be repeated here. BRIEF DESCRIPTION OF DRAWINGS

[0050] The accompanying drawings, which are included to provide a further understanding of the present application, form a part of the present application and illustrate embodiments of the present application and together with the description serve to explain the present application. In the drawings:

[0051] Figure 1 A flowchart of a weather observation sparse parameter determination method provided by the present application;

[0052] Figure 2 A schematic diagram of dividing sparse grid cells by using a sparse parameter provided by the present application;

[0053] Figure 3 L1 level original data with a model resolution of 4km provided by the present application;

[0054] Figure 4 CSR original data with a model resolution of 12km provided by the present application;

[0055] Figure 5The observation distribution map provided by the present application has a sparse parameter value of 20km;

[0056] Figure 6 The observation distribution map provided by the present application has a sparse parameter value of 30km;

[0057] Figure 7 The observation distribution map provided by the present application has a sparse parameter value of 40km;

[0058] Figure 8 The observation distribution map provided by the present application has a sparse parameter value of 50km;

[0059] Figure 9 The observation distribution map provided by the present application has a sparse parameter value of 60km;

[0060] Figure 10 The observation distribution map provided by the present application has a sparse parameter value of 70km;

[0061] Figure 11 The observation distribution map provided by the present application has a sparse parameter value of 80km;

[0062] Figure 12 The observation distribution map provided by the present application has a sparse parameter value of 200km;

[0063] Figure 13 The local difference calculation schematic diagram provided by the present application;

[0064] Figure 14 The probability density distribution schematic diagram of the local difference provided by the present application;

[0065] Figure 15 The local standard deviation probability density distribution diagram corresponding to different sparse parameter values provided by the present application;

[0066] Figure 16 The schematic diagram of the change of the mode range global variance with the resolution provided by the present application;

[0067] Figure 17 The observation and background field difference value probability density distribution schematic diagram of the CSR observation obtained by different degrees of sparse entering the assimilation system provided by the present application;

[0068] Figure 18 The structural schematic diagram of a meteorological observation sparse parameter determination device provided by the present application;

[0069] Figure 19 The structural schematic diagram of a meteorological observation sparse parameter determination device provided by the present application. DETAILED DESCRIPTION

[0070] In order to clearly describe the technical solutions of the embodiments of the present application, in the embodiments of the present application, the terms of "first", "second", etc. are used to distinguish the same or similar items with basically the same function and role. For example, the first threshold and the second threshold are only used to distinguish different thresholds, and the order is not limited. Those skilled in the art can understand that the terms of "first", "second", etc. do not limit the quantity and execution order, and the terms of "first", "second", etc. also do not necessarily mean different.

[0071] It should be noted that in the present application, the words "exemplary" or "for example" are used to represent an example, illustration or description. Any embodiment or design scheme described as "exemplary" or "for example" in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the words "exemplary" or "for example" are used to present the relevant concept in a specific way.

[0072] In the present application, "at least one" means one or more, and "multiple" means two or more. The "and / or" describes the association between the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent the following cases: A exists alone, A and B exist together, and B exists alone, where A and B can be singular or plural. The character " / " generally represents an "or" relationship between the associated objects before and after it. "At least one of the following" or similar expressions means any combination of these items, including any combination of single or multiple items. For example, at least one of a, b or c can represent: a, b, c, the combination of a and b, the combination of a and c, the combination of b and c, or the combination of a, b and c, where a, b and c can be single or multiple.

[0073] Before introducing the embodiments of the present application, the related terms involved in the embodiments of the present application are first explained as follows:

[0074] The background error covariance horizontal correlation characteristic scale is an index for measuring the strength and spatial distribution range of the correlation of the background error in the horizontal direction. In numerical weather prediction, the background error represents the difference between the initial estimate of the numerical prediction model for the atmospheric state and the actual atmospheric state. The distribution and correlation of this difference in the horizontal direction have an important influence on the performance of the assimilation system and the accuracy of the numerical prediction.

[0075] The resolution of the regional numerical prediction system is not only affected by the resolution of the regional numerical prediction model, but also by other links and elements in the system. In addition to the numerical model, the data assimilation process, the accuracy and distribution of observation data, and the output format of the prediction product will all affect the resolution of the system. The resolution of the system reflects the comprehensive precision that the entire regional assimilation prediction system can achieve from data collection, processing, model running to the final presentation of the prediction product. Among them, the horizontal resolution of the numerical prediction model and the data assimilation model is consistent, and the resolution of the analysis field output by the assimilation model is also consistent with the horizontal resolution of the numerical prediction model, which can be directly used as the initial field (initial value) for prediction. Therefore, as long as the horizontal resolution of the observation at the assimilation analysis time meets the matching requirement of the horizontal resolution of the model (including the prediction model and the assimilation model).

[0076] The initial value is an important factor affecting the accuracy of the meteorological numerical prediction model. The initial value of the regional numerical prediction model is usually obtained by fusing various real-time observation information based on the analysis field generated by the global numerical prediction model. With the rapid development of remote sensing technology, meteorological satellite data has become an important source of observation data for numerical weather prediction, especially in areas where conventional observations (such as ground stations and sounding stations) are sparse. Before entering the numerical prediction assimilation system, satellite data must be pre-processed, which includes data thinning.

[0077] Currently, when thinning the satellite data before assimilation, the parameters set in advance by the assimilation system are usually used. With the increase of satellite observation information, the pre-set parameters cannot meet the actual application requirements. For example, for high-resolution observation information, using a pre-set larger thinning parameter for thinning processing will result in a too low observation resolution, which may lose important information and lead to inaccurate final prediction results. If the thinning parameter value is set too small, the observation resolution after thinning is still high, and it is not possible to effectively reduce the processing of redundant information and improve the calculation rate.

[0078] To solve the above problems, the present application provides a meteorological observation thinning parameter determination method, device, equipment and medium, which combines a regional numerical prediction assimilation system and satellite observation data, and provides a strategy for determining the thinning parameter, so as to quickly and accurately give a thinning parameter value that meets the actual requirements in the assimilation application. Next, the present application will be described in conjunction with the accompanying drawings.

[0079] Participate Figure 1 The meteorological observation thinning parameter determination method provided by the present application includes the following steps:

[0080] Step 101: obtaining satellite observation data of a to-be-observed region and system information of a regional numerical prediction assimilation system;

[0081] The system information comprises a background field water vapor related scale and a plurality of preset sparsification parameter values; the background field water vapor related scale is a horizontal correlation characteristic scale of a water vapor background error covariance corresponding to a water vapor field of the background field.

[0082] The sparsification parameter represents a resolution after sparsification of the observation data, i.e., an observation resolution; the preset sparsification parameter values can be randomly generated by the system or set according to needs. Exemplarily, the plurality of preset sparsification parameter values can be 10 km, 20 km, 30 km, 40 km, 50 km, 60 km, 70 km, 80 km and 200 km.

[0083] The satellite observation data is satellite observation data, and the satellite observation data can be clear sky radiation products (CSR products), long-wave radiation products or atmospheric sounding products of FY4B-AGRI.

[0084] Step 102: determining an upper limit value of the sparsification parameter based on the background field water vapor related scale;

[0085] The upper limit value is greater than or equal to 1 / 3 of the background field water vapor related scale and less than or equal to 1 / 2 of the background field water vapor related scale. The background error of the water vapor field has significant spatial correlation within 100 km. In order to effectively assimilate the observation data and reduce the error, the horizontal resolution of the water vapor observation should match the background error correlation scale. Setting 1 / 3 to 1 / 2 of the horizontal correlation characteristic scale of the background error covariance as the upper limit value of the sparsification parameter can ensure that the observation data can provide sufficient details to capture the water vapor changes in the background field. Exemplarily, the horizontal correlation characteristic scale of the water vapor background error covariance is 100 km, and the upper limit value of the resolution of the water vapor observation can be set to about 30 to 50 km, i.e., the upper limit value of the finally determined sparsification parameter can be set to about 30 to 50 km.

[0086] As an optional way, the plurality of preset sparsification parameter values obtained are less than or equal to the upper limit value.

[0087] Step 103: performing horizontal grid division on the satellite observation data by using the sparsification parameter value, to obtain a plurality of sparsification grid units;

[0088] The sparsification grid unit is a near-rhombus, and the sparsification parameter value is the side length of the sparsification grid unit; because the division is on the earth's surface, the lengths of the two adjacent latitudes above and below are inconsistent; but when the division is on the meridian, the lengths are consistent, so the shape of the sparsification grid unit is close to a rhombus. It should be understood that the sparsification grid unit comprises the observation data in the sparsification grid unit.

[0089] Specifically, the satellite observation data is horizontally grid divided by each sparse parameter value, and each sparse parameter value corresponds to a plurality of sparse grid units of sparse sampling. The specific steps of grid dividing the satellite observation data by the sparse parameter are as shown in Figure 2 As shown, the sparse parameter value is first equally divided along the meridian of the to-be-observed region, and then the equally divided points are equally divided along the latitude by the sparse parameter value. Thus, the region surrounded by the four adjacent four equally divided points is a sparse grid unit, marked as a sampling box, and only one observation data in the box is selected as the observation after sparse sampling and entering the assimilation system. The side length of the box is the sparse parameter value. If the sparse parameter value is large, the sparse degree is large, and the number of sparse observation data is small; otherwise, the sparse parameter value is small, the sparse degree is small, and the number of sparse data is large.

[0090] Step 104: calculating the local standard deviation corresponding to each sparse grid unit, and determining the local standard deviation probability density distribution corresponding to each sparse parameter value according to the local standard deviation;

[0091] Specifically, according to formula (1):

[0092]

[0093] The local standard deviation corresponding to each sparse grid unit is calculated;

[0094] wherein, is the local standard deviation, N is the total number of observations in the sparse grid unit, x i is the value of the i th observation point, and μ is the average value of the N observations in the sparse grid unit.

[0095] The local standard deviation is square-rooted to obtain the local standard deviation corresponding to the sparse grid unit, and the local standard deviation probability density distribution is generated according to the local standard deviation, so as to obtain the local standard deviation probability density distribution corresponding to each sparse grid unit.

[0096] Step 105: determining the first target sparse parameter value based on the upper limit value and the local standard deviation probability density distribution corresponding to each sparse parameter value;

[0097] The first target sparse parameter value is less than or equal to the upper limit value.

[0098] Specifically, the sparse parameter value less than the upper limit value and corresponding to the local standard deviation probability density distribution satisfying the preset condition is determined as the first target sparse parameter value.

[0099] The preset condition can be that the skewness of the local standard deviation probability density distribution is less than or equal to a first preset value and the peak value of the local standard deviation probability density distribution is less than or equal to a second preset value. The skewness is the distance between the peak value and the average value of the local standard deviation probability density. The first preset value and the second preset value can be set according to observation requirements. For example, the first preset value can be 0.4, 0.36, etc., and the second preset value can be 1.

[0100] It can be known from the above method that the application can quickly and accurately determine the sparsification parameter value according to requirements, and the sparsification parameter is determined according to the background field water vapor correlation scale of the regional numerical prediction assimilation system, that is, the upper limit value of the sparsification parameter is set to be greater than or equal to 1 / 3 of the horizontal correlation characteristic scale of the vapor background error covariance and less than or equal to 1 / 2 of the horizontal correlation characteristic scale of the vapor background error covariance, thereby reducing the calculation amount of solving the objective function value in the assimilation process, and also making the observation resolution after sparsification match the assimilation system resolution, so that the regional numerical prediction assimilation model can better absorb observation information, reduce representative error, the background field absorbs theoretical observation information, generates a more actual initial field, and further improves the prediction effect of the regional numerical prediction assimilation system.

[0101] As an optional mode, the first target sparsification parameter value is determined based on the upper limit value and the local standard deviation probability density distribution corresponding to each sparsification parameter value, and then the method can further include:

[0102] Step 106: Sparsification processing is performed on the satellite observation data by using the first target sparsification parameter value, to obtain sparsification data in the model range.

[0103] Specifically, in the observation range, the first target sparsification parameter value is used to divide into a plurality of small grids-sparsification grid units box. According to the requirements of the assimilation model, an observation in each box is selected as the observation information of the grid after sparsification, and the observation data is output according to the format requirements of the assimilation system, so as to obtain the sparsification data in the model range.

[0104] Step 107: According to formula (2):

[0105]

[0106] The variance corresponding to the model range is calculated to obtain a global variance. In order to distinguish from the local variance, the variance corresponding to the model range is referred to as the global variance.

[0107] wherein, is the global variance, N1 is the total number of observations of the sparsification data in the model range, x i1 is the value of the i1th observation, and μ1 is the average value of the N1 observations in the model range.

[0108] Step 108: Determine the first target sparsification parameter value corresponding to the global variance that is less than the preset threshold as the second target sparsification parameter value.

[0109] Specifically, the preset threshold can be set according to the observation requirements. For example, the preset threshold can be 1.09, 1.1 or 1.11, etc.

[0110] Accuracy can be improved by further filtering the sparsification parameter values ​​of the first objective by calculating the global variance.

[0111] As an alternative approach, the first target sparsification parameter value is determined based on the upper limit value and the local standard deviation probability density distribution corresponding to each sparsification parameter value. Further sparsification parameter values ​​can then be filtered through the following steps:

[0112] Step 109: Use the regional numerical prediction assimilation system to determine the probability density distribution of the difference between the observation and background fields corresponding to each first target sparsification parameter value;

[0113] Specifically, the satellite observation data of the area to be observed is first sparsified using the first target sparsification parameter to obtain sparsified data within the model range. Then, the sparsified data within the model range is input into the regional numerical prediction assimilation system. During operation, the regional numerical prediction assimilation system outputs the observation-background field difference dataset corresponding to the first target sparsification parameter. Finally, the probability density distribution of the observation-background field difference corresponding to the first target sparsification parameter value is determined based on the observation-background field difference dataset.

[0114] For example, the regional numerical weather prediction assimilation system GRAPES can be used to determine the difference dataset between the observation and background fields. The horizontal resolution of this system is 3 km. The assimilation method of this system is three-dimensional variational assimilation, which is essentially a problem of minimizing the objective function as shown in Equation (3). Equation (3) is as follows:

[0115]

[0116] Where J(x) is the weighted sum of background field error and observation error, and x is the analysis variable, i.e., the analysis field generated after assimilation. b The background field is typically generated by a global model; H is the observation operator, which is the radiative transfer model RTTOV that links model variables and observations; y o For observations, H(x) represents the observed radiation brightness temperature, while for satellite observations, H(x) represents the background field, i.e., the simulated brightness temperature of the background field. oH(x) is the difference between observation and background field, since the observation is brightness temperature, the difference between observation and background field is the difference between observation brightness temperature and background simulated brightness temperature, which will be output during system running; B is the background error covariance matrix, O is the observation covariance matrix, and T represents matrix transpose.

[0117] The observation and background field difference data set corresponding to each first target sparsification parameter value can be determined by using formula (3).

[0118] As can be seen from formula (3), the core of the variational assimilation of the regional numerical prediction assimilation system is to compare the data of the observation point entering the assimilation system with the background field. The second term (H(x)-y o ) T O -1 (H(x)-y o ) of formula (3) is calculated. Since the observation point usually does not fall on the grid point, that is, the observation value and the background value are not in the same spatial position, it is necessary to first perform spatial interpolation to interpolate the grid space of the background value to the observation space, and then convert the mode background value to the brightness temperature value of the observation through physical transformation, that is, convert the background value to the simulated brightness temperature value through the radiative transfer model RTTOV, and then calculate the difference with the brightness temperature data of the observation point. If the number of observations is large, the calculation amount of the second term of formula (3) will increase, and the calculation efficiency will decrease. At the same time, the observation is dense, and when the horizontal interpolation is performed using the mode grid points, the observations on different observation points close to each other will use the same mode grid points, and the correlation will be increased and the representative error will be introduced through different distance weight interpolation. The representative error is the difference between the observation and the background field. The GRAPES prediction system adopts equidistant longitude-latitude grid points and Arakawa-C format in the horizontal direction. The main physical schemes of the mode include SAS (Simplified Arakawa Schubert) cumulus parameterization scheme, WSM6 microphysical process, SLAB land surface process, RRTM longwave radiation, ECMWF shortwave radiation, etc.

[0119] It should be noted that the above-mentioned selected regional numerical prediction assimilation system GRAPES is only for illustration and is not limited, and in actual application, other types of regional numerical prediction assimilation systems can be used to determine the observation and background field difference data set according to the needs.

[0120] The horizontal correlation characteristic scale of the background error covariance matrix is one of the key factors to determine the effect of the variational assimilation, so the horizontal correlation characteristic scale of the background error covariance matrix is determined by the regional numerical prediction assimilation system, and the observation and the background field difference data set is determined by the regional numerical prediction assimilation system, and then the observation and the background field difference probability density distribution is determined according to the data set, and finally the final sparse parameter value is determined according to the observation and the background field difference probability density distribution, and in this step, the resolution of the background error covariance matrix is fully considered, so that the final sparse parameter can make the sparseness reduce the redundant information processing and improve the calculation efficiency, and can match the background field and retain the key observation information.

[0121] As an optional way, determining the observation and background field difference probability density distribution corresponding to the first target sparse parameter value according to the observation and background field difference data set can include: the observation and background field difference data set can be grouped according to appropriate intervals, and the probability of each group is calculated, and these probabilities form the probability density distribution. The appropriate interval can be selected according to the need, or the probability of each observation and background field difference in the data set is calculated, and the probability density distribution is formed according to the probability.

[0122] Step 110: determining the first target sparse parameter value corresponding to the observation and background field difference probability density distribution closest to the normal distribution as the third target sparse parameter value. Specifically, the first target sparse parameter value corresponding to the observation error probability density distribution closest to zero skewness is selected as the third target sparse parameter value.

[0123] In the regional numerical prediction assimilation system, the y o Before the H(x) probability density distribution, it is necessary to confirm the horizontal scale relationship between the observation operator and the model variable, that is, the scale relationship between the CSR observation and the model water vapor (specific humidity), and since the CSR observation of the water vapor channel is to be assimilated, the model water vapor is selected as the model variable.

[0124] The basic principle of the radiative transfer model RTTOV is to solve formula (4):

[0125]

[0126] Where, L Clr (v, θ) is the satellite detected clear sky area radiation energy, B(v, T) is the Planck function of channel i when the environmental temperature is T, τ s (v, θ) is the transmittance from the ground to the top of the atmosphere in the satellite observation direction, ε s (v, θ) is the surface emissivity, T is the average temperature of the model layer, T sis the surface temperature, θ is the satellite observation angle, and v is the channel center wavenumber.

[0127] The key to solve equation (4) is to calculate the transmittance, which is the second term. The main feature of the RTTOV model is to consider the absorption of both the uniform mixing ratio gas (carbon dioxide, etc.) and the variable mixing ratio gas (water vapor) simultaneously. For the absorption of water vapor, the corresponding predictors are selected, and the water vapor related optical thickness is calculated by the statistical regression method, as shown in equation (5):

[0128]

[0129] where σ v,j is the water vapor related optical thickness, a ν,j,k is the regression coefficient, X is the predictor, j is the model layer, v is the channel center wavenumber, and M is the number of predictors.

[0130] Further converted to the transmittance of the water vapor channel, as shown in equation (6):

[0131]

[0132] where τ v,j is the transmittance of the water vapor channel.

[0133] The change in the transmittance of the water vapor channel is calculated by equation (7), as shown in equation (7):

[0134]

[0135] where dτ ν，j is the change in the transmittance of the water vapor channel.

[0136] The selected predictors for the continuous absorption of water vapor in equation (4) are: and W is the model water vapor, and T is the average temperature of the model layer.

[0137] For equation (5), the magnitude of the regression coefficient a ν,j,k is small, and the magnitude of the specific humidity is also small, so the magnitude of the optical thickness value is at most 10 -5, the value of the transmittance of the water vapor channel calculated by formula (6) is close to 1. The change amount of the transmittance calculated by formula (7) is also close to 1. According to formula (4), the horizontal scale of the CSR observation and the model water vapor amount is a 1:1 relationship, and the change of the brightness temperature caused by the change of the water vapor is also a 1:1 relationship. For example, if the resolution of the brightness temperature is 40km, the resolution of the corresponding model water vapor is also 40km. In the CMA-MESO assimilation system, the correlation scale of the given water vapor background error is 100km, which means that the background error of the water vapor field has significant spatial correlation within 100km. In order to effectively assimilate observation data and reduce error, the horizontal resolution of the water vapor observation should match the correlation scale of the background error. According to related research, the observation resolution should be less than the correlation scale of the background error to ensure that the observation data can provide sufficient details to capture the change of the water vapor in the background field. Specifically, the observation resolution can be set to about 1 / 3 to 1 / 2 of the correlation scale of the background error. If the correlation scale of the water vapor background error is 100km, the resolution of the water vapor observation can be set to about 30 to 50km, so the resolution of the FY-4B-AGRI-CSR observation is about 30 to 50km, which can achieve more effective variational assimilation.

[0138] Next, taking the FY4B-AGRI clear sky radiation product (CSR data) as an example, the system resolution is 3km, the background field water vapor correlation scale is 100km, and the upper limit value of the sparsification parameter value is 50km, the contents of steps 103-110 are described in detail.

[0139] Among them, FY4B is the latest Fengyun geostationary satellite operational star, which will drift from east longitude 133 degrees to east longitude 105 degrees from February 1, 2024, providing operational application products. The CSR data is a clear sky optimized data for data assimilation developed by the National Satellite Meteorological Center. See Figures 3-12 Taking 12:00 on May 3, 2024 as an example, the observation distribution of FY4B-AGRI-CSR data after different degrees of sparsification is given. The range in the figure is the model range, and the actual observation area is larger than the model range. After sparsification, only the observations falling within the model range are selected. As shown in Figure 3 and Figure 4 If most of the model area is clear, the L1 data and the CSR data distribution will be similar, and if there are many clouds, the data difference will be obvious, and the CSR distribution obviously does not contain the cloud part. Among them, the L1 data refers to the data product after the original satellite observation data is radiometrically corrected and geometrically corrected. As shown in Figures 5-12As shown, the greater the degree of sparsification, the less the sparsified observation distribution can identify the details of the brightness temperature distribution. When the sparsification parameter value is 200 km, the cloud area and non-cloud area cannot be distinguished, and the details of the brightness temperature distribution cannot be distinguished, and a lot of observation information is lost.

[0140] Referring to Figure 13 , The box in Figure 2 may be regarded as a square box of Figure 13 . The local variance method is to calculate the variance of the observation data in the box. If the sparsification parameter value is small, that is, the degree of sparsification is small, the range of the box is small, the number of observation data contained in the box is small, the distance between Obs2 and Obs4 is small, the degree of similarity is high, the values of the observation data in the box will be closer, the calculated variance is smaller, and the number of boxes in the model range is larger; on the contrary, if the sparsification parameter value is large, that is, the degree of sparsification is large, the range of the box is large, the number of observations contained in the box is large, the distance between Obs2 and Obs4 is large, the degree of similarity is reduced, the difference between the observation data far apart in the box will increase, the calculated variance is larger, and the number of boxes in the model range is smaller.

[0141] For example, in the range of the regional numerical prediction model, the local variance probability density distribution corresponding to the sparsification parameter values of 10 km, 20 km, 30 km, 40 km, 50 km, 60 km, 70 km, 80 km, and 200 km is shown in Figure 14 . When the sparsification parameter value is small, the number of boxes in the model range is large, the number of calculated local variances is large, the shape of the curve of the corresponding local variance probability density distribution is steeper, the tail wing is short, and it is indicated that most of the variance values are too concentrated around the mean value, that is, concentrated around 0, the correlation between observations is high, and the purpose of sparsification cannot be achieved; on the contrary, when the sparsification parameter value is large, the number of boxes in the model range is small, the number of calculated local variances is small, the variance value is large, the shape of the curve of the corresponding local variance probability density distribution is too wide, the tail wing is long, and it is indicated that the degree of deviation of the observation from the mean value is large, although the correlation is reduced, but a lot of observation information is also easily lost.

[0142] Because the local variance probability density distribution is quite different, the local variance probability density distribution can be converted into a local standard deviation probability density distribution, and the standard deviation can more intuitively describe the data distribution. When the first preset value is 0.36 and the second preset value is 1, as shown in Figure 15As shown in the figure, the data in the brackets in the figure respectively represent the value corresponding to the peak of the local standard deviation probability density distribution, the average value of the local standard deviation, the difference between the peak and the average value, and the absolute value of the difference between the peak and the average value is the skewness. The peak of the local standard deviation probability density distribution corresponding to 40 km is 0.85, which is less than the second preset value, and the skewness is 0.36, which is equal to the first preset value. The peak of the local standard deviation probability density distribution corresponding to 30 km is 0.51, which is less than the second preset value, and the skewness is 0.26, which is less than the first preset value. Therefore, 30 km and 40 km are the first target sparse parameter values.

[0143] In combination Figures 3-12 With the distribution comparison, the sparse parameter value of 30-40 km is more in line with the demand and also meets the requirement that the water vapor correlation scale is less than or equal to half of the background field.

[0144] Referring to Figure 16 , the average local variance in the figure is the global variance. Within the range of the regional model, different values of the sparse parameter result in different resolutions of the observation. The variances of the observations with different resolutions after sparse are calculated within the range of the model. As shown in Figure 16 , when the sparse parameter value is less than or equal to 40 km, the variance value does not change significantly. When the sparse parameter value is greater than or equal to 50 km, the variance value increases significantly. When the sparse parameter value is small, the number of observation data retained after sparse is also large, and the small variance value indicates that the difference between the observation data is small and the correlation is high. When the sparse parameter value is large, the number of observation data retained after sparse is small, and the distance between the observation data is relatively far, the difference is large, and the corresponding correlation is reduced, and the representative error is also reduced. The global variances corresponding to the sparse parameter values of 10 km, 20 km, 30 km and 40 km are less than 2, and the global variances corresponding to the sparse parameter values of 50 km, 60 km, 70 km, 80 km and 200 km are greater than 2. The first target sparse parameter values are 30 km and 40 km. Therefore, the second target sparse parameter values obtained by screening are 30 km and 40 km.

[0145] Through the above steps, it can be initially confirmed that for the regional numerical prediction assimilation system with a system resolution of 3 km, the first target sparse parameter value is 30 or 40 km. In order to further confirm the value of the sparse parameter, the CSR observation data are assimilated by using the regional numerical assimilation model, and the difference probability density distribution between the observation brightness temperature and the simulated brightness temperature of the background field obtained in the assimilation process is used for comparative analysis, referring to Figure 17 Within the range of the regional model, different values of the sparse parameter result in different resolutions of the observation. The variances of the observations with different resolutions after sparse are calculated within the range of the model. As shown in Figure 17As shown, the difference distribution between the observation brightness temperature corresponding to the resolution of 40 km and the brightness temperature simulated by the background field is closer to the normal distribution, and the value of the difference between the observation brightness temperature and the brightness temperature simulated by the background field corresponding to the peak is closer to 0.

[0146] In summary, according to the horizontal distribution after the observation is thinned, the local standard deviation probability density distribution of the Coriolis condition, the change of the global variance with the thinning resolution, and the distribution of the difference between the observation brightness temperature and the brightness temperature simulated by the background field after assimilation, it can be determined that the thinning parameter value of 40 km is more suitable for the regional numerical prediction assimilation mode with a resolution of 3 km.

[0147] The embodiments of the present application can divide the functional modules according to the above-mentioned method examples. For example, each functional module can be divided according to each function, or two or more functions can be integrated into one processing module. The integrated module can be realized in the form of hardware or software functional module. It should be noted that the division of the modules in the embodiments of the present application is illustrative, and is only a logical functional division. In actual implementation, another division method can be used.

[0148] In the case of dividing each functional module according to each function, Figure 18 A structure schematic diagram of a meteorological observation thinning parameter determination device provided by the present application is shown. As shown in the figure, Figure 18 The device comprises:

[0149] A data acquisition module 181 is configured to acquire satellite observation data of a to-be-observed region and system information of a regional numerical prediction assimilation system. The system information comprises a background field water vapor related scale and a plurality of preset thinning parameter values.

[0150] A thinning parameter upper limit value determination module 182 is configured to determine an upper limit value of the thinning parameter based on the background field water vapor related scale.

[0151] A horizontal thinning grid division module 183 is configured to perform horizontal grid division on the satellite observation data by using the thinning parameter value, to obtain a plurality of thinning grid cells. The thinning grid cell is a near-rectangular shape, and the thinning parameter value is the side length of the thinning grid cell.

[0152] A local standard deviation probability density distribution result determination module 184 is configured to calculate a local standard deviation corresponding to each thinning grid cell, and determine a local standard deviation probability density distribution corresponding to each thinning parameter value according to the local standard deviation.

[0153] A first target thinning parameter value determination module 185 is configured to determine a first target thinning parameter value based on the upper limit value and the local standard deviation probability density distribution corresponding to each thinning parameter value.

[0154] Optionally, the local standard deviation probability density distribution result determination module 184 can specifically include:

[0155] a local standard deviation calculation unit, configured to calculate a local standard deviation corresponding to each sparse grid unit according to a formula:

[0156]

[0157]

[0158] wherein, is the local standard deviation, N is a total number of observations in the sparse grid unit, x i is a value of the ith observation, and μ is an average value of the N observations in the sparse grid unit;

[0159] a local standard deviation probability density distribution determination unit, configured to perform square root calculation on the local standard deviation to obtain a local standard deviation corresponding to the sparse grid unit, and generate a local standard deviation probability density distribution according to the local standard deviation.

[0160] Optionally, the device further includes a second target sparse parameter value determination module, including:

[0161] a sparse processing unit, configured to perform sparse processing on the satellite observation data by using the first target sparse parameter value, to obtain sparse data in a model range;

[0162] a global variance calculation unit, configured to calculate a variance corresponding to the model range according to a formula:

[0163]

[0164]

[0165] wherein, is the global variance, N1 is a total number of observations of the sparse data in the model range, x i1 is a value of the ith observation, and μ1 is an average value of the N1 observations in the model range;

[0166] a second target sparse parameter value determination unit, configured to determine the first target sparse parameter value corresponding to the global variance less than a preset threshold as a second target sparse parameter value.

[0167] Optionally, the device further includes a third target sparse parameter value determination module, including:

[0168] a regional numerical prediction assimilation unit, configured to determine, based on a regional numerical prediction assimilation system, an observation and background field difference value probability density distribution corresponding to each first target sparse parameter value; ​​

[0169] The third target sparsification parameter value determination unit is configured to determine the first target sparsification parameter value corresponding to the observation and background field difference value probability density distribution closest to the normal distribution as the third target sparsification parameter value.

[0170] Optionally, the regional numerical prediction assimilation unit can be specifically configured to:

[0171] The formula is as follows:

[0172]

[0173] Determine the observation and background field difference value data set corresponding to each first target sparsification parameter value;

[0174] Wherein, J(x) is the weighted sum of the background field error and the observation error, x is the analysis variable, i.e., the analysis field generated after assimilation, x b is the background field; H is the observation operator, y o is the observation field; B is the background error covariance matrix, O is the observation covariance matrix, and T represents the matrix transpose;

[0175] Determine the observation and background field difference value probability density distribution corresponding to the first target sparsification parameter value according to the observation and background field difference value data set.

[0176] Optionally, the upper limit value is half of the water vapor related scale of the background field.

[0177] Optionally, the horizontal scale relationship between the model variable and the observation variable corresponding to the observation operator is 1:1.

[0178] The above mainly introduces the scheme provided by the embodiment of the application from the perspective of the interaction between various modules. It can be understood that, in order to realize the above functions, it contains the hardware structure and / or software module corresponding to the execution of each function. Those skilled in the art should easily realize that, in combination with the units and algorithm steps of the examples described in the embodiments disclosed in the present text, the application can be realized in the form of hardware or the combination of hardware and computer software. Whether a certain function is executed in the form of hardware or computer software driven hardware depends on the specific application and design constraint conditions of the technical solution. The professional technicians can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the application.

[0179] In the case of using the corresponding integration unit, Figure 19 A structure diagram of a meteorological observation sparsification parameter determination device provided by the application is shown in FIG. 1. Figure 19 As shown in the figure, the device comprises:

[0180] The communication unit / communication interface is configured to obtain satellite observation data of a region to be observed and system information of a regional numerical prediction assimilation system, wherein the system information comprises a background field water vapor correlation scale and a plurality of preset sparse parameter values.

[0181] The processing unit / processor is configured to determine an upper limit value of the sparse parameter based on the background field water vapor correlation scale.

[0182] The satellite observation data is horizontally grid-divided using the sparse parameter values to obtain a plurality of sparse grid cells.

[0183] The local standard deviation corresponding to each sparse grid cell is calculated, and the local standard deviation probability density distribution corresponding to each sparse parameter value is determined according to the local standard deviation.

[0184] The first target sparse parameter value is determined based on the upper limit value and the local standard deviation probability density distribution corresponding to each sparse parameter value.

[0185] The processing unit can be a processor or a controller, such as a central processing unit (CPU), a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, a transistor logic device, a hardware component, or any combination thereof. It can implement or execute various exemplary logical blocks, modules and circuits described in combination with the present disclosure. The processor can also be a combination of computing functions, such as one or more microprocessor combinations, DSP and microprocessor combinations, etc. The communication module can be a transceiver, transceiver circuit or communication interface, etc. The storage module can be a memory.

[0186] As shown in Figure 19 the above processor can be a general central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of programs of the present application. The above communication interface can be one or more. The communication interface can use any transceiver device to communicate with other devices or communication networks.

[0187] As shown in Figure 19As shown, the terminal device described above may also include a communication line. The communication line may include a path for transmitting information between the components described above.

[0188] Optional, such as Figure 19 As shown, the terminal device may further include a memory. The memory stores computer execution instructions for implementing the present invention, and the execution is controlled by a processor. The processor executes the computer execution instructions stored in the memory, thereby implementing the method provided in the embodiments of the present invention.

[0189] like Figure 19 As shown, the memory can be read-only memory (ROM) or other types of static storage devices capable of storing static information and instructions, random access memory (RAM) or other types of dynamic storage devices capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed discs, laser discs, optical discs, universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited to these. The memory can exist independently and be connected to the processor via communication lines. The memory can also be integrated with the processor.

[0190] In a specific implementation, as one example, such as Figure 19 As shown, a processor may include one or more CPUs, such as Figure 19 CPU0 and CPU1 in the CPU.

[0191] In a specific implementation, as one example, such as Figure 19 As shown, the terminal device may include multiple processors, such as Figure 19 The processors in the system. Each of these processors can be a single-core processor or a multi-core processor.

[0192] On the one hand, a computer-readable storage medium is provided, which stores instructions that, when executed, enable a method for determining sparsity parameters for meteorological observations.

[0193] In the above embodiments, all or part can be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part can be implemented in the form of a computer program product. The computer program product includes one or more computer programs or instructions. When the computer programs or instructions are loaded on a computer and executed, all or part of the processes or functions described in the embodiments of the present application are performed. The computer can be a general purpose computer, a special purpose computer, a computer network, a terminal, a user equipment or other programmable apparatus. The computer programs or instructions can be stored in a computer readable storage medium or transferred from one computer readable storage medium to another computer readable storage medium, for example, the computer programs or instructions can be transferred from one website site, computer, server or data center to another website site, computer, server or data center through wired or wireless manner. The computer readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center and the like integrated with one or more available media. The available media can be a magnetic medium, such as a floppy disk, a hard disk, a magnetic tape; an optical medium, such as a digital video disc (digital video disc, DVD); and a semiconductor medium, such as a solid state drive (solid state drive, SSD).

[0194] Although the present application has been described in connection with various embodiments thereof, it will be understood that other variations and modifications of the disclosed embodiments can be made by those skilled in the art upon reading the description of the application set forth above. In the claims, the word "comprising" does not exclude other components or steps not mentioned in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. A single processor or other unit can fulfill the functions of several items recited in the claims. Several measures can be combined into one measure. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

[0195] Although the present application has been described in connection with specific features thereof, it will be understood that various modifications and variations can be made by those skilled in the art without departing from the spirit and scope of the application. Accordingly, it is intended that the present description and drawings be considered as illustrative only and that all modifications and variations are intended to be included within the scope of the application as defined by the appended claims. It is also to be understood that the application is not to be limited to the specific examples described and / or shown, as such may vary. It is therefore contemplated to cover by the present application any and all modifications, variations, combinations or equivalents that fall within the scope of the present application.

Claims

1. A method for determining sparsity parameters in meteorological observations, characterized in that, include: Acquire satellite observation data of the area to be observed and system information of the regional numerical weather prediction assimilation system; The system information includes background water vapor correlation scales and multiple preset sparsification parameter values; The upper limit of the sparsification parameter is determined based on the background field water vapor correlation scale; The satellite observation data is divided into horizontal grids using the sparsification parameter values ​​to obtain multiple sparse grid cells; Calculate the local standard deviation corresponding to each of the sparsed grid cells, and determine the local standard deviation probability density distribution corresponding to each sparsed parameter value based on the local standard deviation; The first target sparsification parameter value is determined based on the upper limit value and the local standard deviation probability density distribution corresponding to each sparsification parameter value. Specifically, sparsification parameter values ​​that are less than the upper limit value and whose corresponding local standard deviation probability density distributions satisfy preset conditions are determined as the first target sparsification parameter values. The prediction conditions are that the skewness of the local standard deviation probability density distribution is less than or equal to a first preset value and the peak value of the local standard deviation probability density distribution is less than or equal to a second preset value. The skewness is the distance between the peak value and the average value of the local standard deviation probability density. The first preset value and the second preset value can be set according to observation requirements. The calculation of the local standard deviation corresponding to each of the sparsed grid cells, and the determination of the local standard deviation probability density distribution corresponding to each sparsed parameter value based on the local standard deviation, includes: According to the formula: ； Calculate the local locality of each sparsed grid cell; in, Due to local differences, The total number of observations within the sparse grid cell. For the first The value of each observation, Within sparse grid cells The average of the observations; The local standard deviation corresponding to the sparse grid cell is obtained by taking the square root of the local standard deviation, and a local standard deviation probability density distribution is generated based on the local standard deviation.

2. The method for determining sparsity parameters in meteorological observations according to claim 1, characterized in that, The process of determining the first target sparsification parameter value based on the upper limit value and the local standard deviation probability density distribution corresponding to each sparsification parameter value further includes: The satellite observation data is sparsified using the first target sparsification parameter value to obtain sparsed data within the mode range; According to the formula: ； Calculate the variance corresponding to the range of the patterns to obtain the global variance; in, For global variance, The total number of observations for sparse data within the model range. For the first The value of each observation, Within the mode range The average of the observations; The first objective sparsification parameter value corresponding to the global variance that is less than the preset threshold is determined as the second objective sparsification parameter value.

3. The method for determining sparsity parameters in meteorological observations according to claim 1, characterized in that, The process of determining the first target sparsification parameter value based on the upper limit value and the local standard deviation probability density distribution corresponding to each sparsification parameter value further includes: Based on the regional numerical weather prediction assimilation system, the probability density distribution of the difference between the observation and the background field corresponding to the sparsification parameter value of each first target is determined; The first objective sparsification parameter value corresponding to the probability density distribution of the difference between the observation and the background field that is closest to the normal distribution is determined as the third objective sparsification parameter value.

4. The method for determining sparsity parameters in meteorological observations according to claim 3, characterized in that, The method for determining the probability density distribution of the observation-background field difference corresponding to each first target sparsification parameter value based on the regional numerical prediction assimilation system includes: Formula used: ； Determine the observation-background field difference dataset corresponding to each first objective sparsification parameter value; in, This is a weighted sum of background field error and observation error. For the analysis variables, i.e. the analysis field generated after assimilation, For background scene; For the observation operator, For the observation field; The background error covariance matrix, To observe the covariance matrix, Indicates matrix transpose; The probability density distribution of the observation and background field difference corresponding to the first target sparsification parameter value is determined based on the observation and background field difference dataset.

5. The method for determining sparsity parameters in meteorological observations according to claim 1, characterized in that, The upper limit value is half of the background field water vapor correlation scale.

6. The method for determining sparsity parameters in meteorological observations according to claim 4, characterized in that, The horizontal scale relationship between the model variable and the observation corresponding to the observation operator is 1:

1.

7. A device for determining sparsity parameters in meteorological observations, characterized in that, include: The data acquisition module is used to acquire satellite observation data of the area to be observed and system information of the regional numerical weather prediction assimilation system; The system information includes background water vapor correlation scales and multiple preset sparsification parameter values; The sparsity parameter upper limit determination module is used to determine the upper limit value of the sparsity parameter based on the background field water vapor correlation scale; A horizontal sparse grid partitioning module is used to partition the satellite observation data into horizontal grids using the sparsification parameter values ​​to obtain multiple sparse grid cells. The local standard deviation probability density distribution result determination module is used to calculate the local standard deviation corresponding to each of the sparsed grid cells, and determine the local standard deviation probability density distribution corresponding to each sparsed parameter value based on the local standard deviation. The module for determining the local standard deviation probability density distribution results includes: The local variance calculation unit is used to calculate the variance based on the formula: ； Calculate the local locality of each sparsed grid cell; in, Due to local differences, The total number of observations within the sparse grid cell. For the first The value of each observation, Within sparse grid cells The average of the observations; The local standard deviation probability density distribution determination unit is used to calculate the local standard deviation corresponding to the sparse grid cell by taking the square root of the local standard deviation, and to generate the local standard deviation probability density distribution based on the local standard deviation. The first target sparsification parameter value determination module is used to determine the first target sparsification parameter value based on the upper limit value and the local standard deviation probability density distribution corresponding to each sparsification parameter value. Specifically, sparsification parameter values ​​that are less than the upper limit value and whose corresponding local standard deviation probability density distributions satisfy preset conditions are determined as the first target sparsification parameter values. The prediction conditions are that the skewness of the local standard deviation probability density distribution is less than or equal to a first preset value and the peak value of the local standard deviation probability density distribution is less than or equal to a second preset value. The skewness is the distance between the peak value and the average value of the local standard deviation probability density. The first preset value and the second preset value can be set according to observation requirements.

8. A device for determining sparsity parameters in meteorological observations, characterized in that, include: The communication unit / communication interface is used to acquire satellite observation data of the area to be observed and system information of the regional numerical weather prediction assimilation system. The system information includes background water vapor correlation scales and multiple preset sparsification parameter values; A processing unit / processor is used to determine the upper limit of the sparsification parameter based on the background field water vapor correlation scale; The satellite observation data is divided into horizontal grids using the sparsification parameter values ​​to obtain multiple sparse grid cells; Calculate the local standard deviation corresponding to each of the sparsed grid cells, and determine the local standard deviation probability density distribution corresponding to each sparsed parameter value based on the local standard deviation; The first target sparsification parameter value is determined based on the upper limit value and the local standard deviation probability density distribution corresponding to each sparsification parameter value. Specifically, sparsification parameter values ​​that are less than the upper limit value and whose corresponding local standard deviation probability density distributions satisfy preset conditions are determined as the first target sparsification parameter values. The prediction conditions are that the skewness of the local standard deviation probability density distribution is less than or equal to a first preset value and the peak value of the local standard deviation probability density distribution is less than or equal to a second preset value. The skewness is the distance between the peak value and the average value of the local standard deviation probability density. The first preset value and the second preset value can be set according to observation requirements. The calculation of the local standard deviation corresponding to each of the sparsed grid cells, and the determination of the local standard deviation probability density distribution corresponding to each sparsed parameter value based on the local standard deviation, includes: According to the formula: ； Calculate the local locality of each sparsed grid cell; in, Due to local differences, The total number of observations within the sparse grid cell. For the first The value of each observation, Within sparse grid cells The average of the observations; The local standard deviation corresponding to the sparse grid cell is obtained by taking the square root of the local standard deviation, and a local standard deviation probability density distribution is generated based on the local standard deviation.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed, implement the method for determining sparsity parameters of meteorological observations as described in any one of claims 1-6.

Images