A method for full-sky satellite data assimilation based on weather scale and weather background dependence

Abstract

The present invention discloses a satellite data full-space assimilation method based on weather scale and weather background dependence, comprising the following steps: (1) reading a background error covariance matrix containing information of N sample members, calculating and defining a scale classification operator; (2) reading the background error covariance matrix containing information of N sample members again, and constructing an adaptive partitioned background error covariance of variable scale dependence; (3) calculating a background error covariance based on large-scale variables; (4) calculating a background error covariance based on small-scale variables; (5) reading information containing background error covariance and its diagnostic quantity, satellite observation information, satellite observation error information and model background field, and inputting them into a variational assimilation system; (6) obtaining a model initial field with the best statistical significance; the present invention improves the accuracy of subsequent numerical forecasts.

Description

Technical Field

[0001] The present invention relates to the field of atmospheric science and technology, and in particular to a satellite data full-space assimilation method based on weather scale and weather background dependence. Background Art

[0002] At present, data assimilation is one of the key means to improve the model initial field and thus improve the accuracy of numerical weather forecasts. The assimilation technology for satellite data can be divided into two categories: clear sky assimilation and full sky assimilation. In clear sky assimilation, the observation data in the cloud and rain area will be eliminated, which greatly reduces the utilization rate of satellite data, and the occurrence and development of some disastrous weather systems are always accompanied by cloud and rain processes. Therefore, the development of full sky assimilation technology for satellite data is of great significance to further improve the numerical forecast level of disastrous weather systems. One of the key technologies for successfully assimilating satellite observation data in the full sky (including clear sky areas and cloud and rain areas) is how to reasonably introduce hydrometeor control variables when constructing the background error covariance, and sequentially construct satellite data direct assimilation modules, tangent modules and adjoint modules containing hydrometeor information, so as to realize direct analysis of hydrometeor information.

[0003] Background error covariance plays a very important role in the data assimilation system. It not only determines the ability of observational data to transmit information in the horizontal and vertical directions, but also determines the contribution of observational and background information to the analysis increment together with observational errors. It also includes information transmission and balance relationship between variables, etc. In the construction of the background field error covariance matrix, it is also particularly important to construct the balance relationship between hydrometeors and conventional variables reasonably and appropriately. Through the balance relationship, the coordination between the control variables can be maintained, and the weather information can be transmitted through the implicit dynamic constraint relationship between the variables, and the observation information can be used to the greatest extent. On the other hand, compared with large-scale traditional variables (temperature, humidity, air pressure, wind, etc.), the distribution of hydrometeor variables (cloud water, cloud ice, rain, snow, and sleet) has the characteristics of smaller scale and spatial discontinuity. The background error characteristics of hydrometeors under different weather backgrounds are also very different, and the error characteristics between small-scale hydrometeor variables and large-scale traditional variables are also significantly different.

[0004] At present, there is no research on adaptive assimilation of satellite observation data under all sky based on different weather scales and different weather backgrounds in satellite data assimilation system. At present, most of the existing technologies use isotropic and homogeneous background field error covariance, ignoring the weather system dependence of background field error covariance, that is, ignoring the obvious different characteristics of background field error covariance under different weather scales and different weather backgrounds. In addition, most of the existing technologies do not consider the background error covariance based on hydrometeor control variables, and do not consider the physical balance relationship between hydrometeor control variables and other conventional control variables when constructing the background error covariance. The above existing technical problems will affect the calculation accuracy of the background error covariance value in the assimilation system, further affecting the later assimilation effect, especially the cloud and rain area data assimilation effect. The introduction of ensemble flow-dependent background field error covariance in the assimilation system will be limited by the number of ensemble samples and will consume a lot of computing resources. Therefore, it is very necessary to construct an anisotropic static background error covariance matrix that is more in line with weather characteristics in the variational assimilation system. Summary of the invention

[0005] Purpose of the invention: The purpose of the present invention is to provide a method for full-sky assimilation of satellite data based on weather scale and weather background dependence, to solve the problem of inaccurate calculation of background error covariance, so as to further improve the application effect of full-sky assimilation of satellite data.

[0006] Technical solution: The present invention discloses a method for assimilating satellite data in the whole sky based on weather scale and weather background dependence, comprising the following steps:

[0007] (1) Read the background error covariance matrix containing N sample member information, calculate and define the scale classification operator;

[0008] (2) Read the background error covariance matrix containing the information of N sample members again to construct the variable scale-dependent adaptive partition background error covariance;

[0009] (3) Calculate the background error covariance based on large-scale variables;

[0010] (4) Calculate the background error covariance based on small-scale variables;

[0011] (5) Read the background error covariance and its diagnostic value, satellite observation information, satellite observation error information and model background field, and input them into the variational assimilation system;

[0012] (6) Obtain the model initial field with the best statistical significance.

[0013] Furthermore, step (1) includes the following steps:

[0014] (1);

[0015] Where L is the horizontal characteristic length scale, D represents the variance, Represents a non-equilibrium physical quantity field;

[0016] (12) Read the horizontal characteristic length scale of each variable, diagnose the horizontal characteristic length scale of each variable in the first mode, and define the variable scale classification and discrimination standard, unit: km, the formula is as follows:

[0017] (2) ;

[0018] in, represents a large-scale classification operator, Represents a small-scale classification operator.

[0019] Furthermore, step (2) includes the following steps:

[0020] (21) According to the variable transformation operator in the physical transformation, the mode variables in the covariance matrix are converted into state variables, which include conventional variables: temperature, pressure, humidity, wind, and hydrometeor variables: cloud water, cloud ice, rain, snow, and graupel;

[0021] (22) According to the variable scale classification judgment standard of formula (2), each state variable is classified into large-scale continuous variables and small-scale discontinuous variables, and the large-scale classification operator is constructed based on the sample set divided by the variable scale. and small-scale classification operators ; Among them, large-scale continuous variables include: temperature T, air pressure , pseudo relative humidity , zonal wind U, meridional wind V; small-scale discontinuity variables include: cloud water , Yunbing ,rainwater ,Snow , and graupel ;

[0022] (23) Based on large-scale classification operators and small-scale classification operators , construct the background error covariance matrix of adaptive partitions of different variable scales , the formula is as follows:

[0023] (3);

[0024] in, represents the background error covariance matrix of the large-scale weather area, Represents the background error covariance matrix of the small-scale weather area.

[0025] Furthermore, step (3) includes the following steps:

[0026] (31) According to the equilibrium transformation operator in physical transformation, the equilibrium relationship between various conventional variables is statistically calculated through linear regression, and the formula is as follows:

[0027] (4);

[0028] in, is the complete field control variable, is the unbalanced part of various control variables, is the linear regression coefficient between the two control variables;

[0029] (32) Based on the complete field or non-equilibrium part of the various types of large-scale variables, according to the vertical transformation operator and the horizontal transformation operator, the vertical error eigenvalue and the horizontal error eigenvalue of the large-scale variables are diagnosed and calculated; wherein the vertical error eigenvalue includes an eigenvalue and an eigenvector; and the horizontal error eigenvalue includes a horizontal characteristic length scale.

[0030] Furthermore, step (4) includes the following steps:

[0031] (41) Read N sample sets containing forecast error information, and calculate the total vertical content and average value of the hydrometeoric content of the N sample members. The formula is as follows:

[0032] (5);

[0033] in, represents the total contents of cloud water, cloud ice and rainwater integrated from the top of the model layer to the bottom of the model layer in the vertical direction based on the nth error sample;

[0034] (6);

[0035] in, They represent cloud water, cloud ice and rain water respectively. represents the average total content of cloud water, cloud ice and rainwater calculated based on N samples in the vertical direction from the top of the model layer to the bottom of the model layer;

[0036] (42) Based on the vertically integrated total content and average value of three types of hydrometeors, namely, cloud water, cloud ice and rain water, classification operators for different weather background situations are constructed, including: is the cloud-free area operator, Cloud and water area operator, Cloud ice area operator, Cloud water and cloud ice mixed zone operator, Cloud-rain mixed zone operator;

[0037] (43) Based on the five different hydrometeorological weather situation classification operators and formula (7), the background error covariance matrix based on small-scale variables is adaptively divided into cloud-free areas. , Yunshui District , Yunbing District , cloud water and cloud ice mixed area and mixed cloud and rain areas The sum of five parts:

[0038] (7);

[0039] (44) The background error covariance matrix of each region is calculated. According to the equilibrium transformation operator in physical transformation, the equilibrium relationship between the hydrometeoric control variable and other conventional variables in each region is statistically analyzed by linear regression. The formula is as follows:

[0040] (8);

[0041] in, is the complete field control variable, is the unbalanced part of various control variables, represents the linear regression coefficient between the two control variables;

[0042] (45) Based on the complete field or non-equilibrium part of the various small-scale variables, vertical transformation operators and horizontal transformation operators are applied to diagnose and calculate the vertical error eigenvalue and horizontal error eigenvalue of each variable.

[0043] Furthermore, step (6) is specifically as follows: In the variational assimilation system, based on the hydrometeor information calculated from the model background field, a satellite radiance forward simulation operator, an adjoint operator and a tangent operator containing the hydrometeor information are constructed, and assimilation iterative analysis is performed to generate an analysis increment containing the hydrometeor information.

[0044] The satellite data full-space assimilation system based on weather scale and weather background dependence of the present invention comprises:

[0045] Scale classification operator module: used to read the background error covariance matrix containing N sample member information, calculate and define the scale classification operator;

[0046] Partition background error covariance module: used to read the background error covariance matrix containing N sample member information again, and construct the variable scale-dependent adaptive partition background error covariance;

[0047] Large-scale variable module: used to calculate the background error covariance based on large-scale variables;

[0048] Small-scale variable module: used to calculate the background error covariance based on small-scale variables;

[0049] Reading module: used to read the background error covariance and its diagnostic quantity, satellite observation information, satellite observation error information and model background field, and input them into the variational assimilation system;

[0050] Optimal model initial field module: used to obtain the statistically optimal model initial field.

[0051] An electronic device described in the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded into the processor, any one of the satellite data full-space assimilation methods based on weather scale and weather background dependence is implemented.

[0052] A storage medium described in the present invention stores a computer program, and when the computer program is executed by a processor, any one of the satellite data full-space assimilation methods based on weather scale and weather background dependence is implemented.

[0053] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: effectively constructing anisotropic and inhomogeneous "variable characteristic scale and hydrometeor weather situation dependence" adaptive partitioned background field error covariance, realizing adaptive adjustment of background error covariance under different weather scales and different weather backgrounds, and sequentially constructing satellite data direct assimilation module, tangent module and adjoint module containing hydrometeor information, further realizing adaptive and reasonable analysis of conventional variables and hydrometeor variables, generating corresponding inhomogeneous and anisotropic analysis increments, optimizing the assimilation effect of satellite observation data under the whole sky (including clear sky area and cloud and rain area), and improving the accuracy of later numerical forecasts. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is a framework diagram of the present invention. DETAILED DESCRIPTION

[0055] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0056] like Figure 1 As shown, the embodiment of the present invention provides step 1) at the same time, based on the reanalysis data as the initial field of the numerical forecast model, the model is backward integrated for two types of different forecast durations, and the forecast error sample at the time is obtained according to the difference characteristics between the forecast fields of different forecast durations;

[0057] Step 2) Repeat the above steps at multiple times to construct a set of forecast error sample sets;

[0058] Step 3) Read the forecast error sample set, and construct a background error covariance matrix containing information of N sample members by using the NMC (National Meteorological Center) method.

[0059] Step 4) reading the background error covariance matrix containing the N sample member information, diagnosing and defining the scale classification operator, including the following steps:

[0060] Step 4-1) Apply the horizontal transformation operator and diagnose and calculate the horizontal characteristic length scale L of each variable by solving formula (1):

[0061] (1);

[0062] Where L is the horizontal characteristic length scale, D represents the variance, Represents a non-equilibrium physical quantity field;

[0063] Step 4-2) Read the horizontal characteristic length scale of each variable, diagnose the horizontal characteristic length scale of each variable in the first mode, and define the variable scale classification and discrimination standard, unit: km, the formula is as follows:

[0064] (2) ;

[0065] in, represents a large-scale classification operator, Represents a small-scale classification operator.

[0066] Step 5) reading the background error covariance matrix containing the information of N sample members again, and constructing the variable scale-dependent adaptive partition background error covariance, including the following steps:

[0067] Step 5-1) Apply the variable transformation operator in the physical transformation to transform the mode variables in the covariance matrix into state variables, which include conventional variables: temperature, pressure, humidity, wind, and hydrometeor variables: cloud water, cloud ice, rain, snow, and graupel;

[0068] Step 5-2) According to step 4), each state variable is classified into large-scale continuous variables and small-scale discontinuous variables, and large-scale classification operators are constructed based on the sample sets divided by variable scales. and small-scale classification operators ; Among them, large-scale continuous variables include: temperature T, air pressure , pseudo relative humidity , zonal wind U, meridional wind V; small-scale discontinuity variables include: cloud water , Yunbing ,rainwater ,Snow , and graupel ;

[0069] Step 5-3) Based on large-scale classification operator and small-scale classification operators , construct the background error covariance matrix of adaptive partitions of different variable scales , the formula is as follows:

[0070] (3);

[0071] in, represents the background error covariance matrix of the large-scale weather area, Represents the background error covariance matrix of the small-scale weather area.

[0072] Step 6) Calculate the background error covariance matrix based on the large-scale variables , including the following steps:

[0073] Step 6-1) Apply the balance transformation operator in physical transformation and use formula (4) to linearly regress and calculate the balance relationship between various conventional variables.

[0074] (4);

[0075] in, is the complete field control variable, is the unbalanced part of various control variables, is the linear regression coefficient between the two control variables;

[0076] Step 6-2) Based on the complete field or non-equilibrium part of the various types of large-scale variables, vertical transformation operators and horizontal transformation operators are applied to diagnose and calculate vertical error eigenvalues ​​(eigenvalues ​​and eigenvectors) and horizontal error eigenvalues ​​(horizontal characteristic length scales) based on the large-scale variables;

[0077] Step 6-3) Output the background error covariance based on large-scale variables and its various diagnostic values, and store them in NC format file 1.

[0078] Step 7) Calculate the background error covariance based on the small-scale variables , including the following steps:

[0079] Step 7-1) Read the forecast error sample set in step 2) and calculate the total vertical content and average value of the hydrometeoric content of N sample members according to formula (5) and formula (6):

[0080] (5);

[0081] in, represents the total contents of cloud water, cloud ice and rainwater integrated from the top of the model layer to the bottom of the model layer in the vertical direction based on the nth error sample;

[0082] (6);

[0083] in, They represent cloud water, cloud ice and rain water respectively. represents the average total content of cloud water, cloud ice and rainwater calculated based on N samples in the vertical direction from the top of the model layer to the bottom of the model layer;

[0084] Step 7-2) Based on the vertically integrated total content and average value of three types of hydrometeors, namely, cloud water, cloud ice and rain water, construct classification operators for different weather background situations, including: is the cloud-free area operator, Cloud and water area operator, Cloud ice area operator, Cloud water and cloud ice mixed zone operator, Cloud-rain mixed zone operator;

[0085] Step 7-3) Based on the five different hydrometeorological weather situation classification operators and formula (7), the background error covariance matrix based on small-scale variables is adaptively divided into cloud-free areas , Yunshui District , Yunbing District , cloud water and cloud ice mixed area and mixed cloud and rain areas The sum of five parts:

[0086] (7);

[0087] Step 7-4) Calculate the background error covariance matrix of each region. According to the balance transformation operator in the physical transformation, the balance relationship between the hydrometeor control variables and other conventional variables in each region is statistically analyzed by linear regression. The formula is as follows:

[0088] (8);

[0089] in, is the complete field control variable, is the unbalanced part of various control variables, represents the linear regression coefficient between the two control variables;

[0090] Step 7-5) Based on the complete field or non-equilibrium part of the various small-scale variables, vertical transformation operators and horizontal transformation operators are applied to diagnose and calculate the vertical error eigenvalue and horizontal error eigenvalue of each variable.

[0091] Step 7-6) Output the hydrometeoric characteristic dependence partition background error covariance and its various diagnostic values ​​based on small-scale variables and store them in NC format file 2.

[0092] Step 8) Read the file containing the background error covariance and its diagnostic value, the file containing the satellite observation data information, the file containing the observation error and the model initial guess field, and input the read information into the variational assimilation system at the same time;

[0093] Step 9) In the variational assimilation system, based on the hydrometeor information calculated from the model background field, a satellite observation radiance forward simulation operator, adjoint operator, and tangent operator containing hydrometeor information are constructed to perform assimilation iterative analysis, generate analysis increments containing hydrometeor information, and obtain the optimal model initial field (analysis field).

[0094] Step 10) Deterministic numerical forecast is performed based on the assimilated analysis field, and the forecast field for six hours is used as the model background field and input into the assimilation system to participate in the next cycle assimilation.

[0095] The full-sky assimilation method of satellite data of different weather scales and weather background dependence in this embodiment constructs the background error covariance matrix based on large-scale continuous variables and the background error covariance matrix of small-scale discontinuous variables through variable scale discriminant operator classification, and can adaptively construct the background error covariance matrix according to the horizontal characteristic scale of each variable. For samples of large-scale continuous weather areas, the isotropic and homogeneous background error covariance is calculated. For samples of small-scale discontinuous weather areas, according to the content of various types of hydrometeors in the vertical direction of the background field, a "hydrometeor weather situation dependence" and heterogeneous background error covariance matrix is ​​further constructed, and the balance relationship between the statistical hydrometeor variables and the conventional variables in the weather background area is divided. In addition, a direct assimilation module, a tangent module and an adjoint module of satellite observation data containing hydrometeor information are constructed in sequence. This invention not only realizes the adaptive adjustment of the background error covariance under different weather scales and different weather backgrounds in the variational assimilation system, but also effectively assimilates the satellite observation data based on the direct assimilation, tangent and adjoint operators containing hydrometeor information, further realizes the adaptive and reasonable analysis of conventional variables and hydrometeor variables, generates corresponding heterogeneous and anisotropic analysis increments, and optimizes the assimilation effect of satellite observation data under the whole sky; moreover, the adaptive background error covariance can more reasonably characterize the characteristics of the background error under the whole sky assimilation, and there is no need for repeated statistics, which greatly improves the calculation efficiency.

[0096] The embodiment of the present invention further provides a satellite data full-space assimilation system based on weather scale and weather background dependence, comprising:

[0097] Scale classification operator module: used to read the background error covariance matrix containing N sample member information, calculate and define the scale classification operator;

[0098] Partition background error covariance module: used to read the background error covariance matrix containing N sample member information again, and construct the variable scale-dependent adaptive partition background error covariance;

[0099] Large-scale variable module: used to calculate the background error covariance based on large-scale variables;

[0100] Small-scale variable module: used to calculate the background error covariance based on small-scale variables;

[0101] Reading module: used to read the background error covariance and its diagnostic quantity, satellite observation information, satellite observation error information and model background field, and input them into the variational assimilation system;

[0102] Optimal model initial field module: used to obtain the statistically optimal model initial field.

[0103] An embodiment of the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded into the processor, any one of the satellite data full-sky assimilation methods based on weather scale and weather background dependence is implemented.

[0104] An embodiment of the present invention further provides a storage medium storing a computer program, wherein the computer program, when executed by a processor, implements any one of the satellite data full-sky assimilation methods based on weather scale and weather background dependence.

Claims

1. A method for assimilating satellite data in the whole sky based on weather scale and weather background dependence, characterized in that: The following steps are involved: (1) Read the background error covariance matrix containing N sample member information, calculate and define the scale classification operator; including the following steps: (11) According to the horizontal transformation operator, the horizontal characteristic length scale L of each variable is calculated as follows: (1)； Where L is the horizontal characteristic length scale, D represents the variance, Represents a non-equilibrium physical quantity field; (12) Read the horizontal characteristic length scale of each variable, diagnose the horizontal characteristic length scale of each variable in the first mode, and define the variable scale classification and discrimination standard, unit: km, the formula is as follows: (2) ； in, represents a large-scale classification operator, represents a small-scale classification operator; (2) Read the background error covariance matrix containing the information of N sample members again to construct the variable scale-dependent adaptive partition background error covariance; (3) Calculating the background error covariance based on large-scale variables; including the following steps: (31) According to the equilibrium transformation operator in physical transformation, the equilibrium relationship between various conventional variables is statistically calculated through linear regression, and the formula is as follows: （4）； in, is the complete field control variable, is the unbalanced part of various control variables, is the linear regression coefficient between the two control variables; (32) Based on the complete field or non-equilibrium part of various large-scale variables, according to the vertical transformation operator and the horizontal transformation operator, the vertical error eigenvalue and the horizontal error eigenvalue of the large-scale variables are diagnosed and calculated; wherein the vertical error eigenvalue includes the eigenvalue and the eigenvector; the horizontal error eigenvalue includes the horizontal characteristic length scale; (4) Calculating the background error covariance based on small-scale variables; including the following steps: (41) Read N sample sets containing forecast error information, and calculate the total vertical content and average value of the hydrometeoric content of the N sample members. The formula is as follows: (5)； in, represents the total contents of cloud water, cloud ice and rainwater integrated from the top of the model layer to the bottom of the model layer in the vertical direction based on the nth error sample; (6)； in, They represent cloud water, cloud ice and rain water respectively. represents the average total content of cloud water, cloud ice and rainwater calculated based on N samples in the vertical direction from the top of the model layer to the bottom of the model layer; (42) Based on the vertically integrated total content and average value of three types of hydrometeors, namely, cloud water, cloud ice and rain water, classification operators for different weather background situations are constructed, including: is the cloud-free area operator, Cloud and water area operator, Cloud ice area operator, Cloud water and cloud ice mixed zone operator, Cloud-rain mixed zone operator; (43) Based on the five different hydrometeorological weather situation classification operators and formula (7), the background error covariance matrix based on small-scale variables is adaptively divided into cloud-free areas. , Yunshui District , Yunbing District , cloud water and cloud ice mixed area and mixed cloud and rain areas The sum of five parts: (7)； (44) The background error covariance matrix of each region is calculated. According to the equilibrium transformation operator in physical transformation, the equilibrium relationship between the hydrometeoric control variable and other conventional variables in each region is statistically analyzed by linear regression. The formula is as follows: （8）； in, is the complete field control variable, is the unbalanced part of various control variables, represents the linear regression coefficient between the two control variables; (45) Based on the complete field or non-equilibrium part of various small-scale variables, vertical transformation operators and horizontal transformation operators are applied to diagnose and calculate the vertical error eigenvalue and horizontal error eigenvalue of each variable; (5) Read the background error covariance and its diagnostic value, satellite observation information, satellite observation error information and model background field, and input them into the variational assimilation system; (6) Obtain the model initial field with the best statistical significance.

2. The method for full-space satellite data assimilation based on weather scale and weather background dependence according to claim 1, characterized in that: Step (2) includes the following steps: (21) According to the variable transformation operator in the physical transformation, the mode variables in the covariance matrix are converted into state variables, which include conventional variables: temperature, pressure, humidity, wind, and hydrometeor variables: cloud water, cloud ice, rain, snow, and graupel; (22) According to the variable scale classification judgment standard of formula (2), each state variable is classified into large-scale continuous variables and small-scale discontinuous variables, and the large-scale classification operator is constructed based on the sample set divided by the variable scale. and small-scale classification operators ; Among them, large-scale continuous variables include: temperature T, air pressure , pseudo relative humidity , zonal wind U, meridional wind V; small-scale discontinuity variables include: cloud water , Yunbing ,rainwater ,Snow , and graupel ; (23) Based on large-scale classification operators and small-scale classification operators , construct the background error covariance matrix of adaptive partitions of different variable scales , the formula is as follows: (3)； in, represents the background error covariance matrix of the large-scale weather area, Represents the background error covariance matrix of the small-scale weather area.

3. The method for full-space satellite data assimilation based on weather scale and weather background dependence according to claim 1 is characterized in that: Step (6) is as follows: In the variational assimilation system, based on the hydrometeor information calculated from the model background field, a satellite radiance forward simulation operator, adjoint operator and tangent operator containing the hydrometeor information are constructed, and assimilation iterative analysis is performed to generate an analysis increment containing the hydrometeor information.

4. A system for the full-space assimilation method of satellite data based on weather scale and weather background dependence according to any one of claims 1 to 3, characterized in that: include: Scale classification operator module: used to read the background error covariance matrix containing N sample member information, calculate and define the scale classification operator; Partition background error covariance module: used to read the background error covariance matrix containing N sample member information again, and construct the variable scale-dependent adaptive partition background error covariance; Large-scale variable module: used to calculate the background error covariance based on large-scale variables; Small-scale variable module: used to calculate the background error covariance based on small-scale variables; Reading module: used to read the background error covariance and its diagnostic quantity, satellite observation information, satellite observation error information and model background field, and input them into the variational assimilation system; Optimal model initial field module: used to obtain the statistically optimal model initial field.

5. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the computer program is loaded into a processor, a full-space assimilation method of satellite data based on weather scale and weather background dependence is implemented according to any one of claims 1 to 3.

6. A storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, a full-space assimilation method of satellite data based on weather scale and weather background dependence according to any one of claims 1 to 3 is implemented.

Images