A ensemble normalized mean field relocation method based on radar combination reflectivity

Abstract

The present invention provides a repositioning method for an ensemble assimilated mean field based on radar combination reflectivity. The method continuously improves the ensemble mean field based on the observed radar combination reflectivity, continuously corrects the model forecast initial field, reduces error accumulation, and maintains high accuracy of the forecast results over a long period of time. The method is based on the similarity between the model forecast data and the observed data, obtains the power spectrum of the combined reflectivity through transformation, and establishes a connection between the two using a cross-spectrum. The spatial proximity of the two is determined by the phase difference between the two. The smaller the phase difference, the closer the power spectrum position, and the more stable and similar the two are. The method finds the ensemble member closest to the observed data in the numerical model forecast, that is, the ensemble member with the smallest spatial distance to the observed data is used as the optimal member. The optimal member is used to replace the ensemble mean, and the ensemble member is updated and used as the initial field of the numerical model forecast, thereby improving the accuracy of the ensemble forecast.

Description

Technical Field

[0001] The invention relates to a repositioning method based on an integrated assimilated mean field of radar combined reflectivity, and belongs to the technical field of numerical model prediction in earth science. Background Art

[0002] Ensemble assimilation is a key method for assimilating numerical model forecast data. It has been theoretically proven that if both numerical model forecast and observation errors conform to a normal distribution, the ensemble's average field is optimal among all ensemble members. However, because numerical model forecast errors are difficult to accurately estimate and observation errors do not always conform to a normal distribution, in some cases the average field of the ensemble assimilation is inferior to that of a particular ensemble member. The purpose of ensemble averaging is to combine information from multiple forecast members to provide a more representative and accurate forecast. If the ensemble average is biased, the final forecast value will deviate significantly from actual weather conditions.

[0003] In numerical model forecasts, the background field provides initial meteorological information for ensemble members. Ensemble members are generated from the background field using specific perturbation methods to form multiple forecast states with varying degrees of variance. These states are used in ensemble forecasts to estimate the probability density function and forecast uncertainty of the atmospheric state. The ensemble mean is calculated by averaging the corresponding meteorological elements within each ensemble to obtain the ensemble mean at each time step. This can be calculated using either simple or weighted averaging.

[0004] For areas where the background field does not match the observations, for example, if the observations show that there is convergence of warm and humid air in a certain area, and the water vapor and vertical motion conditions in this area in the background field are weak, the assimilation process will adjust the background field, increase the water vapor content and vertical upward motion in this area, and create favorable conditions for the generation of convection.

[0005] Due to imperfections in initial fields and physical processes, the generation and development of convection predicted by numerical models often exhibit bias. The data assimilation process generates convection in new areas based on observational data. When convection is suppressed in spurious areas, data assimilation requires assimilating observational data at a certain frequency (e.g., hourly) to gradually generate convection in new areas. This process may require the continuous assimilation of data for several hours to achieve results, which is time-consuming. This means that the forecast fields require several assimilation hours to accurately resolve the actual observations, seriously affecting the effectiveness of numerical model forecasts. Summary of the Invention

[0006] In view of the problems existing in the prior art, the present invention provides a repositioning method and system based on the integrated normalized mean field of radar combined reflectivity.

[0007] In order to solve the above problems, the specific technical solutions proposed by the present invention are as follows:

[0008] A relocation method based on an integrated homogenized mean field of radar combined reflectivity, comprising:

[0009] Obtaining observation data of radar reflectivity factor, converting it into pattern space coordinates through spatial transformation and obtaining observation combined reflectivity of radar reflectivity factor;

[0010] Based on the background field data predicted by the numerical model, the ensemble mean and the model combination reflectivity of the ensemble members are calculated; the background field data are matched with the observation data in time and space; the grid point coordinates in the background field data correspond to the grid point coordinates converted into the model space through spatial transformation.

[0011] According to the ensemble mean and the model combination reflectivity of the ensemble members and the observation combination reflectivity, and according to the similarity between the numerical model forecast and the observation data, the ensemble member that is most similar to the observation data is selected as the optimal member;

[0012] The ensemble mean is replaced by the optimal member, and the ensemble members are updated. This is used as the initial field for numerical model forecasts, and ensemble forecasts are carried out.

[0013] The present invention further includes:

[0014] calculating the fairness skill scores of the ensemble average and the ensemble members based on the ensemble average and the pattern combination reflectances of the ensemble members;

[0015] Select the set member with the highest fairness skill score as the optimal member;

[0016] Calculate the Bias score of the best member selected by the fair skill score, and calculate the Bias score of the best member selected based on the similarity between the numerical model forecast and the observed data. Select the two best members with Bias scores closer to 1 as the actual best members, replace the ensemble mean with the actual best member, and form a new ensemble mean and ensemble members.

[0017] The present invention further comprises selecting the ensemble member most similar to the observed data as the optimal member based on the ensemble average and the model combination reflectivity of the ensemble members and the observed combination reflectivity, and according to the similarity between the numerical model forecast and the observed data, specifically:

[0018] S31. Perform Fourier transform on the ensemble mean and the mode combination reflectivity and the observation combination reflectivity of the ensemble members to obtain the power spectrum distribution function. The calculation formula is as follows:

[0019]

[0020] Where, F O(u,v) represents the power spectrum distribution function of the observed combined reflectivity; F f (u, v) represents the power spectrum distribution function of the mode combination reflectivity; M and N represent the number of grid points in the x and y directions, respectively; u and v represent the amplitude in the x and y directions of the Fourier expansion, respectively; CR f (x,y) and CR o x and y in (x,y) are equivalent to i and j in grid point (i,j) respectively;

[0021] S32. Calculate the cross-spectrum of the observed combined reflectivity and the simulated combined reflectivity to obtain the phase difference between the ensemble average and the model combined reflectivity of the ensemble members and the observed combined reflectivity. The calculation formula is as follows:

[0022] G(u,v)=F O (u,v)*F f (u,v)

[0023]

[0024] Where G(u,v) represents the cross spectrum of the observed combination reflectivity and the mode combination reflectivity; f(x,y) represents the phase difference between the mode combination reflectivity and the observed combination reflectivity; G(u,v) is the inverse Fourier transform of the spectrum distribution;

[0025] S33, respectively calculate the similarity between the ensemble mean and the ensemble members’ model combination reflectivity and the observed combination reflectivity, i.e., the spatial distance D i Minimum, calculated as follows:

[0026]

[0027] Where i represents the set member number; f i (x,y) is the grid point of f(x,y) arranged in descending order of amplitude, f0(x,y) is the maximum value, and so on; u 总 represents the sum of the x-direction amplitudes of the Fourier expansion of all grid points; and v 总 represents the sum of the y-direction amplitudes of the Fourier expansion of all grid points;

[0028] The present invention further provides the following calculation formulas for selecting the best member based on the fairness skill score and the actual best member based on the bias score:

[0029] The fairness skill score of the ensemble mean and the combined reflectivity of the ensemble members is calculated as follows:

[0030] ETS i =(NA-R(a)) / (NA+NB+NC-R(a))

[0031]

[0032] Where, ETS i represents the fairness skill score of the i-th set member; R(a) represents the number of correct predictions that may be made under random circumstances; NA represents the number of correct hits; NB represents the number of false positives; NC represents the number of missed negatives; and ND represents the number of correct rejections.

[0033] The Bias score of the best member is calculated as follows:

[0034]

[0035] The present invention further provides that the observation data for obtaining the radar reflectivity factor is converted into pattern space coordinates through spatial transformation and the observed combined reflectivity of the radar reflectivity factor is obtained, specifically:

[0036] S11, read the observed radar reflectivity factor, and use the spatial transformation to transform the radar reflectivity factor Z of the observed space o (r, α, ψ) is converted to the radar reflectivity factor Z in the pattern space o (i,j,k); Z o (r, α, ψ) where r represents the reservoir distance, α represents the azimuth angle, and ψ represents the polar angle; Z o In (i, j, k), i represents the coordinate in the x direction, j represents the coordinate in the y direction, and k represents the coordinate in the z direction;

[0037] S12. Calculate the maximum vertical value of the combined reflectivity factor of the pattern space grid point (i, j) to obtain the observed combined reflectivity CR of the grid point (i, j) o .

[0038] The present invention further comprises calculating the ensemble mean and the reflectivity of the mode combination of the ensemble members based on the background field data predicted by the numerical model, specifically:

[0039] S21. Read the background field data predicted by the numerical model and calculate the model reflectivity factor Z of the model space according to the empirical formula f (i,j,k);

[0040] S22. Calculate the maximum value of the pattern reflectivity factor at the grid point (i, j) in the vertical direction and obtain the pattern combination reflectivity CR of the grid point (i, j) f .

[0041] The present invention further, the said more set members are specifically:

[0042] S51. Calculate the average value of all meteorological elements at each grid point to obtain the ensemble mean field. The calculation formula is as follows:

[0043]

[0044] Where x i Here, i represents the number of the set member, x represents the meteorological element, which is temperature, air pressure, wind field, altitude or water vapor content; n represents the total number of set members;

[0045] S52. Calculate the difference between the ensemble members and the ensemble mean field for all meteorological elements at each grid point to obtain the disturbance field. The calculation formula is as follows:

[0046]

[0047] S53. Re-assign values to all meteorological elements at all grid points. The calculation formula is as follows:

[0048]

[0049] Where x M The meteorological element value for selecting the best member.

[0050] A relocation system based on the collective normalized mean field of radar combination reflectivity, comprising a pre-processing module, an optimal member selection module, and a collective member update module;

[0051] A preprocessing module is configured to obtain observation data of the radar reflectivity factor, convert it into pattern space coordinates through spatial transformation, and obtain the observation combination reflectivity of the radar reflectivity factor; and calculate the ensemble mean and the pattern combination reflectivity of the ensemble members based on the background field data predicted by the numerical model; the background field data is temporally and spatially matched with the observation data;

[0052] An optimal member selection module selects the ensemble member most similar to the observed data as the optimal member based on the ensemble average and the model combination reflectivity of the ensemble members and the observed combination reflectivity, and based on the similarity between the numerical model forecast and the observed data;

[0053] The set member update module replaces the set average with the optimal member and updates the set members.

[0054] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor is used to call and run the computer program stored in the memory to execute the above method.

[0055] A computer-readable storage medium stores a computer program, wherein the computer program implements the steps of the above method when executed by a processor.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] The present invention provides a relocation method based on the ensemble assimilated mean field of radar combination reflectivity. The method continuously improves the ensemble mean field according to the observed radar combination reflectivity, continuously corrects the initial field of the model forecast, reduces the accumulation of errors, and enables the forecast results to maintain a high accuracy over a longer time range.

[0058] The method of the present invention is based on the similarity between model forecast data and observation data. The power spectrum of the combined reflectivity is obtained through transformation, and the connection between the two is established using a cross spectrum. The degree of spatial proximity of the two positions is judged by the phase difference between the two. The smaller the phase difference, the closer the power spectrum positions are, and the more stable and similar the two are. The ensemble member closest to the observation data in the numerical model forecast is found, that is, the ensemble member with the smallest spatial distance to the observation data is taken as the optimal member, the optimal member is used to replace the ensemble average, and the ensemble member is updated as the initial field of the numerical model forecast, thereby improving the accuracy of the ensemble forecast.

[0059] The present invention also uses a public skill score (ETS) to select the optimal ensemble members for numerical model forecasts. The optimal ensemble members selected by the public skill score (ETS) are then subjected to a bias score. The bias scores are compared with the bias scores of the optimal members selected based on the similarity between the numerical model forecast and the observed data. The members with a bias score closer to 1 are selected as the actual optimal ensemble members. The ensemble mean is replaced by the selected actual optimal ensemble members, and the new ensemble members are used as the initial fields for the numerical model forecast, thereby improving forecast accuracy.

[0060] This method continuously improves the ensemble mean field based on the observed radar combination reflectivity, constantly corrects the initial field predicted by the numerical model, reduces the accumulation of errors, and ensures that the forecast results maintain a high degree of accuracy over a longer time range.

[0061] This method combines observation data with different error characteristics with numerical model forecasts in the high-frequency ensemble assimilation of numerical model forecasts, incorporates new observation data into numerical model forecasts, and allows the model to capture the latest changes in the atmosphere in a timely manner, thereby providing more accurate forecasts. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for describing the embodiments or the prior art.

[0063] Figure 1 Schematic diagram of the flow of a relocation method based on the integrated normalized mean field of radar combined reflectivity in an embodiment; DETAILED DESCRIPTION

[0064] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0065] Example 1:

[0066] The present invention is based on a relocation method using a collective normalized mean field of radar combined reflectivity, wherein relocation refers to replacing the original mean field with a new mean field. The method comprises the following steps:

[0067] Obtaining observation data of radar reflectivity factor, converting it into pattern space coordinates through spatial transformation and obtaining observation combined reflectivity of radar reflectivity factor;

[0068] Based on the background field data predicted by the numerical model, the ensemble mean and the model combination reflectivity of the ensemble members are calculated; the background field data are matched with the observation data in time and space; the grid point coordinates in the background field data correspond to the grid point coordinates converted into the model space through spatial transformation.

[0069] According to the ensemble mean and the model combination reflectivity of the ensemble members and the observation combination reflectivity, and according to the similarity between the numerical model forecast and the observation data, the ensemble member that is most similar to the observation data is selected as the optimal member;

[0070] The ensemble mean is replaced by the optimal member, and the ensemble members are updated. This is used as the initial field for numerical model forecasts, and ensemble forecasts are carried out.

[0071] Example 2:

[0072] This embodiment is further designed based on the first embodiment in that, according to the ensemble average and the model combination reflectivity of the ensemble members and the observed combination reflectivity, and according to the similarity between the numerical model forecast and the observed data, the ensemble member that is most similar to the observed data is selected as the optimal member, specifically:

[0073] S31. Perform Fourier transform on the ensemble mean and the mode combination reflectivity and the observation combination reflectivity of the ensemble members to obtain the power spectrum distribution function. The calculation formula is as follows:

[0074]

[0075] Where, F O (u,v) represents the power spectrum distribution function of the observed combined reflectivity; F f (u, v) represents the power spectrum distribution function of the mode combination reflectivity; M and N represent the number of grid points in the x-direction and y-direction, respectively; u and v represent the vibration pairs in the x-direction and y-direction of the Fourier expansion, respectively; CR f (x,y) and CRo x and y in (x,y) are equivalent to i and j in grid point (i,j) respectively;

[0076] S32. Calculate the cross-spectrum of the observed combined reflectivity and the simulated combined reflectivity to obtain the phase difference between the ensemble average and the model combined reflectivity of the ensemble members and the observed combined reflectivity. The calculation formula is as follows:

[0077] G(u,v)=F O (u,v)*F f (u,v)

[0078]

[0079] Where G(u,v) represents the cross spectrum of the observed combination reflectivity and the mode combination reflectivity; f(x,y) represents the phase difference between the mode combination reflectivity and the observed combination reflectivity; G(u,v) is the inverse Fourier transform of the spectrum distribution;

[0080] S33, respectively calculate the similarity between the ensemble mean and the ensemble members’ model combination reflectivity and the observed combination reflectivity, i.e., the spatial distance D i Minimum, calculated as follows:

[0081]

[0082] Where i represents the set member number; f i (x,y) is the grid point of f(x,y) arranged in descending order of amplitude, f0(x,y) is the maximum value, and so on; u 总 represents the sum of the x-direction vibrations of the Fourier expansion of all grid points; and v 总 represents the sum of the y-direction vibration pairs of the Fourier expansion of all grid points; in this example, the cumulative values of f(x,y) of the top 20 are taken respectively. Since the top 20 have more than 90% of the spectral energy of the set members, the specific details are as follows:

[0083]

[0084]

[0085] Example 3:

[0086] This embodiment is further designed on the basis of the second embodiment in that the relocation method based on the collective normalized mean field of the radar combination reflectivity in this embodiment also includes:

[0087] Calculating the fair skill scores (ETS) of the ensemble average and the ensemble members based on the ensemble average and the mode combination reflectances of the ensemble members;

[0088] Select the set member with the highest fairness skill score ETS as the optimal member;

[0089] Calculate the Bias score of the best member selected by the fair skill score ETS, and calculate the Bias score of the best member selected based on the similarity between the numerical model forecast and the observed data. Select the two best members with Bias scores closer to 1 as the actual best members, and replace the ensemble average with the actual best members.

[0090] Example 4:

[0091] This embodiment is further designed based on the third embodiment in that the calculation formula for selecting the best member based on the fairness skill score and the actual best member based on the bias score is as follows:

[0092] The fairness skill score of the ensemble mean and the combined reflectivity of the ensemble members is calculated as follows:

[0093] ETS i =(NA-R(a)) / (NA+NB+NC-R(a))

[0094]

[0095] Where, ETS i represents the fairness skill score of the i-th set member; R(a) represents the number of correct predictions that may be made under random circumstances; NA represents the number of correct hits; NB represents the number of false positives; NC represents the number of missed negatives; and ND represents the number of correct rejections.

[0096] The Bias score of the best member is calculated as follows:

[0097]

[0098] Embodiment 5:

[0099] This embodiment is further designed based on the fourth embodiment in that, in this embodiment, the observation data of the radar reflectivity factor is obtained, converted into pattern space coordinates through spatial transformation, and the observed combined reflectivity of the radar reflectivity factor is obtained, specifically:

[0100] S11, read the observed radar reflectivity factor, and use the spatial transformation to transform the radar reflectivity factor Z of the observed space o (r, α, ψ) is converted to the radar reflectivity factor Z in the pattern space o (i,j,k); Z o (r, α, ψ) where r represents the reservoir distance, α represents the azimuth angle, and ψ represents the polar angle; Z o In (i, j, k), i represents the coordinate in the x direction, j represents the coordinate in the y direction, and k represents the coordinate in the z direction;

[0101] S12. Calculate the maximum vertical value of the combined reflectivity factor of the pattern space grid point (i, j) to obtain the observed combined reflectivity CR of the grid point (i, j) o (i,j).

[0102] Based on the background field data predicted by the numerical model, the ensemble mean and the model combination reflectivity of the ensemble members are calculated, specifically:

[0103] S21, read the background field of the numerical model forecast of each hydrometeor mixing ratio (rain mixing ratio qr, snow mixing ratio qs and graupel mixing ratio qg), temperature T b , air density ρ, calculate the reflectivity factor Zf(i,j,k) of each grid point pattern according to formulas 1 to 3;

[0104] S22. Calculate the maximum value of the pattern reflectivity factor at the grid point (i, j) in the vertical direction and obtain the pattern combination reflectivity CR of the grid point (i, j) f :

[0105] Z f =10lgZ e (1)

[0106] Z e =Z r +Z ds +Z ws +Z g (2)

[0107]

[0108] Among them, x refers to rain r, dry snow ds, wet snow ws, and graupel g, and α x is the coefficient of each hydrometeor determined by the dielectric constant, density and intercept parameters. The coefficient of rainwater is 3.63×10 9 For snow and graupel, the wet snow coefficient is 4.26×10 11 , the dry snow coefficient is 9.80×10 8 The hail coefficient is 1.09×10 9 .

[0109] Example 6:

[0110] This embodiment is further designed based on the fifth embodiment. In this embodiment, after replacing the set average with the optimal member, the specific process of updating the set members is as follows:

[0111] S51. Calculate the average value of all meteorological elements at each grid point to obtain the ensemble mean field. The calculation formula is as follows:

[0112]

[0113] Where x i Here, i represents the number of the set member, x represents the meteorological element, which is temperature, air pressure, wind field, altitude or water vapor content; n represents the total number of set members;

[0114] S52. Calculate the difference between the ensemble members and the ensemble mean field for all meteorological elements at each grid point to obtain the disturbance field. The calculation formula is as follows:

[0115]

[0116] S53. Re-assign values to all meteorological elements at all grid points. The calculation formula is as follows:

[0117]

[0118] Where x M The meteorological element value for selecting the best member.

[0119] Embodiment seven:

[0120] The present invention provides a relocation system based on the collective normalized mean field of radar combined reflectivity, comprising a pre-processing module, an optimal member selection module, and a collective member update module;

[0121] A preprocessing module is configured to obtain observation data of the radar reflectivity factor, convert it into pattern space coordinates through spatial transformation, and obtain the observation combination reflectivity of the radar reflectivity factor; and calculate the ensemble mean and the pattern combination reflectivity of the ensemble members based on the background field data predicted by the numerical model; the background field data is temporally and spatially matched with the observation data;

[0122] An optimal member selection module selects the ensemble member most similar to the observed data as the optimal member based on the ensemble average and the model combination reflectivity of the ensemble members and the observed combination reflectivity, and based on the similarity between the numerical model forecast and the observed data;

[0123] The set member update module replaces the set average with the optimal member and updates the set members.

[0124] Embodiment 8:

[0125] An electronic device of the present invention includes a memory and a processor, wherein the memory stores a computer program, and the processor is configured to call and run the computer program stored in the memory to execute the method described in any one of the above embodiments.

[0126] A computer-readable storage medium of the present invention stores a computer program, which implements the steps of the method described in any of the above embodiments when executed by a processor.

Claims

1. A relocation method based on the collective normalized mean field of radar combined reflectivity, characterized in that: include: Obtaining observation data of radar reflectivity factor, converting it into pattern space coordinates through spatial transformation and obtaining observation combined reflectivity of radar reflectivity factor; Calculating the ensemble mean and the model combination reflectivity of the ensemble members based on the background field data predicted by the numerical model; the background field data is temporally and spatially matched with the observation data; According to the ensemble mean and the model combination reflectivity of the ensemble members and the observation combination reflectivity, and according to the similarity between the numerical model forecast and the observation data, the ensemble member that is most similar to the observation data is selected as the optimal member; Replace the set average with the optimal member and update the set members; The method of selecting the ensemble member most similar to the observed data as the optimal member based on the ensemble mean and the model combined reflectivity of the ensemble members and the observed combined reflectivity, and based on the similarity between the numerical model forecast and the observed data, is specifically as follows: S31. Perform Fourier transform on the ensemble mean and the mode combination reflectivity and the observation combination reflectivity of the ensemble members to obtain the power spectrum distribution function. The calculation formula is as follows: Where, F O (u,v) represents the power spectrum distribution function of the observed combined reflectivity; F f (u, v) represents the power spectrum distribution function of the mode combination reflectivity; M and N represent the number of grid points in the x and y directions, respectively; u and v represent the amplitudes in the x and y directions of the Fourier expansion, respectively; S32. Calculate the cross-spectrum of the observed combined reflectivity and the simulated combined reflectivity to obtain the phase difference between the ensemble average and the model combined reflectivity of the ensemble members and the observed combined reflectivity. The calculation formula is as follows: G(u,v)=F O (u,v)*F f (u,v) Where G(u,v) represents the cross spectrum of the observed combination reflectivity and the model combination reflectivity; f(x,y) represents the phase difference between the model combination reflectivity and the observed combination reflectivity; S33, respectively calculate the similarity between the ensemble mean and the ensemble members’ model combination reflectivity and the observed combination reflectivity, i.e., the spatial distance D i Minimum, calculated as follows: Where i represents the set member number; f i (x,y) is the grid point of f(x,y) arranged in descending order of amplitude, f0(x,y) is the maximum value, and so on; u 总 represents the sum of the x-direction amplitudes of the Fourier expansion of all grid points; v 总 Represents the sum of the y-direction amplitudes of the Fourier expansion of all grid points.

2. The relocation method based on the integrated normalized mean field of radar combined reflectivity according to claim 1, characterized in that: Also includes, Calculating the fairness skill scores of the ensemble average and the ensemble members based on the ensemble average and the mode combination reflectances of the ensemble members; Select the set member with the highest fairness skill score as the optimal member; Calculate the Bias score of the best member selected by fair skill scoring, and calculate the Bias score of the best member selected based on the similarity between the numerical model forecast and the observed data. Select the two best members with Bias scores closer to 1 as the actual best members, and replace the ensemble average with the actual best members.

3. The relocation method based on the integrated normalized mean field of radar combined reflectivity according to claim 2, characterized in that: The calculation formula for selecting the best member based on the fairness skill score and the actual best member based on the bias score is as follows: The fairness skill score of the ensemble mean and the combined reflectivity of the ensemble members is calculated as follows: ETS i =(NA-R(a)) / (NA+NB+NC-R(a)) Where, ETS i represents the fairness skill score of the i-th set member; R(a) represents the number of correct predictions that may be made under random conditions; NA represents the number of correct hits; NB represents the number of false positives; NC represents the number of missed negatives; ND represents the number of correct rejections; The Bias score of the best member is calculated as follows:

4. The relocation method based on the integrated normalized mean field of radar combined reflectivity according to claim 3 is characterized in that: The observation data of the radar reflectivity factor is converted into pattern space coordinates through spatial transformation and the observed combined reflectivity of the radar reflectivity factor is obtained, specifically: S11, read the observed radar reflectivity factor, and use the spatial transformation to transform the radar reflectivity factor Z of the observed space o (r, α, ψ) is converted to the radar reflectivity factor Z in the pattern space o (i,j,k); Z o (r, α, ψ) where r represents the reservoir distance, α represents the azimuth angle, and ψ represents the polar angle; Z o In (i, j, k), i represents the coordinate in the x direction, j represents the coordinate in the y direction, and k represents the coordinate in the z direction; S12. Calculate the maximum value of the combined reflectivity factor of the pattern space grid point (i, j) in the vertical direction to obtain the observed combined reflectivity CR of the grid point (i, j) o .

5. The relocation method based on the integrated normalized mean field of radar combined reflectivity according to claim 4 is characterized in that: The calculation of the ensemble mean and the model combination reflectivity of the ensemble members based on the background field data predicted by the numerical model is specifically as follows: S21. Read the background field data predicted by the numerical model and calculate the model reflectivity factor Z of the model space according to the empirical formula f (i,j,k); S22. Calculate the maximum value of the pattern reflectivity factor at the grid point (i, j) in the vertical direction and obtain the pattern combination reflectivity CR of the grid point (i, j) f .

6. The relocation method based on the integrated normalized mean field of radar combined reflectivity according to claim 1, characterized in that: The update set members are specifically: S51. Calculate the average value of all meteorological elements at each grid point to obtain the ensemble mean field. The calculation formula is as follows: Where x i Here, i represents the number of the set member, x represents the meteorological element, which is temperature, air pressure, wind field, altitude or water vapor content; n represents the total number of set members; S52. Calculate the difference between the ensemble members and the ensemble mean field for all meteorological elements at each grid point to obtain the disturbance field. The calculation formula is as follows: S53. Re-assign values to all meteorological elements at all grid points. The calculation formula is as follows: Where x M The meteorological element value for selecting the best member.

7. A system using the relocation method based on the integrated normalized mean field of radar combined reflectivity according to any one of claims 1 to 6, characterized in that: It includes pre-processing module, optimal member selection module and set member update module; A preprocessing module is configured to obtain observation data of the radar reflectivity factor, convert it into pattern space coordinates through spatial transformation, and obtain the observation combination reflectivity of the radar reflectivity factor; and calculate the ensemble mean and the pattern combination reflectivity of the ensemble members based on the background field data predicted by the numerical model; the background field data is temporally and spatially matched with the observation data; An optimal member selection module selects the ensemble member most similar to the observed data as the optimal member based on the ensemble average and the model combination reflectivity of the ensemble members and the observed combination reflectivity, and based on the similarity between the numerical model forecast and the observed data; The set member update module replaces the set average with the optimal member and updates the set members.

8. An electronic device, characterized in that: The electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor is configured to call and run the computer program stored in the memory to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images