Method for improving sand and dust numerical forecasting through assimilation-aerosol-cloud-radiation

Abstract

The invention discloses a method for improving sand and dust numerical forecasting through assimilation-aerosol-cloud-radiation, and the method comprises the steps: building a sand and dust storm forecasting mode, calculating sand and dust optical parameters based on a spherical particle extinction theory, and achieving the bidirectional feedback of sand and dust to weather; a sand and dust aerosol-cloud interaction ice nucleus nucleation mechanism is given, a sand and dust CCN type aerosol homogeneous freezing process is realized, a radiation variable temperature rate is calculated and fed back to a weather mode power process, and bidirectional feedback of weather-driven sand and dust and influence of sand and dust on weather is formed; constructing an ensemble forecasting module, performing time and space related disturbance on a meteorological initial field, a boundary condition and an aerosol initial concentration, and performing ensemble forecasting after a mode is input; and establishing an ensemble assimilation analysis correction module, constructing a background error covariance matrix by utilizing mode ensemble forecasting, disturbing observation at an analysis moment, and assimilating and correcting the initial concentration of the mode aerosol based on a localized ensemble Kalman filtering method to obtain a forecasting result.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of atmospheric coupling, in particular to a weather-sand-dust two-way feedback numerical model and a combined assimilation prediction method and system. BACKGROUND

[0002] Sandstorm is a global disastrous weather, and Asia is the second largest sand source area in the world. Sand weather can be transported to a very far downstream area after the occurrence of sand source, and has a wide range of influence. The sand-dust numerical prediction model is driven by a series of sand-dust aerosol processes such as weather model sanding, settling, etc., and is an important tool for studying and numerically predicting sandstorms. It is playing an increasingly important role in studying the generation mechanism, transportation and prediction and early warning of sand weather. However, due to the initial and boundary conditions, nonlinearity, sub-grid parameterization and other problems, there are still some errors and uncertainties in the simulation and prediction of numerical model. On the other hand, during the real-time prediction of the sand-dust model, it takes a process to adjust the meteorological field from zero field, and the setting method of the initial concentration of sand-dust at the starting time has an important influence on short-term prediction. We can obtain monitoring data of atmospheric composition, sand weather phenomenon, etc. These observation data can more objectively record the real situation of the atmosphere, but due to the irregular and small number of space-time distribution of the stations, they cannot provide information on each grid point in the area of interest, the average state and are difficult to predict the future situation. By using assimilation technology, we can integrate the prediction results of the model and the observation data and their respective error distribution characteristics to obtain the optimal gridded analysis results, provide more accurate initial field for the model prediction, and improve the prediction level. Once the sand-dust weather occurs, the sand-dust aerosol can exist from the ground to 9 kilometers in height, and has an important influence on solar radiation and cloud physics process. In the model, the influence of the sand-dust aerosol is calculated in detail, and the numerical prediction effect of the sand-dust is also improved. This two-way feedback mechanism between weather and sand-dust and the joint effect of assimilation are more important for improving the prediction effect of sand-dust. SUMMARY

[0003] The purpose of the present application is to provide a weather-sand-dust two-way feedback numerical model and a combined assimilation prediction method.

[0004] In order to achieve the above purpose, the present application is implemented according to the following technical scheme: The present application comprises the following steps: Based on the regional weather prediction model and the online driving of the sanding, transportation and settling of sand-dust of multiple physical mechanisms, a 168-hour prediction model of sandstorm is established, the optical parameters are calculated based on the extinction theory of spherical particles, and the two-way feedback of sand-dust to weather is realized. The optical parameters include optical parameter mass extinction coefficient, sand-dust comprehensive extinction coefficient, sand-dust comprehensive single scattering ratio and sand-dust comprehensive asymmetry factor. Given the ice nucleation mechanism of dust aerosol-cloud interaction, the real-time homogeneous freezing process of dust CCN type aerosol, the optical parameters of the dust are input into the radiation model, the radiation temperature change rate feedback to the dynamic process of the weather model is calculated, and the two-way feedback of weather driving dust and dust affecting weather is formed. Collection forecast module: considering the time and space disturbance of the meteorological initial field and boundary conditions and the initial field of dust aerosol input into the model, the collection forecast is carried out after inputting the model; Collection assimilation analysis revision module: selecting collection samples and error analysis, assimilation analysis and initial field update, using the collection forecast results of the model to construct the background error covariance matrix, disturbing the observation at the analysis time and before, based on the localized collection Kalman filter method, the dust initial field of the model is assimilated and revised to obtain more accurate dust initial field and its influence on the weather.

[0005] Further, the method for establishing a 168-hour dust storm forecast model comprises: Realize the external mixing of N particles of dust in any way, and in Waveband calculation of optical parameters and two-way feedback function: The extinction efficiency, single scattering albedo and asymmetry factor of the nth particle are obtained, according to the spherical particle extinction calculation theory, the optical parameter mass extinction coefficient of dust aerosol under different wavebands and different particle sizes is formed by using the external mixing method for the forecast dust concentration of any N particle diameter of the model, and the expression is: Wherein the dust density is , the dust radius is , the extinction efficiency of the nth particle in waveband is , and the optical parameter mass extinction coefficient is ; The comprehensive extinction coefficient of dust is the sum of the extinction of all N particles, and the expression is: Wherein the comprehensive extinction coefficient of dust is , and ; The comprehensive single scattering ratio of dust is obtained according to the weighted average of the comprehensive extinction coefficient of all particle sizes, and the expression is: Wherein the comprehensive single scattering ratio of dust in waveband is , the single scattering albedo in waveband is , and m is The comprehensive single scattering ratio of the dust is calculated by weighting average of the scattering ratios of the dust with different particle sizes, and the expression is as follows: The comprehensive asymmetric factor of the dust is calculated by weighting average of the comprehensive single scattering ratios of the dust, and the expression is as follows: The comprehensive asymmetric factor of the dust in the wave band is calculated by weighting average of the comprehensive single scattering ratios of the dust, and the expression is as follows: ; The optical parameters quality extinction coefficient, comprehensive extinction coefficient of the dust, comprehensive single scattering ratio and comprehensive asymmetric factor of the dust are inputted into the fixed value in the replacement model in the integral process of the model to replace the model; The radiation transfer model calculates the radiation temperature rate to complete the real-time calculation of the radiation effect of the dust; wherein the radiation temperature rate includes the real-time feedback of the radiation temperature rate affected by the dust to the dynamic process of the model.

[0006] Further, the method for the ice nucleus nucleation of the dust aerosol-cloud interaction mechanism comprises: The ice crystal number concentration generated by the nucleation of the dust aerosol is calculated: The ice crystal number concentration generated by the nucleation of the dust aerosol is calculated as follows: The temperature of the environment atmosphere is T, the unit is K, the number concentration of the SD aerosol with the radius greater than 0.5 m is , the unit is kg -1 , and the constants are a, b and c, respectively, a, b and c are 0.0000594, 3.33 and 0.0033, respectively; The number concentration of the SD aerosol with the radius greater than 0.5 m is calculated: The mass of the SD aerosol particle is m, the unit is kg, the tracer is , the tracer is the mass mixing ratio of the SD aerosol, the number of the tracer SD is num, the average radius is r, the unit is mm, the density is r, the unit is g / cm 3 , the number concentration of the SD aerosol is , the unit is kg -1 , and i, k and j are the positions of the grid points of the model; Then, the calculated Ndust0.5 is substituted into the calculation formula of the ice crystal number concentration generated by the nucleation of the dust aerosol, and the real-time heterogeneous nucleation process of the dust aerosol is completed, so that the real-time ice nucleus nucleation process of the dust-cloud effect is completed.

[0007] Further, the method for the real-time homogeneous freezing process of the dust CCN type aerosol comprises:​ The ice formation process of the dust CCN type aerosol, the total CCN type aerosol number concentration CCN type aerosol homogeneous freezing ice crystal number concentration is calculated: Wherein Nci is the CCN type aerosol homogeneous freezing ice crystal number concentration, unit kg -1 , The homogeneous nucleation rate coefficient is unit cm-3 / s, and the homogeneous nucleation rate coefficient value is mainly related to the ambient atmospheric temperature and the supersaturation, The time step of the model is 100s; NWFA2 is the total CCN type aerosol number concentration, unit kg -1 , mc is the mass of the aerosol, unit kg, and the forecast quantity tracerc is the CCN aerosol mass mixing ratio, unit kg kg -1 , numc is the number of tracerc, rc is the average radius of the aerosol, unit mm, and the density is rc, unit g / cm 3 , i, k, j are the grid point positions of the model.

[0008] Further, the method of the ensemble prediction module comprises: The time and space related perturbations of the meteorological initial field and boundary conditions and the dust aerosol initial field input to the model are considered according to the following formula, Wherein , The variable to be perturbed is i=1,2,...,N, The i-th ensemble sample is The spatial position is The three-dimensional random field satisfying the normal distribution is The three-dimensional random field satisfying the normal distribution at time t and considering the time and space correlation is The standard deviation of the perturbed variable is The parameter representing the size of the time or space correlation is The perturbed meteorological and dust initial concentration of atmospheric composition is input to the weather-dust two-way feedback numerical model for prediction, and the dust ensemble prediction result is obtained.

[0009] Further, the method of the ensemble assimilation system comprises: The matrix Is the ensemble prediction result of the model analysis variable The expression is: wherein is the number of ensemble samples, is the dimension of the model variables; Let is the mean of the matrix, then the anomaly matrix of , the expression is: the model background error covariance matrix , the expression is: Based on the Kalman filter theory, the ensemble Kalman filter method is used for assimilation analysis, and the expression is: wherein is the final solution of the ensemble optimal estimation field, also known as the assimilation analysis field, is the model background field, i.e., the ensemble prediction result, is the observation and its perturbation vector, is the observation operator, which is used to project the model variables into the observation space, is the observation increment, is the analysis increment, is a scalar coefficient for adjusting the relative contribution of the analysis increment and the model background field; is the observation error covariance matrix, is the error covariance matrix of the model background field, which is analyzed and statistically obtained by using the ensemble prediction result in the ensemble assimilation method; Since the atmospheric chemical model includes too many aerosol types and particle sizes, the model analysis variables cannot include all the model variables, and the selected or combined variables are used as the analysis variables for data assimilation to determine the corresponding assimilation analysis field.

[0010] In the second aspect, the ensemble assimilation prediction system based on the weather-sandstorm two-way feedback numerical model comprises: A sandstorm prediction module: based on the regional weather prediction model and the online driving of the sand-raising, transportation and deposition of multiple physical mechanisms, a 168-hour sandstorm prediction model is established, the optical parameters are calculated based on the extinction theory of spherical particles, and the two-way feedback of sand to weather is realized; the optical parameters include extinction coefficient, single scattering ratio and asymmetry factor; Two-way feedback module: for the given sand dust aerosol-cloud interaction mechanism ice nucleus nucleation, real-time sand dust CCN type aerosol homogeneous freezing process, input the optical parameter of the sand dust into the radiation mode, calculate the radiation temperature rate feedback to the weather mode dynamic process, form the two-way feedback of weather driving sand dust and sand dust affecting weather; Collection forecast module: considering the time and space correlation disturbance of the meteorological initial field and boundary condition input by the mode and the initial field of the sand dust aerosol, input the mode to carry out collection forecast; Collection assimilation analysis revision module: carry out selection of collection samples and error analysis, assimilation analysis and initial field update; utilize the collection forecast result of the mode to construct the background error covariance matrix, disturb the observation at the analysis time and before, carry out assimilation revision to the initial field of the mode based on the localized collection Kalman filter method, obtain more accurate sand dust initial field and forecast result and its influence on weather.

[0011] The beneficial effects of the present application are: Compared with the prior art, the present application has the following technical effects: The weather-sand dust two-way feedback mode system of the present application realizes online driving of the sand dust aerosol physicochemical process such as sand dust, transportation, and deposition in the weather mode, carries out sand dust concentration prediction, simultaneously feeds back the sand dust-radiation-cloud interaction to the weather mode based on the predicted sand dust concentration in real time; the collection assimilation system can quickly assimilate the atmospheric composition observation, provide more accurate atmospheric composition initial field for the numerical prediction mode, and reduce the cumulative error of the warm start process of the mode. The collection assimilation joint sand dust-radiation-cloud feedback mechanism to weather can obtain more accurate sand dust prediction result. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 The step flow chart of the weather-sand dust two-way feedback numerical mode and collection assimilation forecast method of the present application is shown in the figure; Figure 2 The mode sand dust aerosol-cloud coupling mechanism in the embodiment of the present application is shown in the figure; Figure 3 The localized collection Kalman filter assimilation system structure diagram of the sand dust mode in the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0013] The present application will be further described below through specific embodiments, and the illustrative embodiments of the present application and the description are used to explain the present application, but not as a limitation of the present application.

[0014] The weather-sand dust two-way feedback numerical mode and collection assimilation forecast method and system of the present application include the following steps: As Figure 1As shown, in the present embodiment, the following steps are included: Based on the regional weather forecast model and the online driving sand, the sand, the transport, the settlement of multiple physical mechanisms of sand, the sandstorm 168 hour forecast model is established, the optical parameters are calculated based on the extinction theory of spherical particles, and the two-way feedback of sand to weather is realized; The optical parameters include optical parameter mass extinction coefficient, sand comprehensive extinction coefficient, sand comprehensive single scattering ratio and sand comprehensive asymmetry factor; Given the ice nucleus nucleation mechanism of dust aerosol-cloud interaction, the real-time homogeneous freezing process of dust CCN type aerosol, the optical parameters of the dust are input into the radiation model, the radiation temperature change rate feedback to the dynamic process of the weather model is calculated, and the two-way feedback of weather driving dust and dust affecting weather is formed; Based on the ensemble forecast module of the weather-dust two-way feedback numerical model, the time and space related perturbations of the meteorological initial field and boundary conditions of the model and the initial concentration of atmospheric aerosols such as dust are considered, and the ensemble forecast is input into the model.

[0015] The assimilation analysis system is constructed, the ensemble sample optimization and the background error matrix construction of the model are carried out, and the local ensemble Kalman filter assimilation analysis is carried out, and the initial field of atmospheric aerosols such as dust is updated; In actual evaluation, first, the observation data quality control and observation error covariance matrix are designed: the observation data are checked for spatial and temporal continuity and threshold, and the abnormal observation and the observation far beyond the range of the model are removed; The observation error covariance matrix R is set as a diagonal matrix, that is, the error of each observation point is not related to other points, and the element on the diagonal is the observation error of each observation point And the representative error ; Set to 7.5% of the observation value, the representative error Is defined as , wherein The model resolution is 10 kilometers; Lr is the representative characteristic scale of observation, which is 2 kilometers for urban sites, 10 kilometers for rural sites, and 20 kilometers for control sites; Secondly, the assimilation analysis variables are constructed: the model aerosol species are black carbon , organic carbon , dust , sea salt , sulfate , nitrate And ammonium salt ; The particle size distribution of the model aerosol particle size N interval, the analysis variables of the model are PM 2.5 And PMC1, the expression is: The observation vector is generated by PM 2.5 and PM 10 -PM 2.5 Concentration observations and their perturbation components; Then, ensemble samples are selected and optimized: the ensemble forecast results of the model forecasts at the analysis time are used as background field information to generate ensemble samples, and the model background error covariance matrix composed of the ensemble samples is optimized. Perform singular value decomposition, the expression is: in Depend on A matrix consisting of all eigenvectors, where each column is an eigenvector. It is a diagonal matrix, and the elements on the diagonal of the diagonal matrix are the corresponding eigenvalues; based on the conditions: 1) the mean of the ensemble sample is close to the observed value; 2) the dispersion of the ensemble sample is comparable to the prediction error, select the eigenvectors corresponding to the N eigenvalues ​​to form a new background error covariance matrix; The next step is to perform localized ensemble Kalman filter assimilation: The observation operator H is calculated to transform the variables in the model space to the observation space, generating observation increments. Localization refers to the fact that during assimilation analysis, a model grid point is influenced by observation points within a certain radius around it. A localization radius L is defined, and for each grid point, only the observation increments within its vicinity of L are considered. For multiple observations within L, a localized weighting system is calculated. Considering the different impacts of observation increments: in This refers to the actual distance from the grid point to the observation station; Before analysis, obtain the effective observations around the grid points, calculate the effective observation increments of the grid points, and return the analyzed variables to the original variable field; localization reduces spurious long-range correlations in the background error covariance matrix and limits the influence of each observation to a certain range of the analysis grid points; The observations are perturbed at the analysis time to obtain... y Vector, calculation mode projected at the observation point observation increment ; Calculate set bias using set samples Then calculate ; Calculate using singular value decomposition The inverse matrix is ​​used to update the mode analysis variables using the Kalman filter formula; Finally, update the initial atmospheric composition field of the model: update the PM... 2.5The incremental analysis of PMC1 concentration is allocated according to the proportion of the aerosol concentration before assimilation. The atmospheric aerosol concentration, including dust, is adjusted accordingly to generate a new initial field file of atmospheric composition for the model. The effect of the assimilation results is verified and evaluated using the correlation coefficient and root mean square error statistic. The updated initial field is then put into the model for the next time period forecast.

[0016] In this embodiment, the method for establishing a 168-hour sandstorm forecast model includes: To achieve N particle levels of sand and dust through external mixing, Band-based optical parameter calculation and two-way feedback function: The extinction efficiency, single-scattering albedo, and asymmetry factor of the nth particle size are obtained. Based on the extinction calculation theory of spherical particles, the mass extinction coefficients of the dust aerosol under different wavelengths and particle sizes are formed by external mixing for N particle sizes of the model-predicted dust concentration. The expression is as follows: The density of sand and dust is The radius of the sandstorm is ,exist The extinction efficiency of the nth band particle is The optical parameter is the mass extinction coefficient. ; The overall extinction coefficient of dust is the sum of the extinctions of all N-level particles, expressed as: The comprehensive extinction coefficient of sand and dust is: ,for ; The comprehensive single scattering ratio of dust particles is obtained by weighting the extinction coefficients of all particle sizes to the comprehensive extinction coefficient. The expression is as follows: Among them The overall single scattering ratio of dust in the band is ,exist The single scattering albedo of particles in band n is ; The comprehensive asymmetry factor of dust is calculated by weighting the single scattering ratio of dust particles with different scattering ratios. The expression is as follows: Where the asymmetric factor is ,exist The comprehensive asymmetry factor of the band dust is ; The optical parameters quality extinction coefficient, dust comprehensive extinction coefficient, dust comprehensive single scattering ratio and dust comprehensive asymmetry factor are replaced by fixed values in the radiation transfer model during the integral process of the model; The radiation transfer model calculates the radiation temperature change rate to complete real-time calculation of the dust radiation effect, wherein the radiation temperature change rate includes the dust-affected radiation temperature change rate, which is fed back to the dynamic process of the model in real time.

[0017] In the embodiment, the ice nucleus nucleation method of the dust aerosol-cloud interaction mechanism comprises: The ice crystal number concentration generated by the dust aerosol nucleation is calculated as follows: The ice crystal number concentration generated by the dust aerosol nucleation is calculated as follows: The temperature of the ambient atmosphere is T, in K, the number concentration of the SD aerosol with a radius greater than 0.5 m is , in kg -1 The constants are a, b and c, and a, b and c are 0.0000594, 3.33 and 0.0033 respectively; The number concentration of the SD aerosol with a radius greater than 0.5 m is calculated as follows: The mass of the SD aerosol particle is m, in kg, the tracer is , the tracer is the SD aerosol mass mixing ratio, the number of the tracer is num, the number of the tracer is numc, the average radius is r, in mm, the density is r, in g / cm 3 , the number concentration of the SD aerosol is , in kg -1 i, k and j are the positions of the grid points; the number of the tracer and related information are shown in Table 1; The number concentration of the SD aerosol with a radius greater than 0.5 m is calculated as follows: In actual evaluation, non-hygroscopic aerosols such as black carbon and dust are ignored for numc, and in the model, only when the temperature of the ambient atmosphere is less than 35.15 DEG C and the supersaturation relative to ice is greater than 0.4, can the CCN type aerosol form ice crystals.

[0018] In the embodiment, the method for the real-time dust CCN type aerosol homogeneous freezing process comprises: The ice formation process of the dust CCN type aerosol, and the ice crystal number concentration of the homogeneous freezing of the CCN type aerosol is calculated: wherein Nci is the ice crystal number concentration of the homogeneous freezing of the CCN type aerosol, unit: kg -1 , the mode NWFA2 calculates the total CCN type aerosol number concentration, unit: kg -1 , is a homogeneous nucleation rate coefficient, unit: cm -3 / s, the value of the homogeneous nucleation rate coefficient is mainly related to the ambient atmospheric temperature and the supersaturation, and dt is the time step of the mode, and the time step is 100s; In the mode NWFA2, the CCN type aerosol can form ice crystals only when the temperature of the ambient atmosphere is less than 35.15℃ and the supersaturation relative to the ice surface is greater than 0.4.

[0019] In the embodiment, the method of the ensemble prediction module comprises: The time and space related perturbation of the meteorological initial field and the boundary condition and the dust aerosol initial field input to the mode is considered according to the following formula, wherein , is a variable that needs to be perturbed, such as wind speed, humidity, i=1,2,...,N, is the i th ensemble sample, is a spatial position, is a three-dimensional random field satisfying a normal distribution, is a random field satisfying a normal distribution at t time and considering time and space correlation, is the standard deviation of the perturbed variable, is a parameter representing the size of the time or space correlation; The perturbed meteorological and dust initial concentration of atmospheric composition is input to the weather-dust two-way feedback numerical mode for prediction, and the dust ensemble prediction result is obtained.

[0020] In the embodiment, the method of the optimal assimilation analysis system comprises: Based on the Kalman filtering theory, the ensemble Kalman filtering method is used for assimilation analysis, and the expression is: wherein is the ensemble optimal estimation field finally solved, also known as the assimilation analysis field, is the mode background field, namely the ensemble prediction result, is an observation and a perturbed vector thereof, For observation operator, the model variables are projected to the observation space, For observation increment, For analysis increment, is a scalar coefficient to adjust the relative contribution of the analysis increment and the model background field; is the observation error covariance matrix, is the error covariance matrix of the model background field, which is obtained by using the ensemble prediction results in the ensemble assimilation method; In the second aspect, the weather-sand-dust two-way feedback numerical model and the ensemble assimilation prediction system comprise: A sandstorm prediction module: based on a regional weather prediction model and online driving of sand-raising, transportation and deposition of multiple physical mechanisms of sand, a 168-hour sandstorm prediction model is established, and optical parameters are calculated based on the extinction theory of spherical particles to realize the two-way feedback of sand to weather; the optical parameters include extinction coefficient, single scattering ratio and asymmetry factor; A two-way feedback module: for a given sand aerosol-cloud interaction mechanism of ice nucleation, a real-time sand CCN type aerosol homogeneous freezing process, the optical parameters of the sand are input into the radiation model, the radiation temperature rate feedback is calculated to the dynamic process of the weather model, and the two-way feedback of weather driving sand and sand affecting weather is formed; An ensemble prediction module: considering the time and space correlation of the perturbation of the meteorological initial field and boundary conditions and the initial field of the sand aerosol input into the model, the ensemble prediction is carried out after inputting the model; An ensemble assimilation analysis correction module: the background error covariance matrix is constructed by using the ensemble prediction results of the model, the observation at the analysis time and before is disturbed, the initial concentration of the atmospheric aerosol such as sand of the model is assimilated and corrected based on the localized ensemble Kalman filter method, and more accurate sand initial field and prediction results and their influence on weather are obtained.

[0021] The above only describes the preferred embodiments of the present application and is not intended to limit the present application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for improving numerical prediction of sand dust by using assimilation-aerosol-cloud-radiation, characterized in that, The method comprises the following steps: A 168-hour dust storm prediction model is established based on a regional weather prediction model and online driving dust, and a plurality of physical mechanisms of dust lifting, transportation and deposition of the dust storm, optical parameters are calculated based on a spherical particle extinction theory, and a two-way feedback of the dust to the weather is realized; the optical parameters include a mass extinction coefficient, a comprehensive extinction coefficient, a comprehensive single scattering ratio and a comprehensive non-symmetry factor of the dust; Given a dust aerosol-cloud interaction mechanism of ice nucleus nucleation, a real-time homogeneous freezing process of dust CCN type aerosol, the optical parameters of the dust are input into a radiation model, a radiation temperature change rate is calculated and fed back to a dynamic process of the weather model, and a two-way feedback of the weather driving the dust and the dust affecting the weather is formed; An ensemble prediction system based on the weather-dust two-way feedback numerical model is constructed, time and space related perturbations of meteorological initial fields and boundary conditions of the model and dust aerosol initial fields are considered, and the model is input for ensemble prediction; An assimilation analysis correction system is established, a background error covariance matrix is constructed by using the ensemble prediction results of the model, observations at the analysis time are perturbed, a dust initial field of the model is assimilated and corrected based on a localized ensemble Kalman filtering method, and more accurate dust initial fields and prediction results and influences of the dust on the weather are obtained.

2. The integrated assimilation prediction method based on the weather-sand-dust two-way feedback numerical model according to claim 1, characterized in that, The method for establishing the 168-hour dust storm prediction model comprises the following steps: The total particle number (N) of sand and dust is realized by external mixing mode in Optical parameters and bidirectional feedback function are calculated in wave band The extinction efficiency, the single scattering albedo and the non-symmetry factor of the nth grade particle are obtained, the optical parameter mass extinction coefficient of the dust aerosol at different wave bands and different particle sizes is formed by using an external mixing mode according to a spherical particle extinction calculation theory, and an expression is as follows: Wherein the sand density is , the sand radius is , the extinction efficiency of the nth grade particle in the waveband is , and the optical parameter mass extinction coefficient is ; The comprehensive extinction coefficient of the dust is the sum of the extinctions of all N grades of particles, and an expression is as follows: wherein the dust integrated extinction coefficient is is ; The comprehensive single scattering ratio of the dust is obtained by weighted average of the extinction coefficients of all particle sizes on the comprehensive extinction coefficient, and an expression is as follows: Wherein in The integrated single scattering ratio of sand and dust in the wave band is , in The single scattering albedo of n-grade particles in the wave band is ; The comprehensive non-symmetry factor of the dust is calculated by weighted average of the scattering ratios of different particle sizes on the comprehensive single scattering ratio, and an expression is as follows: wherein the asymmetry factor is , in the waveband dust integrated asymmetry factor is ; The optical parameter mass extinction coefficient, the comprehensive extinction coefficient, the comprehensive single scattering ratio and the comprehensive non-symmetry factor of the dust are input into a radiation transfer model to replace fixed values in the model during the integral process of the model; The radiation transfer model calculates a radiation temperature change rate to complete real-time calculation of the radiation effect of the dust; the radiation temperature change rate includes the radiation temperature change rate affected by the dust, which is fed back to the dynamic process of the weather model in real time.

3. The integrated assimilation prediction method based on the weather-sand-dust two-way feedback numerical model according to claim 1, characterized in that, The method for the dust aerosol-cloud interaction mechanism of ice nucleus nucleation comprises the following steps: The ice crystal number concentration generated by the dust aerosol nucleation is calculated: The ice crystal number concentration generated by the sand dust aerosol nucleation is The temperature of the ambient atmosphere is T, in K, and the number concentration of the SD aerosol with a radius greater than 0.5 m is , in kg -1 The constants are a, b, and c, respectively, and a, b, and c are 0.0000594, 3.33, and 0.0033, respectively. The number concentration of the SD aerosol with a radius greater than 0.5 m is calculated: where m is the mass of the SD aerosol particle in kg, and the predictor is , where the predictor is the SD aerosol mass mixing ratio, tracerSD, with index num, the average radius, r, in mm, and the density, r, in g / cm 3 , where the predictor is the number concentration of the SD aerosol, N , in units of kg -1 , i, k, j are the model grid point positions; Then, the calculated Ndust0.5 is substituted into the calculation formula of the ice crystal number concentration generated by the dust aerosol nucleation, a real-time dust aerosol heterogeneous nucleation process is completed, and a real-time ice nucleus nucleation process affected by the dust-cloud is completed.

4. The integrated assimilation prediction method based on the weather-sand-dust two-way feedback numerical model according to claim 1, characterized in that, The method for the real-time homogeneous freezing process of the dust CCN type aerosol comprises the following steps: The ice formation process of the dust CCN type aerosol is calculated, and the ice crystal number concentration formed by the homogeneous freezing of the CCN type aerosol is calculated: where Nci is the ice crystal number concentration of CCN type aerosol homogeneous freezing, unit kg -1 , NWFA2 is the total CCN type aerosol number concentration, unit kg -1 , is the homogeneous nucleation rate coefficient, unit cm -3 / s, the value of homogeneous nucleation rate coefficient is mainly related to the ambient atmospheric temperature and supersaturation, dt is the time step of the model, the time step is 100 s; in the calculation of NWFA2, mc is the mass of aerosol, unit kg, the forecast quantity tracerc is the mass mixing ratio of CCN aerosol, unit kg kg -1 , numc is the number of tracerc, rc is the average radius of aerosol, unit mm, the density is rc, unit g / cm 3 , i, k, j are the positions of the grid points of the model.

5. The method and system for integrated assimilation prediction based on the weather-sand-dust two-way feedback numerical model according to claim 1, wherein, The ensemble prediction module comprises the following steps: The meteorological initial field and boundary condition and the initial field of dust aerosol inputted to the mode are disturbed according to the following formula, (12) wherein , is a variable requiring perturbation; i = 1, 2,..., N, is the i-th set sample, is a spatial position, is a three-dimensional random field satisfying a normal distribution, is a three-dimensional random field satisfying a normal distribution at time t and considering time and spatial correlation, is a standard deviation of the perturbation variable, is a parameter representing the size of time or spatial correlation; The disturbed meteorological and dust initial field is inputted to the weather-dust two-way feedback numerical mode to obtain the ensemble prediction result of the concentration of atmospheric aerosol such as dust. 6.The method and system of integrated assimilation prediction based on a weather-sand-dust two-way feedback numerical model according to claim 1, wherein, The method of the ensemble assimilation method and system comprises: matrix is the ensemble prediction result of the pattern analysis variable , and the expression is wherein is the number of samples in the set, is the dimensionality of the pattern variable; Let be the average of the matrix, then the anomaly matrix of , the expression is: Mode background error covariance matrix , is given by the expression Based on Kalman filtering theory, the ensemble Kalman filtering method is used for assimilation analysis, and the expression is: where is the final solution, called the assimilation analysis field, is the model background field, i.e. the ensemble prediction result, is the observation and its perturbation vector, is the observation operator, which projects the model variables into the observation space, is the observation increment, is the analysis increment, is a scalar coefficient, which adjusts the relative contribution of the analysis increment and the model background field; is the observation error covariance matrix, is the error covariance matrix of the model background field, which is obtained by analyzing the ensemble prediction result in the ensemble assimilation method. Because the atmospheric chemical mode includes too many kinds of aerosols and particle sizes, the mode variables are selected or combined, and the selected or combined variables are used as the analysis variables of data assimilation to determine the corresponding assimilation analysis field.

7. A weather-sand-dust two-way feedback numerical model and ensemble convection forecast system for performing the method of any one of claims 1-6, characterized in that, It comprises: The sandstorm prediction module is used for establishing a 168-hour sandstorm prediction mode based on the regional weather prediction mode and the online driving of the sand lifting, transportation and deposition of multiple physical mechanisms of dust; The dust-weather two-way feedback module: the dust-radiation two-way feedback is calculated based on the extinction theory of spherical particles to calculate the optical parameters, realize the feedback of the radiation effect of dust on weather, and realize the real-time homogeneous freezing process of dust CCN type aerosol by increasing the ice nucleation of dust. The ensemble prediction module: the initial field, boundary condition and initial concentration of atmospheric aerosol such as dust are disturbed to input the mode for ensemble prediction; The ensemble assimilation analysis correction module: the background error covariance matrix is constructed by using the ensemble prediction result of the mode, the observation at the analysis time and before is disturbed, the initial field of dust and the concentration of atmospheric aerosol are corrected by using the local ensemble Kalman filtering method, and more accurate initial field of dust and prediction result and its influence on weather are obtained. The ensemble prediction module: the initial field, boundary condition and initial concentration of atmospheric aerosol such as dust are disturbed to input the mode for ensemble prediction; The ensemble assimilation analysis correction module: the background error covariance matrix is constructed by using the ensemble prediction result of the mode, the observation at the analysis time and before is disturbed, the initial field of dust and the concentration of atmospheric aerosol are corrected by using the local ensemble Kalman filtering method, and more accurate initial field of dust and prediction result and its influence on weather are obtained.

Images