Aerosol assimilation method, system and medium based on multi-scale three-dimensional variation

Abstract

This invention discloses a multi-scale three-dimensional variational aerosol assimilation method, system, and medium. The method includes: acquiring an initial aerosol chemical field; inputting the initial aerosol chemical field into an air quality numerical model to obtain an aerosol background field; establishing multiple grid layers using a multigrid method; determining the background error covariance and super-observation data for each grid layer based on the aerosol background field and aerosol observation data; obtaining an updated aerosol analysis field through multi-scale three-dimensional variational assimilation based on the aerosol background field, the background error covariance, and the super-observation data; and obtaining the aerosol assimilation analysis results based on the updated aerosol analysis field. The embodiments of this application are beneficial for improving the accuracy of aerosol analysis, thereby improving the accuracy of air quality forecasting. This method can be widely applied in the field of three-dimensional variational assimilation technology.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional variational assimilation technology, and in particular to a method, system and medium for aerosol assimilation based on multi-scale three-dimensional variation. Background Technology

[0002] With the improvement of living standards, people are paying increasing attention to the impact of the environment on their health; among these concerns, air quality has become a focal point. Aerosols refer to multiphase mixed particulate matter composed of solid or liquid particles and gaseous carriers suspended in the atmosphere, and they play an indispensable role in regional air quality, human health, atmospheric visibility, and climate response. Therefore, accurate prediction of aerosol content in the air is becoming increasingly important. Among related technologies, using atmospheric chemical numerical models to simulate and predict the spatiotemporal distribution and pollution characteristics of aerosols is one of the main methods for studying aerosols. However, high-resolution numerical models cover a wide range of spatial scales, but single variational assimilation methods are difficult to adapt to this; multi-scale assimilation strategies are limited by computational costs in techniques based on aggregated assimilation methods; all of these problems affect the accuracy of aerosol analysis and cannot improve the accuracy of air quality forecasts. Summary of the Invention

[0003] The purpose of this invention is to at least partially solve one of the technical problems existing in the prior art.

[0004] Therefore, the purpose of this invention is to provide a highly accurate aerosol assimilation method, system, and medium based on multi-scale three-dimensional variation.

[0005] To achieve the aforementioned technical objectives, one aspect of this invention provides a multi-scale three-dimensional variational aerosol assimilation method, comprising the following steps: obtaining an initial aerosol chemical field; inputting the initial aerosol chemical field into an air quality numerical model to obtain an aerosol background field; establishing multiple grid layers using a multigrid method; determining the background error covariance and super-observation data for each grid layer based on the aerosol background field and aerosol observation data; obtaining an updated aerosol analysis field through multi-scale three-dimensional variational assimilation based on the aerosol background field, the background error covariance, and the super-observation data; the multi-scale three-dimensional variational assimilation being used for assimilation analysis along grid layers of different resolutions based on the multigrid method; and obtaining the aerosol assimilation analysis results based on the updated aerosol analysis field. This embodiment of the application characterizes different spatial scales of aerosols through multiple grid layers and improves the accuracy of aerosol analysis through multi-scale three-dimensional variational assimilation, thereby improving the accuracy of air quality forecasting.

[0006] In some embodiments, the multi-scale three-dimensional variational aerosol assimilation method of the present invention, wherein the step of obtaining the aerosol analysis update field through multi-scale three-dimensional variational assimilation based on the aerosol background field, the background error covariance, and the super-observation data includes:

[0007] Based on the aerosol background field, background error covariance, and super-observation data of each grid layer, the aerosol concentration of each grid layer is assimilated sequentially from the coarse-resolution grid layer to the fine-resolution grid layer to obtain the updated aerosol analysis field.

[0008] In some embodiments, in one embodiment of the present invention, the step of assimilating the aerosol concentration of each grid layer sequentially from the coarse-resolution grid layer to the fine-resolution grid layer based on the aerosol background field, background error covariance, and super-observation data of each grid layer to obtain the updated aerosol analysis field includes:

[0009] Obtain the initial aerosol field of the Nth grid layer, the Nth background error covariance, and the Nth super observation data. The value of the initial aerosol field of the Nth grid layer is the same as the value of the background aerosol field of the Nth grid layer. N is the number of multigrid layers set by the user.

[0010] The Nth aerosol preliminary field, the Nth background error covariance, and the Nth super observation data are assimilated to obtain the Nth aerosol analysis field.

[0011] The Nth aerosol gain field is interpolated to the (N-1)th grid layer through interpolation processing, and the (N-1)th aerosol background field of the (N-1)th grid layer is superimposed. The superimposed data is used as the (N-1)th aerosol initial guess field. The Nth aerosol gain field is the difference between the Nth aerosol analysis field and the Nth aerosol background field. The Nth grid layer is a coarse resolution grid layer, and the (N-1)th grid layer is a fine resolution grid layer.

[0012] Based on the initial aerosol field of the (N-1)th aerosol, the aerosol concentration of the (N-1)th grid layer is assimilated, and then the assimilation process of each finer grid layer is performed sequentially until the assimilation analysis of the finest resolution grid layer is completed, thus obtaining the updated aerosol analysis field.

[0013] In some embodiments, in one embodiment of the present invention, establishing multiple mesh layers using a multi-mesh method includes:

[0014] Based on the region to be predicted, determine the first simulation range and first resolution of the first grid layer;

[0015] Based on the first grid layer, the second simulation range and the second resolution of the second grid layer are determined, and so on until the Nth simulation range and the Nth resolution of the Nth grid layer are determined; the second resolution is less than the first resolution, and the second simulation range is greater than or equal to the first simulation range.

[0016] In some embodiments, in one embodiment of the present invention, determining the background error covariance and super-observation data for each grid layer based on the aerosol background field and aerosol observation data includes:

[0017] Based on the aerosol background field, the background error covariance of each aerosol component in each grid layer is constructed using the NMC method.

[0018] All aerosol observation data falling within any grid of any grid layer are obtained. Based on the distance between the observation point and the center point of the grid, all aerosol observation data are weighted and averaged to obtain super observation data for the grid. The observation point is the observation point corresponding to the aerosol observation data.

[0019] In some embodiments, in one embodiment of the present invention, the step of inputting the aerosol chemical initial field into an air quality numerical model to obtain an aerosol background field includes:

[0020] Determine the initial boundary field and anthropogenic source emissions;

[0021] The aerosol chemical initial field is input into the air quality numerical model, and the aerosol background field is obtained based on the initial boundary field and the anthropogenic source emissions.

[0022] In some embodiments, in one embodiment of the present invention, the method further includes:

[0023] The aerosol background field is processed by multiple interpolation steps to obtain an aerosol background field with multiple grid layers.

[0024] On the other hand, embodiments of the present invention propose an aerosol assimilation system based on multi-scale three-dimensional variation, comprising:

[0025] The first module is used to obtain the initial chemical field of aerosols;

[0026] The second module is used to input the aerosol chemical initial field into the air quality numerical model to obtain the aerosol background field.

[0027] The third module is used to create multiple mesh layers using the multi-mesh method;

[0028] The fourth module is used to determine the background error covariance and super observation data of each grid layer based on the aerosol background field and aerosol observation data.

[0029] The fifth module is used to obtain the updated aerosol analysis field by multi-scale three-dimensional variational assimilation based on the aerosol background field, the background error covariance, and the super-observation data; the multi-scale three-dimensional variational assimilation is used to perform assimilation analysis along grid layers of different resolutions based on the multigrid method.

[0030] The sixth module is used to obtain aerosol assimilation analysis results based on the aerosol analysis update field.

[0031] On the other hand, embodiments of the present invention provide an aerosol assimilation device based on multi-scale three-dimensional variation, comprising:

[0032] At least one processor;

[0033] At least one memory for storing at least one program;

[0034] When the at least one program is executed by the at least one processor, the at least one processor implements the above-described aerosol assimilation method based on multi-scale three-dimensional variation.

[0035] On the other hand, embodiments of the present invention provide a storage medium storing a processor-executable program, which, when executed by a processor, is used to implement the above-described aerosol assimilation method based on multi-scale three-dimensional variation.

[0036] The method provided in this application includes: acquiring an initial aerosol chemical field; inputting the initial aerosol chemical field into an air quality numerical model to obtain an aerosol background field; establishing multiple grid layers using a multigrid method; determining the background error covariance and super-observation data for each grid layer based on the aerosol background field and aerosol observation data; obtaining an updated aerosol analysis field through multi-scale three-dimensional variational assimilation based on the aerosol background field, the background error covariance, and the super-observation data; the multi-scale three-dimensional variational assimilation is used for assimilation analysis along grid layers of different resolutions based on the multigrid method; and obtaining the aerosol assimilation analysis results based on the updated aerosol analysis field. This application uses multiple grid layers to characterize different spatial scales of aerosols and improves the accuracy of aerosol analysis through multi-scale three-dimensional variational assimilation, thereby improving the accuracy of air quality forecasting. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following description is provided with accompanying drawings of the relevant technical solutions in the embodiments of the present invention or the prior art. It should be understood that the accompanying drawings described below are only for the purpose of clearly illustrating some embodiments of the technical solutions of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0038] Figure 1 This is a schematic flowchart of an embodiment of the aerosol assimilation method based on multi-scale three-dimensional variation provided by the present invention.

[0039] Figure 2 This is a schematic flowchart of another embodiment of the multi-scale three-dimensional variational aerosol assimilation method provided by the present invention.

[0040] Figure 3 A flowchart illustrating an embodiment of data preprocessing provided by the present invention;

[0041] Figure 4 A schematic flowchart of an embodiment of the multi-scale three-dimensional variational assimilation process provided by the present invention;

[0042] Figure 5 This is a schematic diagram of the structure of an embodiment of the multi-scale three-dimensional variational aerosol assimilation system provided by the present invention;

[0043] Figure 6 This is a schematic diagram of one embodiment of the aerosol assimilation device based on multi-scale three-dimensional variation provided by the present invention. Detailed Implementation

[0044] The embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. The step numbers in the following embodiments are set only for ease of explanation, and there is no limitation on the order between the steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0045] Aerosols are multiphase mixed particulate matter (PM) composed of solid or liquid particles and gaseous carriers suspended in the atmosphere. Their main components include black carbon, organic carbon, sulfates, nitrates, ammonium salts, marine elements, and crustal elements. With rapid economic development, air pollution has become increasingly severe, and aerosol pollution has become a major concern. Understandably, aerosol pollution has a significant impact on regional air quality, human health, atmospheric visibility, and climate response, making it a hot topic in atmospheric science research. Using atmospheric chemical numerical models to simulate and forecast the spatiotemporal distribution and pollution characteristics of aerosols is one of the main methods for aerosol research. However, due to significant uncertainties in meteorological dynamic processes, source emission inventories, initial chemical fields, and chemical reaction mechanisms, current numerical models for aerosol simulation and forecasting still face considerable challenges. Therefore, effectively improving the accuracy of aerosol simulation and forecasting has significant scientific and practical value. Data assimilation can reduce the uncertainty of the initial field of a model by using observational data, thereby improving the accuracy of simulation forecasts. It is a key technology for improving the analysis and forecasting capabilities of atmospheric models.

[0046] Three-dimensional variational assimilation (3DVar) methods can simultaneously perform global optimal analysis on multiple observational data, handle complex nonlinear observation operators, and have low computational cost, making them widely used in aerosol assimilation research. Background error covariance (BEC) is a crucial component of 3DVar, determining the influence weight, range, spatial smoothness, and balance between different control variables of the analysis gain field, directly impacting the data assimilation effect and the accuracy of subsequent model forecasts. High-resolution numerical models cover a wide range of spatial scales, which single variational assimilation methods struggle to handle because a single BEC rarely provides background error information across all scales. Multi-scale assimilation strategies have been studied in meteorological and oceanic assimilation, producing more accurate forecasts than traditional assimilation schemes. Atmospheric chemical assimilation without considering multi-scale information is merely a suboptimal use of observational data; multi-scale assimilation is a significant direction for the development of atmospheric chemical assimilation. Therefore, research on multi-scale aerosol assimilation is of great scientific importance to the development of atmospheric chemical assimilation. However, research on multi-scale aerosol assimilation remains scarce to date.

[0047] Meanwhile, regarding observational data, the complexity of aerosol composition and the relatively limited variety of observation types make the types of observational data and their application a key factor restricting assimilation effectiveness. In recent years, the construction of aerosol observation networks has expanded the sources of observational data. Although various aerosol observational data have been applied to data assimilation research and have shown positive improvement effects, most of these studies only differentiate between different types of data at the level of observational error covariance. Within the observational error range, each observation contains information from various scales in the real atmosphere. How to reflect the multi-scale information contained in observational data in aerosol assimilation has not yet been thoroughly studied, but it has important scientific guiding significance for in-depth research on multi-source aerosol data assimilation.

[0048] In atmospheric data assimilation, observed values ​​and model background values ​​are typically treated as random vectors, and optimal analysis is obtained based on the minimum error variance solution or maximum likelihood estimation. The quality of this analysis depends on the accuracy of the error covariance (BEC) associated with the model background field. With the development of numerical models, models with horizontal resolutions reaching the kilometer level have been applied to analyze cloud systems in the atmosphere and sub-mesoscale circulation systems in the ocean. These high-resolution numerical models encompass a wide range of spatial scales, and their dynamic and statistical characteristics differ across these scales. Since a single BEC is insufficient to provide background error information at all scales, traditional single assimilation methods are ill-suited for such fine-resolution models. Therefore, multi-scale data assimilation schemes have emerged to minimize the error information of the model background field at different scales during assimilation.

[0049] Applying multi-scale assimilation strategies to ensemble-based assimilation techniques (such as EnKF) is constrained by enormous computational costs, especially for high-resolution model systems. Variational assimilation methods, however, offer advantages such as low computational cost and the ability to perform globally optimal analysis on various datasets, making them commonly used in multi-scale assimilation research. Existing research in meteorology and oceanography utilizes multigrid methods to implement sequential multi-scale assimilation in 3DVar, reducing background error from large to small scales. The multigrid method represents different scales within the same simulation region by controlling the number of grid cells. Fewer grid cells and larger grid spacing represent larger scales, and vice versa. Similar strategies have been applied to the assimilation of sea surface temperature and radar radial wind. Some studies have proposed a novel multi-scale three-dimensional variational assimilation scheme, decomposing the cost function of variational assimilation into large-scale and small-scale components and applying it to ocean data assimilation. However, this method is challenging and difficult to extend to atmospheric chemical assimilation analysis. Other research has applied multigrid methods to four-dimensional variational assimilation in meteorology, leading to the development of a new meteorological assimilation forecasting system. However, four-dimensional variational assimilation is developed based on the numerical model used, which is highly dependent on a certain numerical model and is difficult to extend its application.

[0050] Some studies have achieved multi-scale three-dimensional variational assimilation of aerosols, but these studies employ a numerical model to simultaneously perform two-grid simulations of aerosols. The outer grid has a large range but low spatial resolution, while the inner grid has a small range but high spatial resolution. Although this study yields more accurate aerosol simulation and forecast results than a single-grid model, it requires two-grid simulations, increasing computational costs. Furthermore, the assimilation results from the two grids are not correlated, essentially performing separate assimilations, failing to transfer large-scale information to smaller scales, and the two-grid model does not provide additional spatial scale information. Moreover, this study does not consider utilizing multi-scale information from aerosol observation data.

[0051] The results of the above studies all indicate that multi-scale assimilation schemes can produce more accurate forecasts compared to traditional single-variable assimilation. However, to date, research on multi-scale assimilation has primarily focused on meteorology and oceanography, with very little research on aerosol multi-scale assimilation. Furthermore, there is currently no research on three-dimensional variational multi-scale assimilation of aerosols based on multigrid methods. This research, however, has significant scientific implications and reference value for the assimilation of atmospheric chemical data and the development of air quality forecasting systems.

[0052] In summary, the defects and shortcomings of existing technologies are as follows:

[0053] 1. A single background error covariance is insufficient to provide background error information at all scales, making traditional single assimilation methods unsuitable for high-resolution model systems;

[0054] 2. The multi-scale assimilation rate is limited by computational cost in techniques based on aggregate assimilation methods;

[0055] 3. Existing studies only distinguish different types of data at the level of constructing covariance of observation errors, without considering the multi-scale information carried by the observation data. Atmospheric chemical assimilation without considering multi-scale information is only a suboptimal use of the observation data.

[0056] 4. Among the existing multi-scale assimilation methods, although strategies such as the multigrid method have been attempted to be applied to assimilation in meteorological and oceanographic fields, the difficulty in implementing these methods makes it hard to extend their application to atmospheric chemical assimilation analysis.

[0057] 5. Existing multi-scale three-dimensional variational assimilation of aerosols uses a two-layer grid simulation in numerical models, increasing computational costs. Furthermore, the assimilation analysis processes within the two-layer grid are unrelated, essentially performing separate assimilation operations. Moreover, the two-layer grid cannot transmit sufficient spatial scale information, hindering the full utilization of the multi-spatial scale information contained in the observational data.

[0058] To address this issue, this invention discloses a multi-scale three-dimensional variational assimilation scheme for aerosols based on the multigrid method and the WRF / Chem numerical model, belonging to the field of environmental monitoring, early warning, and forecasting technology. Specifically, the multigrid method characterizes the multi-scale information of aerosols in the atmosphere with different horizontal grid resolutions. Taking into account both numerical model errors and observational data errors, the aerosol composition in the numerical model is corrected using multi-scale three-dimensional variational assimilation data to reduce the uncertainty of the initial field, thereby improving the accuracy of the numerical model's aerosol simulation and forecasting.

[0059] The following describes in detail, with reference to the accompanying drawings, the aerosol assimilation method and system based on multi-scale three-dimensional variation proposed according to embodiments of the present invention. First, the aerosol assimilation method based on multi-scale three-dimensional variation proposed according to embodiments of the present invention will be described with reference to the accompanying drawings.

[0060] Reference Figure 1This invention provides a multi-scale three-dimensional variational aerosol assimilation method. This method can be applied to terminals, servers, or software running on either. Terminals can be tablets, laptops, desktop computers, etc., but are not limited to these. Servers can be independent physical servers, server clusters or distributed systems composed of multiple physical servers, or cloud servers providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The multi-scale three-dimensional variational aerosol assimilation method in this invention mainly includes the following steps:

[0061] S100: Obtain the initial chemical field of aerosols;

[0062] S200: Input the aerosol chemical initial field into the air quality numerical model to obtain the aerosol background field;

[0063] S300: Multiple mesh layers are established using the multi-mesh method;

[0064] S400: Based on the aerosol background field and aerosol observation data, determine the background error covariance and super observation data for each grid layer;

[0065] S500: Based on the aerosol background field, the background error covariance, and the super-observation data, the aerosol analysis update field is obtained through multi-scale three-dimensional variational assimilation; the multi-scale three-dimensional variational assimilation is used for assimilation analysis along grid layers of different resolutions based on the multigrid method;

[0066] S600: Based on the aerosol analysis update field, obtain the aerosol assimilation analysis results.

[0067] In some possible implementations, the purpose of this invention is to address the deficiencies and shortcomings of the prior art and provide a widely applicable aerosol multi-scale three-dimensional variational assimilation scheme based on the multigrid method and the WRF / Chem numerical model. Specifically, referring to... Figure 2As shown, a multigrid method is used to represent different horizontal spatial scales of aerosols with varying grid horizontal resolution, reflecting the multi-scale information of aerosols in the real atmosphere. Considering both numerical model errors and observational data errors, and utilizing aerosol observation data, a multi-scale three-dimensional variational assimilation method is employed to optimize the aerosol components of the numerical model at different horizontal resolution grid layers. This reduces the uncertainty of the initial field, thereby improving the accuracy of subsequent aerosol simulation and forecasting. The multigrid-based aerosol multi-scale three-dimensional variational data assimilation method created in this invention can be efficiently extended to different aerosol numerical models and assimilation systems, improving the scalability and applicability of aerosol multi-scale assimilation. This invention mainly includes a data assimilation preprocessing module and an aerosol multi-scale three-dimensional variational assimilation module. It is understood that the air quality numerical model in the embodiments of this application can be a WRF / Chem numerical model, a CMAQ model, or a DAMx model; this application does not limit the specific choice of numerical model. In this embodiment, the multiple mesh layers are multiple mesh layers with different resolutions. Based on this, when performing three-dimensional variational assimilation processing, the mesh layers can be assimilated sequentially according to the order of resolution coarseness.

[0068] Reference Figure 2 As shown, it can be understood that the embodiments of this application determine the aerosol chemical initial field based on an idealized chemical initial field; obtain the aerosol background field (which can also be considered as the aerosol forecast field) through an air quality numerical model; obtain the aerosol analysis update field through multi-scale three-dimensional variational assimilation analysis; if the current time has not reached the set termination time, the obtained aerosol analysis update field is used as the new aerosol chemical initial field (which can be considered as the aerosol assimilation analysis result), and a new aerosol prediction analysis is performed through an air quality numerical model to obtain the aerosol forecast field. The aerosol forecast field can be used for air quality forecasting. It should be noted that the embodiments of this application determine the termination condition of the multi-scale three-dimensional variational assimilation analysis through a user-set termination time. Those skilled in the art can set the termination condition of the multi-scale three-dimensional variational assimilation analysis according to actual needs, and this application does not impose specific limitations.

[0069] Optionally, in one embodiment of the present invention, obtaining the updated aerosol analysis field through multi-scale three-dimensional variational assimilation based on the aerosol background field, the background error covariance, and the super-observation data includes:

[0070] Based on the aerosol background field, background error covariance, and super-observation data of each grid layer, the aerosol concentration of each grid layer is assimilated sequentially from the coarse-resolution grid layer to the fine-resolution grid layer to obtain the updated aerosol analysis field.

[0071] In some possible implementations, data assimilation can be performed on each grid layer in order from coarse-resolution grid layer to fine-resolution grid layer; of course, in other embodiments, data assimilation can also be performed on each grid layer in order from fine-resolution grid layer to coarse-resolution grid layer. Those skilled in the art can choose an appropriate order to assimilate the data corresponding to each grid layer according to actual needs.

[0072] Optionally, in one embodiment of the present invention, establishing multiple mesh layers using the multi-mesh method includes:

[0073] Based on the region to be predicted, determine the first simulation range and first resolution of the first grid layer;

[0074] Based on the first grid layer, the second simulation range and the second resolution of the second grid layer are determined, and so on until the Nth simulation range and the Nth resolution of the Nth grid layer are determined; the second resolution is less than the first resolution, and the second simulation range is greater than or equal to the first simulation range.

[0075] Optionally, in one embodiment of the present invention, determining the background error covariance and super-observation data for each grid layer based on the aerosol background field and aerosol observation data includes:

[0076] Based on the aerosol background field, the background error covariance of each aerosol component in each grid layer is constructed using the NMC method.

[0077] All aerosol observation data falling within any grid of any grid layer are obtained. Based on the distance between the observation point and the center point of the grid, all aerosol observation data are weighted and averaged to obtain super observation data for the grid. The observation point is the observation point corresponding to the aerosol observation data.

[0078] Optionally, in one embodiment of the present invention, the step of inputting the aerosol chemical initial field into the air quality numerical model to obtain the aerosol background field includes:

[0079] Determine the initial boundary field and anthropogenic source emissions;

[0080] The aerosol chemical initial field is input into the air quality numerical model, and the aerosol background field is obtained based on the initial boundary field and the anthropogenic source emissions.

[0081] Optionally, in one embodiment of the present invention, the method further includes:

[0082] The aerosol background field is processed by multiple interpolation steps to obtain an aerosol background field with multiple grid layers.

[0083] In some possible implementations, refer to Figure 3 As shown, the data preprocessing process of the data assimilation preprocessing module specifically includes steps S11 to S16:

[0084] Step S11: Build the WRF / Chem numerical model;

[0085] Step S12: Based on the study area, set the simulation range and horizontal grid resolution (defined as G_1 grid with a horizontal grid spacing of d1 km) for the WRF / Chem model, select the aerosol chemical mechanism, run the numerical model of step S11, and generate the initial aerosol field of the G_1 grid.

[0086] Step S13: Based on the higher horizontal resolution grid (G_1 grid) generated in step S12, double the horizontal grid spacing to construct a coarser horizontal resolution simulation grid (G_2 grid, with a horizontal grid spacing of d2 km, which is twice that of the previous layer grid (G_1 grid) with a coverage area no less than that of the previous layer grid (G_1 grid).

[0087] Step S14: Based on the set number of multi-mesh layers N, repeat step S13 to generate mesh layers G_i (i = 3, 4, ..., N) with the horizontal mesh spacing increasing by a factor of two;

[0088] Step S15: Using the National Weather Center (NMC) method, dynamic background error covariances (BEC_G_1, BEC_G_2, ..., BEC_G_N) are constructed for each aerosol component in the numerical model at each grid layer (G_1, G_2, ..., G_N) to reflect the background error information of the aerosol components in the numerical model. The classification of aerosol components varies depending on the aerosol scheme and mainly includes sulfate components, nitrate components, ammonium salt components, sodium salt components, chloride components, organic aerosols, black carbon aerosols, sea salt components, dust components, etc.

[0089] Step S16: Combining the numerical model grid layers G_i (i = 1, 2, ..., N) constructed in steps S12 to S14, construct corresponding super observation data OBS_G_i (i = 1, 2, ..., N) based on different grid layers of aerosol observation data. Specifically, determine whether the aerosol observation data falls within a certain grid of a certain grid layer based on its latitude and longitude, and then perform a weighted average based on the distance between the observation point and the center point of that grid point to obtain super observation data of aerosols in different grid layers, so as to reflect aerosol observation information at different spatial scales.

[0090] Optionally, in one embodiment of the present invention, the step of assimilating the aerosol concentration of each grid layer according to the aerosol background field, background error covariance, and super-observation data of each grid layer, sequentially from the coarse-resolution grid layer to the fine-resolution grid layer, to obtain the updated aerosol analysis field includes:

[0091] Obtain the initial aerosol field of the Nth grid layer, the Nth background error covariance, and the Nth super observation data. The value of the initial aerosol field of the Nth grid layer is the same as the value of the background aerosol field of the Nth grid layer. N is the number of multigrid layers set by the user.

[0092] The Nth aerosol preliminary field, the Nth background error covariance, and the Nth super observation data are assimilated to obtain the Nth aerosol analysis field.

[0093] The Nth aerosol gain field is interpolated to the (N-1)th grid layer through interpolation processing, and the (N-1)th aerosol background field of the (N-1)th grid layer is superimposed. The superimposed data is used as the (N-1)th aerosol initial guess field. The Nth aerosol gain field is the difference between the Nth aerosol analysis field and the Nth aerosol background field. The Nth grid layer is a coarse resolution grid layer, and the (N-1)th grid layer is a fine resolution grid layer.

[0094] Based on the initial aerosol field of the (N-1)th aerosol, the aerosol concentration of the (N-1)th grid layer is assimilated, and then the assimilation process of each finer grid layer is performed sequentially until the assimilation analysis of the finest resolution grid layer is completed, thus obtaining the updated aerosol analysis field.

[0095] In some possible implementations, refer to Figure 4 The assimilation process performed by the aerosol multi-scale three-dimensional variational assimilation module specifically includes steps S21 to S27:

[0096] Step S21: Using the WRF / Chem numerical model built in step S11, perform aerosol numerical simulation based on the simulation grid G_1 generated in step S12 and the selected aerosol chemical mechanism to generate the aerosol numerical background field BAK_G_1.

[0097] Step S22: Based on the aerosol scheme selected in step S12 and the number N of multigrids set by the user, BAK_G_1 generated in step S21 is interpolated sequentially to each grid G_i (i = 2, 3, ..., N) constructed in steps S14 to S15 using a linear horizontal interpolation method, thereby generating BAK_G_2, BAK_G_3, ..., BAK_G_N sequentially;

[0098] Step S23: Starting from the coarse resolution grid layer G_coarser (if this step is being performed for the first time, G_coarser is G_N; otherwise, coarser = N-1, ..., 2, 3), read the initial aerosol field FG_G_coarser of this grid layer (FG_G_coarser of the coarsest grid layer is equivalent to BAK_G_N), and call the aerosol forward operator, tangent linear operator, and adjoint operator of the corresponding chemical mechanism;

[0099] Step S24: The aerosol background error covariance BEC_G_coarser generated in steps S13-S14, the aerosol super observation OBS_G_coarser constructed in step S16, the aerosol initial guess field read in step S22, and the aerosol forward operator, tangent linear operator, and adjoint operator called in step S23 are optimized by three-dimensional variational assimilation to obtain the analysis field ANA_G_coarser of the concentration of each aerosol component on the coarser layer grid (if this step is performed for the first time, coarser = N, that is, G_coarser is G_N; otherwise, coarser = N-1, ..., 3, 2).

[0100] Step S25: Determine whether the assimilation analysis of the finest horizontal resolution grid layer has been completed and generate ANA_G_1. If yes, proceed directly to step S27. If not, subtract the aerosol background field BAK_G_coarser generated in step S22 from the aerosol analysis field ANA_G_coarser generated in step S24 to obtain the gain field AmB_G_coarser on the coarse resolution grid layer G_coarser. Then, interpolate AmB_G_coarser to the grid layer G_finner (finner = coarser-1) with doubled resolution using horizontal linear interpolation. Add this to the background field of the grid layer G_finner generated in step S22 to generate the initial aerosol field FG_G_finner of the grid layer G_finner. Assign G_finner to G_coarser for the next round of loop operation.

[0101] Step S26: Repeat steps S23 to S25 sequentially until the three-dimensional variational optimization of the finest horizontal resolution mesh layer G_1 is completed, and ANA_G_1 is obtained;

[0102] Step S27: Use ANA_G_1, the finest horizontal resolution grid layer, as the initial field for aerosols, and drive the WRF / Chem numerical model in a hot-start manner to perform subsequent aerosol simulation and prediction.

[0103] The three-dimensional variational assimilation process can be viewed as a quadratic functional minimization process of the distances between the analysis field and the background and observation fields of the state variable x in the numerical model, in order to minimize the error of x. Generally, this objective function is defined as:

[0104]

[0105] Where, x b Let x be the background field, i.e., the initial estimate field of the numerical model; y be the observation field; B be the background error covariance; and R be the observation error covariance. H(x) is the observation operator, which connects the observation information with the control variables of the numerical model. For regular observations, H(x) can be a simple horizontal and vertical interpolation, interpolating the regular grid simulation data to the location of the observation station.

[0106] High-resolution numerical models cover a wide range of spatial scales, which a single three-dimensional variational assimilation method cannot adequately address, as a single BEC (i.e., B in the above formula) can only provide background error information at a single scale. This invention achieves multi-scale variational assimilation of aerosols using a multigrid method and a three-dimensional variational assimilation method. The idea behind the multigrid method is to represent different spatial scales within the same simulation region by controlling the increase or decrease of the number of grids. Fewer grids and larger grid spacing represent larger spatial scales; more grids and smaller grid spacing represent smaller spatial scales. With each grid transformation from coarse to fine resolution, the grid spacing in the same horizontal direction is reduced by a factor of 2. This invention uses the three-dimensional variational assimilation method multiple times, starting assimilation from the coarse-resolution grid layer and progressing sequentially to the fine-resolution grid layer. After each coarse-resolution grid layer assimilation analysis, the aerosol variation values ​​at the coarse-resolution grid layer are obtained. These values ​​are then linearly interpolated to a finer-resolution grid layer and superimposed with the aerosol background field of that grid layer to obtain the initial aerosol prediction field. This process transfers aerosol assimilation information from a large scale to a smaller scale. Subsequently, a new three-dimensional variational assimilation analysis is performed. This process continues until the finest-resolution grid layer assimilation analysis is completed, obtaining the aerosol analysis field at the finest resolution. This invention, without altering the original three-dimensional variational assimilation code, can continuously transfer large-scale assimilation analysis information to small-scale assimilation analyses. This invention has strong applicability and scalability, and increases the feasibility of aerosol multi-model ensemble forecast assimilation.

[0107] The method provided in this application will be described in detail below with reference to a specific embodiment:

[0108] The execution order of each step in the example can be adapted according to the understanding of those skilled in the art.

[0109] Reference Figure 3The data assimilation preprocessing module steps are as follows:

[0110] Step S101: Set up the operating environment for running the WRF / Chem numerical model, including installing the Linux operating system, compiler, parallel software and necessary function libraries, and installing the WRF / Chem model;

[0111] Step S102: Set up the corresponding simulation region grid G_1 according to the specific study area;

[0112] Step S103: Double the horizontal grid spacing and construct a grid layer G_coarser, whose coverage is no less than that of the previous grid layer and should be kept as similar as possible;

[0113] Step S104: Determine whether G_coarser is the set coarsest horizontal resolution grid layer. If yes, proceed to step S105; otherwise, proceed to step S103.

[0114] Step S105: Collect the grid layers G_1, G_2, ..., G_N generated in steps S102 to S104 with progressively increasing horizontal grid spacing, where N is the number of multi-grid layers set by the user;

[0115] Step S106: For each grid layer in step S105, prepare the corresponding numerical model input file, including the initial meteorological boundary field and anthropogenic source emissions, and select the same aerosol scheme to run the WRF / Chem model to generate a series of aerosol forecast fields with different forecast lead times.

[0116] Step S107: Using the aerosol forecast field generated in step S106, construct the background error covariances BEC_G_1, BEC_G_2, ..., BEC_G_N of each aerosol component in each grid layer using the National Weather Center (NMC) method.

[0117] Step S108: Combine the aerosol observation data with each grid layer of step S105, and perform a weighted average of the aerosol observations at the same grid point in the same grid layer based on the distance between the observation point and the center of the grid point to construct the super aerosol observation data OBS_G_1, OBS_G_2, ..., OBS_G_N for each grid layer.

[0118] Reference Figure 4 The steps of the aerosol multiscale three-dimensional variational assimilation module are as follows:

[0119] Step S201: Run WRF / Chem to generate the aerosol numerical background field BAK_G_1 (at this time, BAK_G_1 is BAK_G_finner);

[0120] Step S202: Using the horizontal linear interpolation method, interpolate BAK_G_finner to a grid G_coarser with doubled horizontal grid spacing to generate BAK_G_coarser;

[0121] Step S203: Determine whether the BAK_G_coarser generated in step S202 is the background field of the coarsest horizontal resolution mesh layer. If yes, proceed to step S204; otherwise, use the BAK_G_coarser generated in step S202 as BAK_G_finner and proceed to step S202.

[0122] Step S204: Collect the background fields BAK_G_1, BAK_G_2, ..., BAK_G_N of the grid layers with progressively increasing horizontal grid spacing generated in steps S202 to S203;

[0123] Step S205: Prepare for cyclic assimilation, starting from the coarsest horizontal resolution grid layer, i.e., coarser = N, aerosol initial guess field FG_coarser = BAK_G_coarser, and assign finner = coarser-1;

[0124] Step S206: Take the initial aerosol concentration guess field FG_coarser generated in step S205 or step S210, the background error covariance BEC_G_coarser of each aerosol component in the corresponding grid layer generated in step S107, and the aerosol super-observation data OBS_G_coarser of the corresponding grid layer generated in step S108, and perform aerosol three-dimensional variational assimilation of the grid layer G_coarser to generate the analysis field ANA_G_coarser of each aerosol component;

[0125] Step S207: Determine whether the assimilation analysis of the finest resolution mesh layer has been completed, i.e., whether coarser equals 1; if yes, then ANA_G_coarser is the final aerosol analysis field, and the operation of this aerosol multi-scale three-dimensional variational assimilation module ends; if not, jump to step S208.

[0126] Step S208: Subtract the BAK_G_coarser of the corresponding grid layer in step S204 from the ANA_G_coarser generated in step S206 to obtain the aerosol assimilation gain field AmB_G_coarser.

[0127] Step S209: Interpolate the AmB_G_coarser from step S208 to the grid layer G_finner with a grid spacing reduced by half using the horizontal linear interpolation method, and add it to BAK_G_finner to obtain the initial aerosol field FG_G_finner of the grid layer G_finner.

[0128] Step S210: Reassign values ​​for aerosol variational assimilation in the next grid layer: coarser = finner, FG_G_coarser = FG_G_finner, finner = coarser-1; jump to step S206.

[0129] Reference Figure 2 The steps for performing aerosol assimilation forecasting cyclic operation using this invention are as follows:

[0130] Step S301: Using the WRF / Chem numerical model built in step S101, based on the G_1 simulation region grid constructed in step S102, input the meteorological initial boundary field and anthropogenic emission files corresponding to the G_1 simulation grid, and use the idealized chemical initial field as the aerosol initial field.

[0131] Step S302: Input the prepared data file into WRF / Chem mode, run mode, and perform forward simulation integration to obtain the aerosol forecast field, which is used as the aerosol background field BAK_G_1 (equivalent to step S201);

[0132] Step S303: The aerosol background field BAK_G_1 generated in step S302, the background error covariances of each aerosol component in each grid layer constructed in step S107 BEC_G_1, BEC_G_2, ..., BEC_G_N, and the super observation data of each aerosol in each grid layer constructed in step S108 OBS_G_1, OBS_G_2, ..., OBS_G_N are used as input data files. The aerosol multi-scale three-dimensional variational assimilation module of the present invention is called to perform aerosol assimilation analysis to obtain the optimized analysis update field of each aerosol component.

[0133] Step S304: Use the aerosol component optimization analysis field obtained in step S303 as the aerosol initial field, and prepare the corresponding meteorological initial boundary field and anthropogenic source emission files;

[0134] Step S305: Determine whether the current simulation time is equal to the user-set termination time. If yes, end the run; otherwise, proceed to step S302 for looping.

[0135] The beneficial effects of this invention are:

[0136] 1. Based on the specific simulation period and grid layers with different horizontal resolutions, this invention constructs dynamic aerosol background error covariance at different spatial scales, taking into account the background error information characteristics of the actual simulation area, simulation period, and multi-spatial scale aerosol numerical model.

[0137] 2. This invention achieves multi-scale variational assimilation of aerosols through the multi-grid method and the three-dimensional variational assimilation method. The multi-grid method achieves the presentation of different spatial scales by controlling the grid spacing. Based on this, by performing three-dimensional variational assimilation multiple times, the assimilation analysis information of large scales can be continuously and sequentially transferred to the assimilation analysis of small scales.

[0138] 3. This invention couples aerosol observation data with grid layers of different resolutions and obtains super aerosol observation data of different grid layers through a distance-weighted averaging method, thereby realizing the utilization of aerosol observation data at multiple spatial scales;

[0139] 4. The aerosol multi-scale three-dimensional variational assimilation module constructed in this invention is simple and direct, requiring no modification to the internal code of the assimilation analysis. Only the input file of the assimilation analysis needs to be changed, which has strong applicability and scalability, enhances the application prospects of aerosol multi-scale assimilation in atmospheric chemical numerical models, and increases the feasibility of multi-model multi-scale ensemble forecast assimilation of aerosols.

[0140] In summary, the above method includes: obtaining an initial aerosol chemical field; inputting the initial aerosol chemical field into an air quality numerical model to obtain an aerosol background field; establishing multiple grid layers using a multigrid method; determining the background error covariance and super-observation data for each grid layer based on the aerosol background field and aerosol observation data; obtaining an updated aerosol analysis field through multi-scale three-dimensional variational assimilation based on the aerosol background field, the background error covariance, and the super-observation data; the multi-scale three-dimensional variational assimilation is used for assimilation analysis along grid layers of different resolutions based on the multigrid method; and obtaining the aerosol assimilation analysis results based on the updated aerosol analysis field. This application's embodiments characterize different spatial scales of aerosols through multiple grid layers and improve the accuracy of aerosol analysis through multi-scale three-dimensional variational assimilation, thereby improving the accuracy of air quality forecasts.

[0141] Secondly, refer to the appendix Figure 5 This invention describes an aerosol assimilation system based on multi-scale three-dimensional variation, proposed according to embodiments of the present invention.

[0142] Figure 5 This is a schematic diagram of a multi-scale three-dimensional variational aerosol assimilation system according to an embodiment of the present invention. The system specifically includes:

[0143] The first module 510 is used to acquire the initial chemical field of aerosols;

[0144] The second module 520 is used to input the aerosol chemical initial field into the air quality numerical model to obtain the aerosol background field.

[0145] The third module, 530, is used to establish multiple mesh layers using the multi-mesh method.

[0146] The fourth module 540 is used to determine the background error covariance and super observation data of each grid layer based on the aerosol background field and aerosol observation data.

[0147] The fifth module 550 is used to obtain the updated aerosol analysis field by multi-scale three-dimensional variational assimilation based on the aerosol background field, the background error covariance, and the super observation data; the multi-scale three-dimensional variational assimilation is used to perform assimilation analysis along grid layers of different resolutions based on the multigrid method.

[0148] The sixth module 560 is used to obtain aerosol assimilation analysis results based on the aerosol analysis update field.

[0149] It is evident that the content of the above method embodiments is applicable to this system embodiment. The specific functions implemented in this system embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0150] Reference Figure 6 This invention provides an aerosol assimilation device based on multi-scale three-dimensional variation, comprising:

[0151] At least one processor 410;

[0152] At least one memory 420 is used to store at least one program;

[0153] When the at least one program is executed by the at least one processor 410, the at least one processor 410 implements the aerosol assimilation method based on multi-scale three-dimensional variation.

[0154] Similarly, the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0155] This invention also provides a computer-readable storage medium storing a processor-executable program, which, when executed by a processor, is used to perform the above-described aerosol assimilation method based on multi-scale three-dimensional variation.

[0156] Similarly, the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0157] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order shown in the operation diagrams. For example, depending on the functions / operations involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts of this invention are provided by way of example to provide a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is altered and sub-operations described as part of a larger operation are executed independently.

[0158] Furthermore, although the invention has been described in the context of functional modules, it should be understood that, unless otherwise stated, one or more of the functions and / or features may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the invention. Rather, given the properties, functions, and internal relationships of the various functional modules in the apparatus disclosed herein, the actual implementation of the module will be understood within the scope of conventional skill of an engineer. Therefore, those skilled in the art can implement the invention as set forth in the claims using ordinary techniques without excessive experimentation. It is also understood that the specific concepts disclosed are merely illustrative and not intended to limit the scope of the invention, which is determined by the full scope of the appended claims and their equivalents.

[0159] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several programs to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0160] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a ordered list of executable programs for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, a program execution system or apparatus (such as a computer-based system, a processor-included system, or other system that can retrieve and execute a program from or in conjunction with such a program execution system or apparatus). For the purposes of this specification, "computer-readable medium" can mean any means that can contain, store, communicate, propagate, or transmit a program for use by or in conjunction with a program execution system or apparatus.

[0161] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0162] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable program execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0163] In the foregoing description of this specification, references to terms such as "one embodiment," "another embodiment," or "some embodiments" indicate that a specific feature, structure, material, or characteristic described in connection with an embodiment or example is included in at least one embodiment or example of the present invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0164] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

[0165] The above is a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of the present invention.

Claims

1. A multi-scale three-dimensional variational aerosol assimilation method, characterized in that, Includes the following steps: Obtain the initial chemical field of aerosols; The aerosol chemical initial field is input into the air quality numerical model to obtain the aerosol background field; Multiple mesh layers are established using the multi-mesh method; Based on the aerosol background field and aerosol observation data, determine the background error covariance and super observation data for each grid layer; Based on the aerosol background field, the background error covariance, and the super-observation data, the updated aerosol analysis field is obtained through multi-scale three-dimensional variational assimilation; the multi-scale three-dimensional variational assimilation is used to perform assimilation analysis along grid layers of different resolutions based on the multigrid method; Based on the aerosol analysis update field, the aerosol assimilation analysis results are obtained; The process of obtaining the updated aerosol analysis field through multi-scale three-dimensional variational assimilation based on the aerosol background field, the background error covariance, and the super-observation data includes: Based on the aerosol background field, background error covariance and super observation data of each grid layer, the aerosol concentration of each grid layer is assimilated sequentially from coarse resolution grid layer to fine resolution grid layer to obtain the updated aerosol analysis field. The process involves assimilating the aerosol concentration at each grid layer based on the aerosol background field, background error covariance, and super-observation data, sequentially from coarse-resolution to fine-resolution grid layers, to obtain the updated aerosol analysis field, including: Obtain the initial aerosol field of the Nth grid layer, the Nth background error covariance, and the Nth super observation data. The value of the initial aerosol field of the Nth grid layer is the same as the value of the background aerosol field of the Nth grid layer. N is the number of multigrid layers set by the user. The Nth aerosol preliminary field, the Nth background error covariance, and the Nth super observation data are assimilated to obtain the Nth aerosol analysis field. The Nth aerosol gain field is interpolated to the (N-1)th grid layer through interpolation processing, and the (N-1)th aerosol background field of the (N-1)th grid layer is superimposed. The superimposed data is used as the (N-1)th aerosol initial guess field. The Nth aerosol gain field is the difference between the Nth aerosol analysis field and the Nth aerosol background field. The Nth grid layer is a coarse resolution grid layer, and the (N-1)th grid layer is a fine resolution grid layer. Based on the initial aerosol field of the N-1th aerosol, the aerosol concentration of the N-1th grid layer is assimilated, and the assimilation process of each finer grid layer is performed sequentially until the assimilation analysis of the finest resolution grid layer is completed, thus obtaining the updated aerosol analysis field. The step of determining the background error covariance and super-observation data for each grid layer based on the aerosol background field and aerosol observation data includes: Based on the aerosol background field, the background error covariance of each aerosol component in each grid layer is constructed. All aerosol observation data falling within any grid of any grid layer are obtained. Based on the distance between the observation point and the center point of the grid, all aerosol observation data are weighted and averaged to obtain super observation data for the grid. The observation point is the observation point corresponding to the aerosol observation data.

2. The aerosol assimilation method based on multi-scale three-dimensional variation as described in claim 1, characterized in that, The method of establishing multiple mesh layers using a multi-mesh approach includes: Based on the region to be predicted, determine the first simulation range and first resolution of the first grid layer; Based on the first grid layer, the second simulation range and the second resolution of the second grid layer are determined, and so on until the Nth simulation range and the Nth resolution of the Nth grid layer are determined; the second resolution is less than the first resolution, and the second simulation range is greater than or equal to the first simulation range.

3. The aerosol assimilation method based on multi-scale three-dimensional variation as described in claim 1, characterized in that, The step of inputting the initial aerosol chemical field into the air quality numerical model to obtain the aerosol background field includes: Determine the initial boundary field and anthropogenic source emissions; The aerosol chemical initial field is input into the air quality numerical model, and the aerosol background field is obtained based on the initial boundary field and the anthropogenic source emissions.

4. The aerosol assimilation method based on multi-scale three-dimensional variation as described in claim 1, characterized in that, The method further includes: The aerosol background field is processed by multiple interpolation steps to obtain an aerosol background field with multiple grid layers.

5. An aerosol assimilation system based on multi-scale three-dimensional variation, characterized in that, include: The first module is used to obtain the initial chemical field of aerosols; The second module is used to input the aerosol chemical initial field into the air quality numerical model to obtain the aerosol background field. The third module is used to create multiple mesh layers using the multi-mesh method; The fourth module is used to determine the background error covariance and super observation data of each grid layer based on the aerosol background field and aerosol observation data. The fifth module is used to obtain the updated aerosol analysis field by multi-scale three-dimensional variational assimilation based on the aerosol background field, the background error covariance, and the super-observation data; the multi-scale three-dimensional variational assimilation is used to perform assimilation analysis along grid layers of different resolutions based on the multigrid method. The sixth module is used to obtain aerosol assimilation analysis results based on the aerosol analysis update field; The process of obtaining the updated aerosol analysis field through multi-scale three-dimensional variational assimilation based on the aerosol background field, the background error covariance, and the super-observation data includes: Based on the aerosol background field, background error covariance and super observation data of each grid layer, the aerosol concentration of each grid layer is assimilated sequentially from coarse resolution grid layer to fine resolution grid layer to obtain the updated aerosol analysis field. The process involves assimilating the aerosol concentration at each grid layer based on the aerosol background field, background error covariance, and super-observation data, sequentially from coarse-resolution to fine-resolution grid layers, to obtain the updated aerosol analysis field, including: Obtain the initial aerosol field of the Nth grid layer, the Nth background error covariance, and the Nth super observation data. The value of the initial aerosol field of the Nth grid layer is the same as the value of the background aerosol field of the Nth grid layer. N is the number of multigrid layers set by the user. The Nth aerosol preliminary field, the Nth background error covariance, and the Nth super observation data are assimilated to obtain the Nth aerosol analysis field. The Nth aerosol gain field is interpolated to the (N-1)th grid layer through interpolation processing, and the (N-1)th aerosol background field of the (N-1)th grid layer is superimposed. The superimposed data is used as the (N-1)th aerosol initial guess field. The Nth aerosol gain field is the difference between the Nth aerosol analysis field and the Nth aerosol background field. The Nth grid layer is a coarse resolution grid layer, and the (N-1)th grid layer is a fine resolution grid layer. Based on the initial aerosol field of the N-1th aerosol, the aerosol concentration of the N-1th grid layer is assimilated, and the assimilation process of each finer grid layer is performed sequentially until the assimilation analysis of the finest resolution grid layer is completed, thus obtaining the updated aerosol analysis field. The step of determining the background error covariance and super-observation data for each grid layer based on the aerosol background field and aerosol observation data includes: Based on the aerosol background field, the background error covariance of each aerosol component in each grid layer is constructed. All aerosol observation data falling within any grid of any grid layer are obtained. Based on the distance between the observation point and the center point of the grid, all aerosol observation data are weighted and averaged to obtain super observation data for the grid. The observation point is the observation point corresponding to the aerosol observation data.

6. An aerosol assimilation device based on multi-scale three-dimensional variation, characterized in that, include: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the aerosol assimilation method based on multi-scale three-dimensional variation as described in any one of claims 1 to 4.

7. A computer-readable storage medium storing a processor-executable program, characterized in that, The processor-executable program, when executed by the processor, is used to implement the multi-scale three-dimensional variational aerosol assimilation method as described in any one of claims 1 to 4.

Images