A workflow-based computational method for achieving high performance in CALPUFF

Abstract

The high-performance computing method of CALPUFF is realized based on workflow, which relates to the fields of atmospheric environment and computer processing technology. It solves the problems that the existing meteorological models have poor processing capabilities for abnormal situations, cannot support concurrency in calculations, have low computing efficiency, and are insufficient in analyzing and processing calculation results; and cannot support calculations for complex situations of multiple models, and can only solve single problems with a single script. The present invention is implemented based on an atmospheric model computing service system. According to the workflow model, a computing task is divided into multiple task nodes, which are completed one by one. This method integrates complex data, data analysis and processing, and realizes cloud service sharing. The atmospheric model computing service system of the present invention adopts a numerical simulation method to analyze the causes of atmospheric pollution under different climatic conditions and at different times within the range of the numerical model, and to identify the contribution rates of various pollution sources, various types of pollution sources, and each grid pollution source that affect atmospheric pollution, providing a decision-making support system for atmospheric pollution prevention and control.

Description

Technical Field

[0001] The present invention relates to the fields of atmospheric environment and computer processing technology, and in particular to a method for implementing CALPUFF high-performance computing through workflow and platformization. Background Art

[0002] There are meteorological models of different scales, from large to small: WRF, CALMET, and CALPUFF. If you want accurate CALPUFF calculation results at a small scale, you can first calculate the WRF model, then the CALMET model, and then use the results as the pre-calculation data for the CALPUFF model.

[0003] Currently, most calculations in this model are performed manually, resulting in heavy workloads, low computational efficiency, and the risk of errors. A small number of methods utilize simpler scripting tools to orchestrate the entire process, but this approach struggles with handling various anomalies, lacks support for concurrent calculations, suffers from low computational efficiency, and lacks the ability to analyze and process results. Furthermore, it lacks support for complex multi-model calculations, requiring only a single script to solve a single problem.

[0004] Current market demand calls for solutions that can stably and efficiently handle complex computational processes, with strong capabilities for integrating diverse data sources and analyzing and processing data. The core function of the atmospheric model computing service platform is to leverage an advanced system architecture to stably and efficiently handle complex computational flows. By integrating various data sources, this platform significantly improves the automation of complex computational processes and the ability to process and analyze the final results. Furthermore, the atmospheric model computing service platform can be made publicly available through cloud services, providing high-quality services to customers. Summary of the Invention

[0005] This invention aims to address the problems of existing meteorological models, such as their poor ability to handle abnormal situations, a lack of concurrent computing, low computational efficiency, insufficient ability to analyze and process results, and an inability to support complex multi-model calculations, requiring only a single script to solve a single problem. This invention provides a workflow-based high-performance CALPUFF computational method.

[0006] A workflow-based high-performance CALPUFF computing method is implemented based on an atmospheric model computing service system, which includes an atmospheric model management system, an atmospheric model computing system, and a task scheduling management system.

[0007] The third-party system interacts with the entire atmospheric model calculation service system through the atmospheric model management system;

[0008] The atmospheric model management system is used for user-oriented use and interaction with third-party systems; it is used to implement the management of computing models, case management, statistical analysis of model data, management of computing service interfaces, and system management functions;

[0009] The atmospheric model calculation system is responsible for completing the model calculation process and subsequent processing of the calculation results; specifically including WRF model calculation, CALMET model calculation, CALPUFF model calculation and CALPOST model calculation;

[0010] The atmospheric model calculation system divides a calculation task into multiple task nodes according to the workflow mode and completes them one by one; the task nodes include WRF calculation subnode, CALMET calculation subnode, CALPUFF calculation subnode, data processing subnode and image rendering subnode;

[0011] The task of the WRF calculation subnode is to complete the CALWRF model calculation; the task of the CALMET calculation subnode is to complete the CALMET model calculation; the task of the CALPUFF calculation subnode is to complete the CALPUFF model calculation; the task of the data processing subnode is to analyze and store the data of the CALPUFF model calculation results; the task of the image rendering subnode is mainly to render the image of the calpuff result data;

[0012] The task scheduling management system is responsible for the scheduling and management of various tasks in the atmospheric model calculation system; the task scheduling management system groups tasks, implements operations such as pause, immediate execution, execution according to time rules, and resumption of execution, and is also used to view task execution logs.

[0013] Beneficial effects of the present invention:

[0014] The method described in the present invention integrates complex computing processes: by integrating complex computing processes through a workflow engine. This not only allows for efficient coupling of various computing links, but also allows for flexible integration of computing links to cope with changes in computing processes in different computing scenarios.

[0015] Integrate complex data: Various data are used and generated throughout the entire complex computing process. Through independent data modules, these complex data sources can be effectively managed and used to provide strong data support for complex computing scenarios.

[0016] Data analysis and processing: Data processing and analysis occurs at two levels. The first level involves analyzing and processing CALPUFF calculation results. The second level combines CALPUFF calculation results with various business data for secondary analysis. By integrating these different levels and various business data sources, we can meet data analysis needs at a wider range of levels.

[0017] Cloud service sharing: Through cloud services, we can integrate with third-party platforms to provide convenient services to third-party users. Services can have two dimensions: one is functional, which can meet the business needs of different users; the second dimension is performance. The platform's concurrency advantages can provide users with efficient computing, significantly saving users' time costs.

[0018] The atmospheric model calculation service system described in the present invention uses a numerical simulation method to analyze the causes of atmospheric pollution under different climatic conditions and at different times within the scope of the numerical model, and to determine the contribution rates of various pollution sources, various types of pollution sources, and various grid pollution sources that affect atmospheric pollution, thereby providing a decision-making support system for atmospheric pollution prevention and control.

[0019] The atmospheric model calculation service system consists of multiple subsystems. Each subsystem needs to be deployed independently and shares different tasks. Through mutual collaboration, it completes functions such as model management, model calculation, result post-processing, and data analysis, and can provide secondary development interfaces and services in various ways.

[0020] The atmospheric model computing service system can run on a variety of operating systems, such as Linux and Windows. The platform adopts advanced technical architecture, supports concurrent computing, massive data processing, and real-time rendering of various types of images. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is the overall business architecture diagram of the atmospheric model computing service platform;

[0022] Figure 2 This is a diagram of the network architecture of the atmospheric model computing service system and its external systems in the workflow-based high-performance CALPUFF computing method described in the present invention;

[0023] Figure 3 The overall technical architecture diagram of the atmospheric model calculation service system;

[0024] Figure 4 This is the framework diagram of the MVC model;

[0025] Figure 5 This is the operating mode principle diagram of the WRF model;

[0026] Figure 6This is the parameter block diagram of the WRF model operation mode;

[0027] Figure 7 This is the effect diagram of the simulation using the WRF model;

[0028] Figure 8 This is the effect diagram of the simulation using the CALMET model;

[0029] Figure 9 This is the calculation principle diagram of the CALMET model. DETAILED DESCRIPTION

[0030] Specific implementation method 1. Combination Figures 1 to 9 This embodiment describes a workflow-based high-performance computing method for CALPUFF. The method is based on an atmospheric model computing service system (platform) and consists of three systems, specifically an atmospheric model management system, an atmospheric model computing system, and a task scheduling management system.

[0031] The third-party system interacts with the entire atmospheric model calculation service system through the atmospheric model management system;

[0032] The atmospheric model management system is used for user-oriented use and interaction with third-party systems; it is used to implement the management of computing models, case management, statistical analysis of model data, management of computing service interfaces, and system management functions;

[0033] The atmospheric model calculation system is responsible for completing the model calculation process and subsequent processing of the calculation results; specifically, it includes WRF model calculation, CALMET model calculation, CALPUFF model calculation, CALPOST model calculation and subsequent processing of the model results; the subsequent processing includes image rendering and contribution rate analysis.

[0034] After the case calculation is complete, the results can be rendered into various images. Users can pre-set rendering schemes for various images so that they can be referenced during the calculation process. Various statistical analyses can be performed on the calculation results.

[0035] The atmospheric model calculation system divides a calculation task into multiple task nodes according to the workflow mode and completes them one by one; the task nodes include WRF calculation subnode, CALMET calculation subnode, CALPUFF calculation subnode, data processing subnode, image rendering subnode and CALPOST calculation subnode; used for CALPOST model calculation;

[0036] The task of the WRF calculation subnode is used to complete the CALWRF model calculation; the task of the CALMET calculation subnode is used to complete the CALMET model calculation; the task of the CALPUFF calculation subnode is used to complete the CALPUFF model calculation; the task of the data processing subnode is used to analyze and store the data of the CALPUFF model calculation results; the task of the image rendering subnode is mainly to render the image of the CALPUFF result data; the CALPOST calculation subnode is used to calculate the CALPOST model.

[0037] The task scheduling management system is responsible for the scheduling and management of various tasks in the atmospheric model calculation system; the task scheduling management system groups tasks, implements operations such as pause, immediate execution, execution according to time rules, and resumption of execution, and is also used to view task execution logs.

[0038] Combine Figure 2 To explain this embodiment, the atmospheric model management system and the atmospheric model calculation system do not communicate directly, but rather interact through the model database and cache server. The task scheduling system interacts with the atmospheric model calculation system to schedule and allocate various computing tasks within the atmospheric model calculation system. The atmospheric model management system also interacts with the basic database, primarily to provide the various basic data required for case creation.

[0039] In this embodiment, the atmospheric model management system includes a model management module, a case management module, an emission reduction simulation module, a system management module, a traceability simulation module, and a computing service interface module;

[0040] The model management module is used to create and maintain models. Model creation requires users or third-party systems to set various customizable parameters for the model, including terrain parameters, pollutant parameters, model calculation parameters, and image rendering parameters. Based on the set model parameters, the model's base configuration file, terrain file, and workspace are generated. Users can add, delete, modify, and query already generated models.

[0041] The case management module is used to create, calculate, maintain, and query case results. Users or third-party systems create new models in the model management module and then create customized cases under the model. The system classifies cases with different calculation processes into different products, and users can select different calculation models and calculation processes to meet specific needs.

[0042] By customizing the case parameters, you can map most of the parameters in the CALPUFF model calculation, such as pollution parameters, emission data, and calculation time;

[0043] After a case is created, it is used to initiate calculations for the case, query the case calculation status, and the time spent on each calculation link; after the case calculation is completed, it is used to retrieve the case calculation results and various statistical analysis data based on the case calculation results;

[0044] Finally, the user performs maintenance operations on the case;

[0045] The emission reduction simulation module is used by the user to customize the emission reduction simulation plan, select relevant information of the emission reduction simulation, reduce emissions for a certain parameter, the emission reduction ratio, and reduce emissions for a certain monitoring station;

[0046] Users can see case-related information and status in the emission reduction simulation list, and the emission reduction simulation results are displayed graphically;

[0047] The computing service interface module registers the information published by the computing platform in the service registration center. Users can obtain the configuration information of the access platform related interface by querying the information of the service registration center. The third-party platform accesses the service through multiple protocols.

[0048] The system management module is used for user and authority management, third-party system interface authority management, log management, etc.

[0049] The traceability simulation module can simulate the scenario of unknown sudden emissions and calculate the possible range of sudden unknown emission sources based on simulation data and mathematical modeling.

[0050] In this embodiment, the task scheduling system includes a task management module, a task scheduling module and a log management module;

[0051] In the task management module, users create multiple task executors, each of which is responsible for executing all scheduled tasks under it. Users set the executor name, serial number, IP address, and port number. Users can view each executor through the list page and modify its parameters.

[0052] The user creates a new task, sets the corresponding executor, routing strategy, operation mode, task parameters, person in charge, and alarm email information;

[0053] The task scheduling module allows users to group tasks by executor and assign multiple tasks to a certain executor to isolate scheduled tasks of different businesses;

[0054] In the log management module, users can query log information through query conditions.

[0055] Specific implementation method 2: Figures 3 to 9This embodiment is an example of the workflow-based high-performance CALPUFF computing method described in Specific Implementation 1.

[0056] Combine Figure 3 To illustrate this embodiment, the overall technical architecture of the atmospheric model computing service platform adopts the design methods of MVC and SOA. The overall architecture can be divided into four layers, namely the basic layer, data layer, support layer and application layer.

[0057] The MVC pattern is a framework that forces the application's input, processing, and output to be separated. Using MVC, an application is divided into three core components: model, view, and controller. Each of them handles its own tasks. The relationship between the three layers is as follows: Figure 4 shown.

[0058] The view layer is the interface that users see and interact with. For existing web applications, the view is composed of HTML elements. In new web applications, HTML still plays a key role in the view, but new technologies are emerging, including Flash, markup languages like XHTML, XML / XSL, WML, and web services. The benefit of MVC is that it can handle many different views for an application. No actual processing occurs in the view; whether the data is stored online or a list of employees, the view simply serves as a way to output data and allow the user to manipulate it.

[0059] The model layer represents enterprise data and business rules. Of the three components of MVC, the model handles the most processing tasks. For example, it might use component objects such as EJBs and ColdFusion Components to work with the database. The data returned by the model is neutral, meaning the model is independent of the data format. This allows a model to provide data to multiple views. Since the code applied to the model only needs to be written once and can be reused by multiple views, code duplication is reduced.

[0060] The control layer (controller + service) accepts user input and calls models and views to meet user needs. Therefore, when clicking a hyperlink in a web page or sending an HTML form, the controller itself does not output anything or do any processing. It simply receives the request and decides which model component to call to handle the request, and then determines which view to use to display the returned data.

[0061] Service-Oriented Architecture (SOA) is a component model that decomposes an application's functional units (called services) and connects them through well-defined interfaces and contracts. Interfaces are defined in a neutral manner, independent of the hardware platform, operating system, and programming language used to implement the services. This allows services built on a variety of systems to interact in a unified and universal manner.

[0062] Service-oriented architecture (SOA) enables the distributed deployment, composition, and use of loosely coupled, coarse-grained application components across a network as needed. The service layer, the foundation of SOA, can be directly invoked by applications, effectively controlling the human dependencies that interact with software agents within the system.

[0063] SOA is a coarse-grained, loosely coupled service architecture where services communicate through simple, precisely defined interfaces, without involving underlying programming interfaces or communication models. SOA can be seen as a natural extension of the B / S model and XML / Web Services technologies.

[0064] SOA will help software engineers gain a deeper understanding of the development and deployment of various components within an enterprise architecture. It will also help enterprise system architects build entire business systems more quickly, reliably, and reusably. Compared to previous models, SOA-based systems can more easily adapt to rapid business changes.

[0065] The foundational layer construction is the foundational guarantee for project construction. In this implementation, the foundational layer primarily refers to the various software environments upon which the atmospheric model computing platform operates. Examples include operating systems, databases, modeling software, application server software, and other software. The following describes the various software components of the foundational layer.

[0066] In this implementation, the atmospheric model computing platform runs on the Linux operating system. This is primarily due to its exceptional stability, high computational efficiency, and robust security. Its modular core design allows for easy porting across different hardware platforms, resulting in high scalability. Its comprehensive network capabilities surpass those of other operating systems in terms of communication and networking. Portability is exceptional, meaning that an operating system can be ported from one platform to another and still function normally.

[0067] In this implementation, the WRF (Weather Research Forecast) model is a next-generation mesoscale forecast model and assimilation system developed and researched by scientists from numerous US research institutions and universities. The WRF model is a fully compressible, non-hydrostatic model that utilizes the Arakawa C grid, combines advanced numerical methods and data assimilation techniques, employs improved parameter schemes for multiple physical processes, and features multiple nesting capabilities and easy localization to different geographic locations. This integrated model system, combining numerical weather forecasting, atmospheric simulation, and data assimilation, is capable of improving simulation and forecasting of mesoscale weather from meters to thousands of kilometers.

[0068] Key features:

[0069] (1) The static model adds non-static items based on the MM5 model;

[0070] (2) In the horizontal direction, WRF-ARW uses the Arakawa C grid;

[0071] (3) The vertical coordinates are based on the topographic coordinate system, with the mean sea level pressure as the reference surface;

[0072] (4) It has a variety of physical parameterization schemes, including microphysical process scheme, cumulus convection parameterization scheme, land surface process scheme, boundary layer scheme, atmospheric radiation scheme, etc.

[0073] (5) The basic principles follow various laws such as Newton's second law, the conservation equations of mass and energy, and the gas test law;

[0074] (6)Multi-physics options;

[0075] (7) unidirectional and bidirectional nesting;

[0076] WRF model dynamics framework:

[0077] (1) Fast wave horizontal propagation: front difference-back difference scheme;

[0078] (2) Vertical propagation of sound waves: implicit scheme;

[0079] (3) Horizontal direction: Adams-Bashforth scheme;

[0080] (4) Vertical direction: Crank-Nicholson scheme;

[0081] (5) Turbulent kinetic energy (TKE), each phase of water: explicit, iterative, flux correction (called once every two time steps);

[0082] T, U, V (space) advection:

[0083] (1) Horizontal direction: energy and vorticity pseudo-energy conservation, quadratic conservation (square conservation), second order;

[0084] (2) Vertical direction: quadratic conservation, second order;

[0085] (3) Turbulent kinetic energy (TKE), water in all phases: reversible, flux-corrected, positive definite, and conservative.

[0086] Combine Figure 5 To explain this embodiment, Figure 5 The WRF Preprocessing System (WPS) is a module consisting of three programs that prepare input fields for real-world simulations. The three programs are used as follows: GEOGRID defines the model region (including the region's latitude and longitude, center point coordinates, grid nesting, number of horizontal grid points, and resolution) and interpolates topographic data (including terrain elevation, land use type, vegetation cover, soil type, etc.) from static terrain data onto the grid points; UNGRIB extracts meteorological feature fields from GRIB-formatted data; and METGRID horizontally interpolates these extracted meteorological feature fields onto the grid points defined by GEOGRID.

[0087] The WRF-ARW / NMM system is the core module of the WRF model. It has the advantages of high efficiency, easy to master and parallel operation. Before running the WRF program, the data needs to be processed by the real program. The meteorological element data interpolated by the METGRID program is recognized by the real program of the WRF model and vertically interpolated to the eta layer of the WRF model to generate the required boundary layer file and initialize the boundary conditions.

[0088] Combine Figure 6 and Figure 7 This implementation describes the WRF model. In this implementation, WRF model operation settings include basic data input, control file setup, and physical parameter settings. Basic data includes terrain, land use type, GFS (grid forecast data), FNL (historical meteorological reanalysis data), and surface and upper-air observation data. Simulation settings include basic settings such as nested grid range, map projection, and simulation time. The WRF model offers multiple physical parameterization schemes. Selecting the appropriate one can improve the simulation and forecast of mesoscale weather.

[0089] Taking the WRF simulation of Xingtai City, Hebei Province, as an example, a three-layer nested grid is created. The outer layer provides the boundary field for the inner layer, providing higher-precision meteorological data for the region. Grid resolutions are 36 km, 12 km, and 4 km. The first layer covers the entire country, the second layer covers most of eastern China, and the third layer covers most of the Beijing-Tianjin-Hebei region and surrounding areas. The model is divided into 23 vertical layers, with the top layer having an air pressure of 50 kPa. The model map uses the Lambert map projection, which is suitable for mid-latitude regions.

[0090] In the present embodiment, CALMET model is the diagnostic wind field calculation mode developed by Sigma Research Corporation (now a subsidiary of Earth Tech, Inc) recommended by U.S. EPA. CALMET utilizes the mass conservation continuity equation, a meteorological module that describes hourly wind field and temperature field in a three-dimensional grid simulation domain, its core part comprises diagnostic wind field and micro-meteorological field pattern, the diagnostic wind field module carries out terrain dynamics, overland flow, terrain blocking effect adjustment to initial guess wind field (mesoscale model output meteorological field, conventionally monitored ground and high-altitude meteorological data), produces the first step wind field, imports observation data, and produces final wind field by interpolation, smoothing, vertical velocity calculation, divergence minimization etc. CALMET module has considered the dynamic influence of terrain, oblique airflow and blocking effect in detail in the three-dimensional wind field simulation process.

[0091] Combine Figure 8 To illustrate this implementation, CALMET model modeling is performed, and a model model is created by taking Xingtai City, Hebei Province as an example. The simulation grid covers the area of Xingtai City, the grid resolution is 1km*1km, the grid range is 200km in the east-west direction and 150km in the north-south direction, the number of vertical layers is set to 10 layers, and there are 11 height planes, from 0 meters to 3000 meters above the ground, namely 0m, 20m, 40m, 80m, 160m, 320m, 640m, 1000m, 1500m, 2200m, and 3000m, and the time resolution is 1 hour.

[0092] The WRF mesoscale model simulation results are used as the initial meteorological field input of the CALMET model. Part of the grid meteorological field data is extracted and converted into the CALMET input format file. The extracted grid range covers the CALMET grid area (that is, covering the Xingtai city area).

[0093] In this embodiment, the CALPUFF model is an air quality diffusion model developed by Sigma Research Corporation (now a subsidiary of EarthTech, Inc.) and recommended by the U.S. Environmental Protection Agency (EPA). It consists of three parts: the CALMET meteorological module, the CALPUFF puff diffusion module, and the CALPOST post-processing module. It is an air quality diffusion model used to simulate unstable multi-layer, multi-species pollution (such as SO2, NO x The Gaussian puff diffusion model (e.g., a model of the air mass flow) considers the migration and diffusion of different pollutants under temporally and spatially varying meteorological conditions, as well as their dry and wet deposition processes and basic chemical conversion processes. It also accounts for the influence of complex terrain, overwater transport, coastal interface effects, and building subsidence, simulating pollutants emitted from source emissions through advection diffusion to estimate concentrations and deposition at pre-determined points.

[0094] The CALPUFF puff diffusion model has the following characteristics:

[0095] a) It can handle time-varying point source and non-point source pollution;

[0096] b) It can simulate areas ranging from tens of meters to hundreds of kilometers;

[0097] c) Pollutant concentrations can be predicted from one hour to one year;

[0098] d) A wide variety of pollutants can be simulated;

[0099] e) The linear removal process and chemical transformation mechanism of pollutants are taken into account, and the secondary generation of pollutants can be simulated, which is suitable for simulation under rough and complex terrain conditions.

[0100] Combine Figure 9 To explain this implementation, the CALPUFF model calculation is primarily divided into two parts: the CALMET meteorological processing module and the CALPUFF puff diffusion module. The CALMET module is used to generate the meteorological field files required by the CALPUFF main module. The CALPUFF module is the primary module of this model and is a multi-level Gaussian diffusion model used to simulate or predict multiple pollutants under unsteady and non-steady conditions. This module allows for concentration diffusion calculations, but requires external input of relevant data from pollution emission sources to determine the mass concentration distribution of pollutants (such as SO2 and NOx) under meteorological factors that vary over time and space.

[0101] Taking Xingtai City, Hebei Province as an example, a model is created. The simulation grid covers the area of Xingtai City with a grid resolution of 1km*1km. The grid range is 200km in the east-west direction and 150km in the north-south direction. The number of vertical layers is set to 10, and there are 11 height planes, from 0 meters to 3000 meters above the ground, namely 0m, 20m, 40m, 80m, 160m, 320m, 640m, 1000m, 1500m, 2200m, and 3000m.

[0102] Using CALMET results as input, combined with Xingtai City's pollution source emission inventory data, we simulated air quality and developed a customized air quality model. The simulation took into account both dry and wet deposition of pollutants, including both gaseous and particulate matter deposition.

[0103] In this implementation, the CALPUFF pollution sources are simulated using Xingtai City, Hebei Province as an example. The pollution source data is based on the 2017 pollutant emission inventory of Xingtai City. Pollutants include sulfur dioxide, nitrogen oxides, particulate matter PM10, fine particulate matter PM2.5, VOCs, carbon monoxide, etc. The emission inventory calculates pollution emissions from major industries, including fossil fuel stationary combustion sources, process sources, mobile sources, solvent use sources, agricultural sources, and dust sources. Among them, fossil fuel stationary combustion sources and process sources cover key industrial sectors in Xingtai City, including electric heating, industrial boilers, civilian boilers, steel, cement, chemical fiber, glass, coking, etc. Mobile sources are divided into road mobile sources and non-road mobile sources, and dust sources are divided into soil dust, road dust, construction dust, and yard dust.

[0104] Due to the large number of industrial enterprises in Xingtai, this simulation selected key pollutant-emitting enterprises as point source inputs. This totals approximately 360 point sources, primarily emitting sulfur dioxide, nitrogen oxides, particulate matter, and carbon monoxide. Mobile and dust sources were geographically divided into small grids for the area source input model. This area source counted approximately 4,000. Mobile sources primarily emitted nitrogen oxides, while dust sources were the primary sources of PM10 and PM2.5. This CALPUFF model simulated the pollution diffusion concentrations of sulfur dioxide, nitrogen oxides, PM10, fine particulate matter PM2.5, and carbon monoxide.

[0105] In this implementation, the model data database in the base layer uses PostgreSQL. PostgreSQL is a fully featured free software object-relational database management system based on POSTGRES version 4.2, developed by the Department of Computer Science at the University of California. Many of POSTGRES's advanced concepts only appeared later in commercial website databases. PostgreSQL supports most of the SQL standard and provides many other modern features, such as complex queries, foreign keys, triggers, views, transaction integrity, and multi-version concurrency control. PostgreSQL is also extensible in many ways, such as by adding new data types, functions, operators, aggregate functions, prime methods, and procedural languages. Furthermore, due to its flexible license, anyone can use, modify, and distribute PostgreSQL for any purpose free of charge.

[0106] The main advantages of PostgreSQL are as follows:

[0107] (1) The operating system supports WINDOWS, Linux, UNIX, MAC OS X, and BSD.

[0108] (2) In terms of basic functions, it supports ACID, relational integrity, database transactions, and Unicode multi-language.

[0109] (3) In terms of tables and views, PostgreSQL supports temporary tables, and materialized views can be simulated using stored procedures and triggers in PL / pgSQL, PL / Perl, PL / Python, or other procedural languages.

[0110] (4) In terms of indexing, it fully supports R- / R+tree indexes, hash indexes, reverse prime indexes, partial prime indexes, Expression indexes, GiST, and GIN (used to accelerate full-text retrieval). Starting from version 8.3, it supports bitmap indexes.

[0111] (5) For other objects, data domains are supported, as well as stored procedures, triggers, functions, external calls, and cursors. 7) In terms of data table partitioning, four types of partitioning are supported, namely range, hash, mixed, and list.

[0112] (6) In terms of transaction support, the support for transactions has undergone more thorough testing than that of MySQL.

[0113] (7) Regarding MyISAM table handling, MySQL uses table locking for MyISAM tables without transactions. A long-running query is likely to hinder updates to the table, while PostgreSQL does not have such a problem.

[0114] (8) From the perspective of stored procedures, PostgreSQL supports stored procedures, while MySQL does not currently support them. This is because the existence of stored procedures also avoids the transmission of a large number of original SQL statements over the network, and this advantage is obvious.

[0115] (9) In terms of subquery support, MySQL does not support subqueries.

[0116] (10) In terms of user-defined function extension, PostgreSQL can more conveniently use UDF (user-defined function) for extension.

[0117] Nginx (enginex), a high-performance HTTP and reverse proxy web server, also provides IMAP / POP3 / SMTP services. Its features include low memory usage and strong concurrency. In fact, Nginx's concurrency is indeed better than other web servers of the same type. Website users using Nginx include Baidu, JD.com, Sina, NetEase, Tencent, and Taobao. It has many excellent features:

[0118] (1) Nginx can be compiled and run on most Unix and Linux OS, and there is a Windows port.

[0119] (2) In the case of high connection concurrency, Nginx is a good alternative to Apache service: Nginx is one of the software platforms often chosen by business owners in virtual hosting in the United States. It can support responses with up to 50,000 concurrent connections.

[0120] (3) Nginx as a load balancing service: Nginx can directly support Rails and PHP programs to provide external services internally, and can also support external services as an HTTP proxy service. Nginx is written in C and is much better than Perl in terms of system resource overhead and CPU efficiency.

[0121] (4) Process static files, index files and automatic indexing; open file descriptor buffer.

[0122] (5) Non-cached reverse proxy acceleration, simple load balancing and fault tolerance.

[0123] (6) FastCGI, simple load balancing and fault tolerance.

[0124] (7) Modular structure. This includes gzipping, byte ranges, chunked responses, and filters such as the SSI filter. If multiple SSIs in a single page are processed by FastCG or other proxy servers, the processing can run in parallel without waiting for each other.

[0125] Tomcat, a core project within the Apache Software Foundation's Jakarta project, was developed jointly by Apache, Sun, and several other companies and individuals. Thanks to Sun's involvement and support, Tomcat consistently implements the latest Servlet and JSP specifications. Tomcat 5 supports the latest Servlet 2.4 and JSP 2.0 specifications. Due to its advanced technology, stable performance, and free nature, Tomcat is a favorite among Java enthusiasts and has gained recognition from some software developers, making it a popular web application server.

[0126] The Tomcat server is a free, open-source, lightweight web application server. It's widely used in small and medium-sized systems and in environments with limited concurrent users. It's the preferred choice for developing and debugging JSP programs. For beginners, think of it this way: once the Apache server is configured on a machine, it can be used to respond to HTML (an application program in the Standard Generalized Markup Language) page requests. Tomcat is actually an extension of the Apache server, but it runs independently. Therefore, when you run Tomcat, it actually runs as a separate process from Apache.

[0127] The trick is that, when configured correctly, Apache serves HTML pages, while Tomcat actually runs JSP pages and servlets. Tomcat, like web servers like IIS, has the ability to process HTML pages, and it's also a servlet and JSP container. A standalone servlet container is Tomcat's default mode. However, Tomcat's ability to handle static HTML is not as good as that of Apache. The latest version of Tomcat is 9.0.

[0128] Redis in the base layer is a key-value storage system. Similar to Memcached, it supports storing relatively more value types, including string, list, set, zset (sorted set) and hash. These data types support push / pop, add / remove, intersection, union, difference and richer operations, and these operations are atomic. On this basis, Redis supports various different sorting methods. Like Memcached, in order to ensure efficiency, data is cached in memory. The difference is that Redis will periodically write updated data to disk or write modification operations to additional record files, and on this basis, master-slave synchronization is implemented.

[0129] Redis is a high-performance key-value database. Its emergence largely offsets the shortcomings of key / value stores like memcached, and in some cases, it can complement relational databases. It offers easy-to-use clients for Java, C / C++, C#, PHP, JavaScript, Perl, Objective-C, Python, Ruby, and Erlang.

[0130] Redis supports master-slave synchronization. Data can be synchronized from a master to any number of slaves, and a slave can be a master connected to other slaves. This allows Redis to perform single-level tree replication. Data can be written to disk intentionally or unintentionally. Thanks to a fully implemented publish / subscribe mechanism, slaves can subscribe to a channel and receive a complete record of messages published by the master when synchronizing the tree anywhere. Synchronization greatly aids read scalability and data redundancy.

[0131] ZooKeeper, a distributed, open-source coordination service for distributed applications, is an open-source implementation of Google's Chubby and a key component of Hadoop and HBase. It provides consistency services for distributed applications, including configuration maintenance, domain name services, distributed synchronization, and group services. ZooKeeper's goal is to encapsulate complex and error-prone critical services, providing users with a simple, easy-to-use interface and a high-performance, stable system.

[0132] In this implementation, the GFS forecast data in the data layer uses the global forecast GFS data from the NCEP (National Centers for Environmental Prediction) official website as the initial boundary condition for the future weather forecast model. The forecast data can predict the weather for the next 8 days, a total of 192 hours, with a spatial resolution of 0.5°×0.5°. It is updated every 6 hours, with a forecast interval of 3 hours, and is reported four times a day (00UTC, 06UTC, 12UTC, and 18UTC). The data content includes parameters such as air pressure, temperature, wind speed, and humidity at different altitudes. The system uses an automated script to download the latest weather forecast data on a daily basis, automatically starting the WRF mode to forecast the future weather in Lanzhou.

[0133] In this embodiment, the WRF model data in the data layer includes meteorological data and basic geographic data. The meteorological data used to simulate historical weather uses meteorological reanalysis data FNL data. This is global gridded data jointly produced by the National Centers for Environmental Prediction (NCEP) and the National Center for Atmospheric Research (NCAR) of the United States and published on the NCEP official website. It uses the most advanced global data assimilation system and a complete database to perform quality control and assimilation processing on observation data from various data sources (ground, ship, radiosonde, satellite, etc.), thereby obtaining a complete set of reanalysis data with the characteristics of multiple time periods, high density, strong continuity, high resolution, and rich content. It can effectively make up for the shortcomings of conventional observation data in the analysis of severe weather.

[0134] FNL data has a spatial resolution of 1°×1° and a temporal resolution of 6 hours. Global data analysis is performed daily at 0:00, 0:00, 12:00, and 18:00 UTC. Data includes air pressure, temperature, relative humidity, rainfall, and other information. The data is available in two formats: GRIB1 and GRIB2. GRIB1 data runs from July 30, 1999, to December 6, 2007, while GRIB2 data runs from December 6, 2007, to the present day and is continuously updated. The meteorological data provided here is in GRIB2 format.

[0135] Basic geographic data includes terrain elevation, land use type, and other underlying surface information. Topographic data (GTOP030) has a resolution of 30°; land use type data (USGS and MODIS) includes 24 land types; and satellite land cover product data has a resolution of up to 30°. Other underlying surface data includes vegetation type, soil type, soil moisture, and soil texture.

[0136] Point source emission data includes each source's location, effective height, altitude, emission temperature, emission rate, units, and emission period. PTEMARB.DAT is a point source parameter file containing point source emission data and detailed, optionally variable emission parameters. In the point source file, chimney parameters and emission rates can be specified, but the plume rise height must be simulated and calculated using the CALPUFF model's formulas.

[0137] Point source files contain a range of time-invariant and time-varying quantities. Time-invariant quantities include chimney height, diameter, coordinates, building downwash flags, user-defined codes, and building height and width. Horizontal and vertical systems are used independently within the model. Horizontal coordinates are specified according to the meteorological grid, while vertical layer source information is derived from internal model calculations of plume lift. Time-varying quantities include outlet temperature, outlet velocity, and emission rates of various pollutants.

[0138] Point source parameters need to be obtained by processing the regional pollutant emission inventory. The information that needs to be read is:

[0139] Name, number of sources, types of pollutants emitted, UTM zone, start date, start time, end date, end time, CALPUFF version used, and simulation range tag.

[0140] The chemical formula and molecular weight of the pollutant.

[0141] A. Quantities that do not change with time: coordinates; chimney geometric height, diameter; altitude, building downwash mark and custom mark. When representing the building downwash mark,

[0142] 0 means not considering the downwash effect of buildings, 1 means considering the downwash effect of buildings.

[0143] B. Quantities that change with time: outlet temperature, outlet velocity and emission rates of various pollutants.

[0144] Area source emission data, including the BAEMARB.DAT area source parameter file, includes the location, effective height, altitude, initial diffusion coefficient, units, and emission period of each area source. This file contains detailed area source emission data, as well as variable emission parameters. In the area source file, source emission parameter values and emission rates can be specified at each step of the run, while the plume rise height for each source is simulated and calculated using the CALPUFF model's calculation formula.

[0145] The BAEMARB.DAT file is an ASCII data file that contains a header and data blocks. The data blocks contain variables that vary with time, as well as variables that do not vary with time.

[0146] Monitoring station data includes: monitoring station coordinates, name, and type. Air quality monitoring station data includes daily hourly monitoring values for conventional pollutants such as PM10, PM2.5, SO2, NO2, O3, and CO. Monitoring value types include hourly monitoring values and 24-hour averages for each pollutant. For ozone, the 8-hour sliding value and 24-hour sliding value are included. The Air Quality Index (AQI) is calculated based on the monitoring values for each pollutant. Air quality is then classified into different levels based on the AQI values and standards.

[0147] Emission enterprise information data

[0148] Information on pollutant-emitting enterprises includes:

[0149] The region where the enterprise belongs; the coordinates of the enterprise;

[0150] Altitude of the location;

[0151] Annual GDP of the enterprise;

[0152] Industry;

[0153] Pollutant emissions;

[0154] Discharge outlet information: discharge height, outlet temperature, outlet velocity, etc.

[0155] In this embodiment, the framework of the support layer is specifically composed of the following parts:

[0156] Spring Framework, Spring boot, Eventbus, Mybatis, Redisson, Netty, Velocity, Dubbo, HikariCP, CXF, Proj4j, Work Flow, Ganymed, Json and Caffeine;

[0157] The Spring Framework is a framework for building a multi-layer J2EE system based on IOC and AOP, allowing you to choose to use one of its modules according to your needs.

[0158] Spring Framework is the core framework of the entire platform, and various other components are integrated into the Spring Framework in a modular way.

[0159] Spring Boot is a new framework from the Pivotal team designed to simplify the initial setup and development of new Spring applications. The framework uses a specific approach to configuration, eliminating the need for developers to define boilerplate configurations. In this way, Spring Boot aims to become a leader in the rapidly developing field of rapid application development.

[0160] Spring Boot can make the configuration management of the entire platform very fast, with better flexibility and scalability.

[0161] EventBus is Guava's event handling mechanism and an elegant implementation of the Observer pattern (producer / consumer programming model) in the design pattern. For event listening and publish-subscribe patterns, EventBus is a very elegant and simple solution.

[0162] EventBus is responsible for message sending, caching, routing and other functions on this platform.

[0163] MyBatis is an excellent persistence framework that supports custom SQL, stored procedures, and advanced mapping. MyBatis eliminates nearly all JDBC code and manual parameter setting and result set fetching. MyBatis uses simple XML or annotations to configure and map native information, mapping interfaces and Java POJOs (Plain Ordinary Java Objects) to database records.

[0164] Mybatis is mainly responsible for the ORM function on this platform, responsible for database data reading and mapping.

[0165] Redisson is a Java in-memory data grid built on Redis. Built on the NIO-based Netty framework, Redisson leverages the advantages of the Redis key-value database. Building on the common interfaces of the Java utility package, it provides users with a series of common utility classes with distributed features. This enables a toolkit originally designed to coordinate single-machine, multi-threaded concurrent programs to coordinate distributed, multi-machine, multi-threaded concurrent systems, significantly simplifying the design and development of large-scale distributed systems. Furthermore, by combining distinctive distributed services, Redisson further simplifies the collaboration between programs in a distributed environment.

[0166] Redisson on this platform is mainly responsible for connecting to the Redis cache server, distributed locking, and performing various atomic operations on cache objects in the Redis cache server.

[0167] Netty is an open source Java framework provided by JBOSS. Netty provides an asynchronous, event-driven network application framework and tools for the rapid development of high-performance, highly reliable network server and client programs. In other words, Netty is a client and server programming framework based on NIO. Using Netty ensures that you can quickly and easily develop a network application, such as a client or server application that implements a specific protocol. Netty simplifies and streamlines the programming and development process for network applications, such as developing socket services based on TCP and UDP. "Fast" and "simple" do not necessarily lead to maintainability or performance issues. Netty is a carefully designed project that draws on the implementation experience of multiple protocols (including FTP, SMTP, HTTP, and other binary text protocols). Ultimately, Netty has successfully found a way to ensure ease of development while also guaranteeing the performance, stability, and scalability of its applications.

[0168] Netty is responsible for the connection, maintenance, and operation of basic HTTP, FTP, and TCP communication channels on this platform, allowing the platform to communicate with other external systems through various protocols.

[0169] Velocity is a Java-based template engine. It allows anyone to reference objects defined in Java code using a simple template language. It can generate SQL, PostScript, and XML from templates. It can also be used as a standalone tool to generate source code and reports, or as an integrated component of other systems.

[0170] Velocity is mainly responsible for managing the templates of various model parameter files on this platform and dynamically generating various parameter files based on the model parameter templates when the system is running.

[0171] Dubbo is an open-source, high-performance service framework developed by Alibaba. It enables applications to implement service input and output functions through high-performance RPC and can be seamlessly integrated with the Spring Framework. Dubbo is a high-performance, lightweight, open-source Java RPC framework that provides three core capabilities: interface-oriented remote method invocation, intelligent fault tolerance and load balancing, and automatic service registration and discovery. Its main features are as follows:

[0172] (1) High-performance RPC calls for interface proxies. Provides high-performance proxy-based remote call capabilities. The service is granular with interfaces, shielding developers from the underlying details of remote calls.

[0173] (2) Intelligent load balancing. Multiple built-in load balancing strategies can intelligently sense the health of downstream nodes, significantly reducing call latency and improving system throughput.

[0174] (3) Automatic service registration and discovery. Supports multiple registration center services and real-time perception of service instance online and offline.

[0175] (4) Highly scalable. Following the microkernel + plug-in design principle, all core capabilities such as Protocol, Transport, and Serialization are designed as extension points, treating built-in implementations and third-party implementations equally.

[0176] (5) Traffic scheduling during operation. Built-in routing strategies such as conditions and scripts can be configured to easily implement grayscale releases and prioritize the same data center by configuring different routing rules.

[0177] (6) Visualized service governance and operation and maintenance. Provides a wealth of service governance and operation and maintenance tools: query service metadata, service health status, and call statistics at any time, issue routing policies in real time, and adjust configuration parameters.

[0178] The Dubbo framework is mainly responsible for the release of microservices on this platform. The service can be used by other subsystems within the platform and by various third-party application systems on the Internet.

[0179] HiKariCP is a rising star in the database connection pool and is a database connection pool component with excellent performance.

[0180] In this platform, HiKariCP is mainly responsible for the creation, management, and scheduling of database connection pools. Because each system in the platform needs to frequently interact with the database, the database connection pool can greatly improve the efficiency of the system-database connection.

[0181] Apache CXF is an open-source services framework that helps you build and develop services using frontend programming APIs like JAX-WS. These services can support multiple protocols, such as SOAP, XML / HTTP, RESTful HTTP, or CORBA, and can run over multiple transports, such as HTTP, JMS, or JBI. CXF greatly simplifies the creation of services and, inheriting the XFire tradition, seamlessly integrates with Spring. XF includes a wide range of features, but focuses on the following areas:

[0182] Support for Web Services Standards: CXF supports multiple Web Services standards, including SOAP, Basic Profile, WS-Addressing, WS-Policy, WS-ReliableMessaging, and WS-Security. Frontends: CXF supports multiple "Frontend" programming models. CXF implements the JAX-WS API (following the JAX-WS 2.0 TCK version). It also includes a "simple frontend" that allows the creation of clients and EndPoints without annotations. CXF supports both WSDL-first development and code-first development models from Java. Easy to Use: CXF is designed to be more intuitive and easy to use. There are a large number of simple APIs for quickly building code-first services. Various Maven plug-ins also make integration easier, supporting the JAX-WS API and Spring 2.0's more simplified XML configuration method. Support for binary and legacy protocols: CXF is designed as a pluggable architecture that supports both XML and non-XML type bindings, such as JSON and CORBA.

[0183] CXF is mainly responsible for providing web service functions on this platform. This allows the system to provide services to third-party systems developed in various languages on the Internet.

[0184] Proj4j is the most famous map projection library for open-source GIS. Projections in software such as GRASS GIS, MapServer, PostGIS, Thuban, OGDI, Mapnik, TopoCad, and GDAL / OGR all directly or indirectly rely on Proj4. Proj4's main functions include converting longitude and latitude coordinates to geographic coordinates, and converting coordinate systems, including datum transformations. The following illustrates how to use this conversion function, both from the command line and through programming.

[0185] Proj4j is mainly responsible for the conversion functions of various coordinate systems and some GIS-related operations on this platform.

[0186] A workflow is a computational model of a workflow. It represents the logic and rules for organizing tasks sequentially within a workflow using an appropriate computer model and then performs computations on them. The primary problem that workflows address is how to automatically communicate information among multiple participants using computers according to predetermined rules to achieve a business goal. Workflows are part of the field of Computer Supported Cooperative Work (CSCW), which generally studies how groups of people can collaborate with the help of computers.

[0187] In this implementation, the workflow used is based on an open-source workflow engine, primarily responsible for managing and executing model calculations. Through workflow model management, the computational process can be modularized, making it flexible and adaptable to a variety of computing scenarios and requirements.

[0188] Ganymed SSH-2 for Java is a package that implements the SSH-2 protocol in pure Java. It can be used to connect to SSH servers directly from Java programs.

[0189] In this implementation, each system is deployed on a different Linux server. During system use, many functions rely on SSH protocol operations. Ganymed is primarily responsible for completing these operations through SSH protocol, assisting in the implementation of system functions.

[0190] JSON (JavaScript Object Notation) is a lightweight data exchange format. Based on a subset of ECMAScript (the JS specification developed by the European Computer Society), it uses a text format that is completely independent of the programming language to store and represent data. Its simplicity and clear hierarchical structure make JSON an ideal data exchange language. It is easy for humans to read and write, as well as for machines to parse and generate, and it significantly improves network transmission efficiency.

[0191] In this implementation, many interfaces and direct system data interactions use Json. Different Json components are used in different application scenarios based on their characteristics, such as Gson, fastJson, and JackJson.

[0192] Caffeine is a high-performance Java cache component that supports various caching strategies and offers superior performance.

[0193] In this implementation, Caffeine is responsible for caching small amounts of content. Combined with Redis, it can improve system operating efficiency in many aspects.

[0194] In this embodiment, the application layer integrates model management, case management, case calculation, meteorological data post-processing, source contribution rate analysis, emission reduction simulation, image rendering management, projection coordinate conversion, model result statistical analysis, task scheduling and system management;

[0195] The model management can support multiple models online computing services at the same time. Users can manage these models and modify model parameters on this platform.

[0196] Case management is used to create various calculation cases. Users can manage these cases on this platform, modify the parameters of these cases, and submit the calculations of the cases.

[0197] Case calculation can perform calculations on various cases, such as single case calculations, composite case calculations, and composite normalized case calculations.

[0198] After the meteorological data post-processing is completed for the atmospheric computing service platform, various result files are generated, such as CON files and LST files. The platform needs to parse these files and process them to generate the data files required for subsequent steps.

[0199] Source contribution rate analysis is used in the atmospheric computing service platform. Through composite case calculation or composite normalized case calculation, it can statistically analyze the contribution rate of each parameter of each pollution source in the case to a certain monitoring station.

[0200] Emission reduction simulations used in the atmospheric computing service platform can generate different emission reduction plans based on specified emission reduction targets, such as proportional emission reduction or GDP-optimized emission reduction.

[0201] Image rendering management is used to render the results of atmospheric computing service platform cases into various images after they are completed. Users can pre-set rendering schemes for various images so that they can be referenced during the calculation process.

[0202] Projection coordinate conversion is used to support the conversion of various projection coordinate systems in the atmospheric computing service platform. Through projection coordinate conversion, data from different projection coordinate systems can be standardized and used uniformly.

[0203] Statistical analysis of model results Various statistical analyses can be performed on the data of the calculation results.

[0204] Task scheduling is used to provide the functions of scheduling the execution, management, and log viewing of computing tasks on each computing node.

[0205] System management mainly includes user and permission management, third-party system interface permission management, log management, etc.

[0206] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the patent for this invention shall be determined by the appended claims.

Claims

1. A workflow-based high-performance computational method for CALPUFF, characterized by: The method is implemented based on an atmospheric model calculation service system, which includes an atmospheric model management system, an atmospheric model calculation system, and a task scheduling management system; The third-party system interacts with the entire atmospheric model calculation service system through the atmospheric model management system; The atmospheric model management system is used for user-oriented use and interaction with third-party systems; It is used to implement computing model management, case management, statistical analysis of model data, management of computing service interfaces, and system management functions; The atmospheric model management system includes a model management module, a case management module, an emission reduction simulation module, a system management module and a computing service interface module; The model management module is used to create and maintain models. Model creation requires users or third-party systems to set various customized model parameters, including terrain parameters, pollutant parameters, model calculation parameters, and image rendering parameters. Based on the set model parameters, the model's base configuration file, terrain file, and workspace are generated. Users can add, delete, modify, and query generated models. The case management module is used to create, calculate, maintain, and query case results. Users or third-party systems create new models in the model management module and then create customized cases under the model. The system classifies cases with different calculation processes into different products, and users can select different calculation models and calculation processes to meet their needs. By customizing case parameters, you can map the pollution parameters, emission data, and calculation time used in the CALPUFF model calculation. After a case is created, it is used to initiate calculations for the case, query the case calculation status, and the time spent on each calculation link; after the case calculation is completed, it is used to retrieve the case calculation results and various statistical analysis data based on the case calculation results; Finally, the user performs maintenance operations on the case; The emission reduction simulation module is used by users to customize emission reduction simulation plans and select relevant information for emission reduction simulation; Users can see case-related information and status in the emission reduction simulation list, and the emission reduction simulation results are displayed graphically; The computing service interface module registers the information published by the computing platform in the service registration center. Users can obtain the configuration information of the access platform related interface by querying the information of the service registration center. The third-party platform accesses the service through multiple protocols. The system management module is used for user and authority management, third-party system interface authority management, and log management; The atmospheric model calculation system is responsible for completing the model calculation process and subsequent processing of the calculation results; the calculation process includes WRF model calculation, CALMET model calculation, CALPUFF model calculation and CALPOST model calculation; Subsequent processing of the calculation results includes image rendering and contribution rate analysis; The atmospheric model calculation system divides a calculation task into multiple task nodes according to the workflow mode and completes them one by one; The task nodes include WRF calculation subnodes, CALMET calculation subnodes, CALPUFF calculation subnodes, data processing subnodes and image rendering subnodes; The task of the WRF calculation subnode is to complete the CALWRF model calculation; the task of the CALMET calculation subnode is to complete the CALMET model calculation; the task of the CALPUFF calculation subnode is to complete the CALPUFF model calculation; the task of the data processing subnode is to analyze and store the data of the CALPUFF model calculation results; the task of the image rendering subnode is mainly to render the image of the calpuff result data; The task scheduling management system is responsible for the scheduling and management of various tasks in the atmospheric model calculation system; the task scheduling management system groups tasks, implements operations such as pause, immediate execution, execution according to time rules, and resumption of execution, and is also used to view task execution logs.

2. The workflow-based high-performance computing method for CALPUFF according to claim 1, characterized in that: The atmospheric model management system does not communicate directly with the atmospheric model calculation system, but interacts through the model database and cache server; the task scheduling management system interacts with the atmospheric model calculation system to schedule and allocate various computing tasks in the atmospheric model calculation system; the atmospheric model management system interacts with the basic database to provide various data required to create cases.

3. The workflow-based high-performance computing method for CALPUFF according to claim 1, characterized in that: The task scheduling management system includes a task management module, a task scheduling module and a log management module; In the task management module, users create multiple task executors, each of which is responsible for executing all scheduled tasks under it. Users set the executor name, serial number, IP address, and port number. Users can view each executor through the list page and modify its parameters. The user creates a new task, sets the corresponding executor, routing strategy, operation mode, task parameters, person in charge, and alarm email information; The task scheduling module allows users to group tasks by executor and assign multiple tasks to a certain executor to isolate scheduled tasks of different businesses; In the log management module, users can query log information through query conditions.

Images