集合预报产品综合分析显示平台关键技术与实现

摘要

针对集合预报方法在天气预报业务中的应用，开发了具有自主知识产权的集合预报产品综合分析显示平台。以集合预报模式输出数据量大、气象图表显示效率和质量要求高两个主要需求为出发点，采用客户端服务器架构设计。服务器端将原始数据转换为产品数据以提高客户端执行效率。该文详细分析了平台关键技术，针对数据延时问题，轮询式数据处理技术实时检查原始数据变化状态并更新产品，采用生产者消费者互斥方法解决多线程锁死问题。为提高图表美观程度，动态页面布局显示技术对所有图形要素进行分类，并给出显示属性的抽象描述，结合图形渲染技术，实现了看图模式和出图模式的动态切换。该平台为预报员和服务决策者提供了宝贵的不确定性信息，在中小尺度极端天气预报、台风路径预报中发挥了重要作用。%In response to the impendent requirement of ensemble forecast applications in modern weather forecast operations,an ensemble forecast product analysis and display platform named NUMBERS (NUmerical Model Blending and Ensemble foRecast System)is developed.The application background,requirement analysis,design of system architecture and function implementation are discussed in details.In addition, some key technologies such as dynamic page layout rendering and data pooling,are also described. First of all,the ensemble forecast platform is designed using the client-server architecture.On the server side,there is a data processing program that converts large amounts of ensemble numerical model output into product data to ensure the performance of client data visualization program.On the client side, there is a data visualization program and a management console program.The data visualization program provides features including ensemble product data analysis,blending of multiple deterministic models,cus-tomized geographic information service,layer-based graphics rendering,interactive configuration of graph-ics layers,and exporting of weather maps.The management console program provides a unified user inter-face to help users manage all settings of the platform. As there is a large difference in computing resource throughout the meteorology department,the en-semble forecast platform is designed to be cross-platform by employing a modular and stratified design ap-proach with C++ is programming language. In order to enhance graphics rendering quality of weather maps,an innovative page layout rendering technique is proposed,allowing a flexible configuration of graphics elements like layers,titles and legends and creation of professionally-looking weather maps. As an effective abstraction of ensemble data may significantly improve working efficiency of forecaste-rs,advanced ensemble prediction algorithms and graphics rendering technologies are incorporated into the platform,which support nearly all popular products including statistics quantities,probability forecast, stamps,spaghetti,plume and box-whiskers. Since September 2013,the platform has been put into operation in central meteorological observatory and nearly all observatories of province capitals.Ever since,the platform is further improved by adding new features and fixing bugs based on user feedbacks. In the future,the platform will be improved by incorporating the latest algorithms in ensemble predic-tion,enhancing support for professional forecast such as typhoon track prediction and short-term strong weather prediction,introducing ensemble prediction validation and improving interactive performance. Furthermore,all the features of the platform will be incorporated into the fourth edition MICAPS,which will play an important role in flourishing China ensemble forecast applications and improving operational capability.