支撑气象大数据实时预报业务应用的MICAPS4系统概述
MICAPS4 System Overview for Supporting Meteorological Big Data Real-Time Forecasting Business Application
中文摘要：
国家气象中心自2013年正式启动了MICAPS4系统的研发，MICAPS4体系采用了客户端/服务器的系统架构，一方面建立了基于大数据的气象实时预报应用技术体系，解决了气象大数据处理、存储、分析和显示效率的关键技术难题；另一方面搭建了多个满足不同复杂预报业务需求的专业化版本预报平台，基于MICAPS4基础框架的多个专业版客户端系统在中央台和部分省台进行应用推广。MICAPS4服务器系统利用分布式存储与分布式计算技术，可容纳上百TB气象实时数据，为数百用户提供并发数据访问服务，在国内率先实现全部气象实时数据由文件系统到数据库、从集中式系统到分布式系统的迁移。在海量气象数据和大量用户并发访问的环境下，表现出很高的稳定性和优越的读写性能，极大减轻了系统维护工作量，显著提升了数据解析及数据存储访问效率。
中文关键词：
MICAPS4，海量数据存储，智能网格预报，SWAN，台风和海洋预报
KeyWords：
MICAPS4, massive data storage, intelligent grid forecasting, SWAN, typhoon and ocean forecasting
Abstract：
The National Meteorological Center has officially launched MICAPS4 system since 2013. The MICAPS4 system uses a client/server system architecture. On the one hand, a real-time forecasting system based on Big Data technology to solve the key technical problems in processing, storage, analysis and display of meteorological big data is established; On the other hand, several specialized forecast platforms to meet the needs of different complex forecasting business is built. Based on MICAPS4 basic framework, a number of professional client have been applied to promote in NMC and part of the province. Utilizing distributed storage and distributed computing technology, the MICAPS4 server system can hold hundreds of TBs meteorological real-time data, providing data service for hundreds of concurrent users. MICAPS4 server system be the first to transits all meteorological real-time data storage from file system to database, from centralized system to distributed system in China. Under the condition of massive meteorological data and concurrent access of many users, MICAPS4 server system shows high stability and excellent read-write performance, it reduces the system maintenance workload greatly, and improves data analysis and data storage access efficiency significantly.