Along with the continuous development of the modernization level of meteorological service, meteorological data has been steadily on the increase, which results in a higher and higher demand of meteorological department for meteorological data storage, management, and read. Through analyzing the framework of Distributed File System, Data Warehouse Tool Hive of the open source cloud platform — Hadoop, the Build Process of the Hadoop meteorological cloud platform is studied. Based on Naïve Bayes algorithm research, a kind of method which can transplant it to the Hadoop platform is found.