The approach allows to organize distributed storage of large amounts of diverse data in order to further their parallel processing in high performance cluster systems for problems of climatic processes analysis and forecasting. For different classes of data was used the practice of using meta descriptions - like formalism associated with certain categories of resources. Development of a metadata component was made based on an analysis of data of surface meteorological observations, atmosphere vertical sounding, atmosphere wind sounding, weather radar observing, observations from satellites and others. A common set of metadata components was formed for their general description. The structure and content of the main components of a generalized meta descriptions are presented in detail on the example of reporting meteorological observations from land and sea stations.