数值预报产品释用业务系统历史数据存储与检索

Historical Data Storage and Retrieval in the Operational System for Application and Interpretation of Numerical Weather Forecasting Products

该文介绍了在国家气象中心中短期精细化数值预报产品释用业务系统中的历史数据存储结构的设计原理。基于该数据存储结构,依照用户的要求,可方便地选择区域大小,检索出某要素某段时间的数据,还可通过计算机编程实现国家级数值预报释用业务需求的历史数据处理和预报因子分析功能,表明新的历史数据存储结构简炼合理,灵活实用,能够极大地提高数值预报产品释用的工作效率。

The prediction products and weather observation information are the major processing objects of the application and interpretation of numerical prediction products, so the storage and processing of their historical data are very important throughout the operation procedure. In the past, National Meteorological Center adopts the Operational System of Meteorological Database as the interface of historical data storage and retrieval. But in the actual operation, this database＇s user interface is hard to operate and there are too many computing resources occupied. To simplify the operation procedure and enhance the efficiency of application and interpretation of numerical prediction products, the effective experience of database administration is taken as a reference, a series of data storage technology is adopted including index management, formless directly data storage, data dictionary and so on, therefore a theory of the design is put forward for new historical data storage structure. Based on this theory, the ground observation database and two-dimensional grid-point field （which is called the predictor） factor database of numerical forecasting products are established respectively. Each of them is composed of several files, and because of the organizational difference between the grid data and the site data, two types of files have different record structures. Based on the characteristics of the grid data, two-dimensional grid-point field factor database assumes that all the forecasting data of one physical quantity in selected period and level in a year are saved as a whole file, and based on the site data＇s organizational characteristics, the observation database assumes that all the ground observation data of one station in a year are saved as a whole file which is named by its station number. As an important component of the operational system for the application and interpretation of medium-short range fine numerical weather forecasting products which is adopted by the National Meteorological Center, its main function is to manage a greatnumber of meteorological data uniformly which is needed in the operation of application and interpretation, and to provide convenient and quick data interface for the platform for data processing. At the same time,this data storage structure is of very strong compatibility. It allows fully that the numerical forecasting patterns and meteorological observation factors increase and change with the development of the technology and can provide an environment to increase extensible databases such as satellite remote sensing data and radar data in the future. Based on this data storage structure and according to user＇s requirement, the region size can be conveniently chosen, the data of selected essential factors in the selected period can be retrieved, and functions such as historical data processing and forecast factors analysis are realized which satisfy the operational needs of the state-level application and interpretation of numerical weather forecasting products through computer programming as well. In a word, the new historical data storage structure is simple and practical and can largely enhance efficiency of the application and interpretation of numerical weather forecasting products.