Abstract

Over the last 45 years, the Computational and Information Systems Laboratory of NCAR has collected a large upper-air data archive. It contains over 40 million unique observations covering the period 1920 to the present. Most observations are from the global land station network, however, there ar... Show moreOver the last 45 years, the Computational and Information Systems Laboratory of NCAR has collected a large upper-air data archive. It contains over 40 million unique observations covering the period 1920 to the present. Most observations are from the global land station network, however, there are also observations from permanent and moving ships. This archive has served as input for several global atmospheric reanalysis projects. Over the years NCAR has received data from over 30 different sources, including individual countries and various institutions. As such, the original data are in many different formats with varying station metadata, measurement units, and quality control procedures. To make this collection more useful it has been organized in a relational database that allows easy access in multiple ways. The station metadata have been vastly improved through careful analysis and comparison of various existing station libraries, and computed versus reported station surface elevations. The data quality tests have been made using a robust estimation of location (mean) and scale (standard deviation) based on the Huber weighting function, and the measurement units have been homogenized. This upper-air data archive is not static, but will continue to grow with monthly updates and new unique datasets as they become available. These long-term historical data are part of NCAR’s Research Data Archive. Public access to the data will be through a web interface where users can select data by station, source, time, and geographic domain. The current output format is simple ASCII text that is sorted by station or chronologically. The current status, capability, and advantages of managing diverse upper-air data in this manner will be discussed. The public opening of the archive is planned for January 2011. Show less