This project currently focuses on the first goal. Building on available metadata sets, such as the Integrated Global Radiosonde Archive (IGRA) metadata file and the on-line WMO Catalogue of Radiosondes and Upper-Air Wind-Finding Systems, a very large amount of additional metadata has been found in sources such as journals, field experiment and technical reports, and web sites. So far, over 1900 radiosonde models and related upper air meteorological instruments, and over 4000 stations (or other platforms, such as ships) which have made upper air observations, have been documented.
Station metadata is of limited use for climate research if it does not specify the location, surface elevation, instrument type, and the beginning and ending date of each parameter. Most available metadata is a "snapshot," such as a report that a certain radiosonde model was in use at a certain time. In addition, the number of discrepancies increases as the number of sources increases. Therefore, this project examines the archived data, down to the level of individual observations, to validate available metadata, resolve discrepancies, and fill in missing metadata such as starting dates for each instrument.
To identify instrument types and pinpoint transitions as precisely as possible, it is best to focus on variables which are very sensitive to different instruments. The most sensitive variables are moisture-related, such as the lowest relative humidity reported per sounding, or the lowest pressure or temperature with a reported dew point. These variables are of little climate interest because reported values result from a mixture of inherent instrument characteristics (such as humidity sensor responsiveness), operating practices (such as not reporting the dew point when the temperature is below -40 C), and real climatic variations between stations and seasons. However, if the timing and nature of instrument changes at many stations is confidently established using any combination of variables and supporting information, then it is possible to develop temperature adjustments for each instrument type to a hypothetical common reference with a similar level of confidence, even if instrument-caused temperature differences are considerably smaller than natural short-term climate variations.
This presentation will report on monitoring the ongoing radiosonde network modernization efforts in China, India, and Russia, which are only partly documented in the metadata sources mentioned above. The new radiosondes cause significant drying in reported moisture, so their operational use at each station is easily detected. Accurate identification of the starting times of new instruments at each station should improve the assessment of the effects of these instruments on temperature trends.