The longest-running atmospheric sounding data base is that from pilot balloon observations (“pibals”), whose routine observations date from the early 20th century. Such observations are still made today, though mostly in developing countries in Asia and Africa. Although pibals are subject to limitations (cloudiness and darkness being the two most obvious ones), their low cost has historically allowed for more frequent and more widespread use than radiosondes. Much less well-known, and less appreciated, is that pibals have some advantages over radiosonde winds for studies of climate variability. The procedure for making a pibal has not fundamentally changed in 100 years, unlike wind observations obtained from rawinsonde systems – which have used radiotheodolites, Omega, LORAN and now GPS. Wind profiles obtained from these different systems show somewhat differing characteristics, which can complicate identifying historical trends in wind data.

Making effective use of historical pilot balloon data for climate studies requires availability of the original angle data and some metadata - such as the characteristics of the balloons and inflation procedures. We discuss the difficulty in obtaining both of these; a major effort at digitizing the global data base of pilot balloon observations will be required. Studies are underway (results will be shown) to evaluate the uncertainties introduced into mean wind profiles due to missing data (generally due to clouds), and the observational errors that characterize different pibal networks. These need to be known to determine the limitations of pibals for describing and monitoring regional climate variations.