Quality control of pilot balloon data for climate monitoring

Special observations have been made since 1997 to monitor the variability of the wind field over the intertropical Americas with a project known as the Pan-American Climate Studies Sounding Network (PACS-SONET), supported by by the PACS program (now part of the the Climate Prediction Program for the Americas, CPPA) of NOAA's Office of Global Programs. These observations consist of once or twice-daily pilot balloon observations made at up to 20 locations, currently in 8 countries ranging from Mexico to Paraguay. Since the data base now consists of more than 40,000 observations and is sufficiently long to be starting to be useful for evaluations of short-term climate variations, the quality control of the data set is of special concern.

This presentation will describe the procedures we have developed to produce the research quality data set now being prepared for more widespread distribution. These steps include the use of special software at the observation site to reduce the inevitable observer errors, the plotting of the data and calculation of means to detect possible errors, and the comparison of nearby radiosonde observations and global analyses for detection of possible systematic errors. The frequency of erroneous data and the sources of these errors will be discussed, as well as their impact on the usefulness of the overall data set. The findings we describe can be applied to other pilot balloon data sets that continue to be generated in many parts of Asia and Africa.