Because radiosonde instrument changes cause data discontinuities, upper-air climate trends calculated from, or calibrated by, radiosonde data are controversial. Incomplete instrument metadata (a list of instruments used at each station) makes it difficult to avoid biases. A companion presentation in the 12th Symposium on Meteorological Observations and Instrumentation describes methods to infer complete metadata and determine instrument adjustments at over 1200 stations from 1973 to 2002. This presentation focuses on global precipitable water trends incorporating these preliminary instrument adjustments.

Because radiosonde data is sparse in many areas, especially over oceans, global average precipitable water is computed by filling in a 2.5-degree grid from available observations. First, climatological averages are determined for each month of the year, and then daily observations are expressed as anomalies from monthly means. Empty grid boxes are filled in by weighting the nearest observations, favoring a zonal direction. The exact grid filling method is not important for determining trends, but other methods produce less realistic spatial patterns, especially in the tropical eastern Pacific data void. Grids and time series averages can be produced both from adjusted and unadjusted radiosonde data.

An earlier examination of data from 1973 to 1996 produced an unadjusted global average of 2.55 cm of precipitable water, and an adjusted average of 2.51 cm. The average correction was about -4 percent in the 1970s, but near zero by 1996, so an unadjusted trend artificially favors drying. Precipitable water showed relatively steplike changes coinciding with documented global climate shifts around 1976-77 and 1988-89, so trends here discuss differences between three periods: 1973-76, 1978-87, and 1989-95. The late 1970s shift produced about 3.4 percent moistening (1978-87 minus 1973-76), while the late 1980s shift produced 0.8 percent drying (1989-95 minus 1978-87), or a total moistening of 2.6 percent (1989-95 minus 1973-76). Without instrument adjustments, the 2.6 percent moistening would have appeared to be 0.4 percent drying.

Other researchers have hypothesized a strong global moistening trend by studying known homogeneous stations, which are mostly in the Northern Hemisphere extratropics, but the adjusted trends show noticeable drying in the tropics and Southern Hemisphere extratropics since the late 1980s. At the conference, results should be available into late 2002, and it may be possible to determine if the climate regime starting 1989 has shifted to a different state, and if global average moistening has resumed.