Two different methods for retrieving upper tropospheric humidity (UTH) from
the TIROS Operational Vertical Sounder (TOVS) instruments aboard NOAA pola
r orbiting satellites are presented and compared. The first one, from the E
nvironmental Technology Laboratory, computed by J. J. Bates and D. L. Jacks
on, estimates UTH from a simplified radiative transfer analysis of the uppe
r tropospheric infrared water vapor channel at 6.7 mum wavelength measured
by High-resolution Infrared Radiation Sounder (HIRS). The second one result
s from a neural network analysis of the TOVS (HIRS and Microwave Sounding U
nit (MSU)) data developed at the Laboratoire de Meteorologie Dynamique. Alt
hough the two methods give very similar retrievals in temperate regions (30
degrees -60 degreesN and S), the latter is larger by up to 16% in the trop
ics. The two data sets have also been compared with the UTH retrievals from
infrared radiance measurements at 6.3 mum wavelength from the geostationar
y satellite Meteosat. These products are taken from the archive without any
reprocessing that would take care of known biases. Since the Meteosat UTH
in 1989 was confined to clear-sky areas, it has a dry bias. The differences
observed among the three data sets can be explained. UTH computation is se
nsitive to assumed air temperature and humidity profiles. Despite the biase
s the spatial and temporal correlations are very good. Overall, the compari
son of the two TOVS retrievals provides an assessment of the UTH uncertaint
ies, about 15-25% (relative). With regard to the Meteosat UTH it is conclud
ed that the archived product performs well in depicting spatial and tempora
l changes. For future quantitative analyses, a reprocessing of the Meteosat
UTH is suggested.