C. Kopken et al., ASSESSMENT OF THE QUALITY OF TOVS RETRIEVALS OBTAINED WITH THE 3I ALGORITHM FOR ANTARCTIC CONDITIONS, JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES, 100(D3), 1995, pp. 5143-5158
The quality of layer-mean temperatures and humidities retrieved with t
he improved initialization inversion (31) (version 2) algorithm from m
easurements of the TIROS operational vertical sounder (TOVS) radiomete
r has been studied for two summer periods in 1987 and 1990 in the Wedd
ell Sea region (Antarctica) by means of collocations with radiosoundin
gs, European Centre for Medium-Range Weather Forecasts (ECMWF) analyse
s and special sensor microwave / imager (SSM/I) retrievals. Appropriat
e corrections of the atmospheric transmission functions have been used
for the two periods. A significant influence on the accuracy of the r
etrievals has been observed. Collocation statistics with radiosounding
s for layer-mean temperatures based on 160 cases show very satisfying
results. The standard deviation is very close to the deviations expect
ed from collocation errors alone, but an overall negative bias of the
retrieved temperatures is observed, Considerably smaller biases in cle
ar conditions suggest that the frequent cloudy situations are one reas
on. The atmospheric stability index derived from the retrievals corres
ponds well to the observations. However, surface inversions and invers
ions capping the boundary layer (being characteristics of the polar en
vironment) cannot be resolved by the TOVS. Layer-mean humidities and i
ntegrated water vapor content are calculated by the 31 algorithm with
a good absolute accuracy and errors are comparable to those occurring
at lower latitudes. But problems have been found in the ability to cor
rectly retrieve varying water vapor co?tents in different situations w
hich is caused by weak signals in the very dry Antarctic conditions. C
omparisons of the total water vapor content to results based on SSM/I
measurements show a good agreement between the two different instrumen
ts and retrieval methods. Comparisons of temperature and humidity retr
ievals to ECMWF analyses show a better correspondence between retrieva
ls and analyses for 1990, related probably to model and assimilation s
ystem changes. The ECMWF analyses show deviations from radiosoundings
comparable to those of the retrievals emphasizing the value of TOVS re
trievals as an important source of information in this data-sparse reg
ion.