Aj. Gasiewski et Jt. Johnson, STATISTICAL TEMPERATURE PROFILE RETRIEVALS IN CLEAR-AIR USING PASSIVE118-GHZ O-2 OBSERVATIONS, IEEE transactions on geoscience and remote sensing, 31(1), 1993, pp. 106-115
Linear statistical temperature profile retrievals from nadiral passive
118-GHZ O2 spectra are demonstrated using time- and space-coincident
microwave observations and radiosonde profiles. The nadiral aircraft-b
ased brightness observations were made using the Millimeter-wave Tempe
rature Sounder (MTS) instrument aboard the NASA ER-2 high-altitude pla
tform during the Genesis of Atlantic Lows Experiment (GALE) and the Co
operative Huntsville Meteorological Experiment (COHMEX). Separate retr
ievals are demonstrated for winter and summer midlatitude conditions i
n clear-air over land; observations during both day and night are incl
uded. The retrieval operator is a linear-statistical minimum mean-squa
red-error estimator. A purely statistical retrieval operator circumven
ts the effects of radiative transfer model uncertainties and biases in
either the radiosonde or MTS data. The retrieved profile root-mean-sq
uare errors are approximately 0.7 - 1.2 K for either the winter or sum
mer tropospheres. These are substantially smaller than the a priori te
mperature profile variations, demonstrating that 118-GHz aircraft or s
atellite observations can provide useful information on atmospheric ve
rtical thermal structure. However, classification of retrieval operato
rs by prevailing meteorological conditions appears to be important for
operational purposes. Correlations between MTS antenna temperatures a
nd radiosonde kinetic temperatures are nearly consistent with calculat
ions based on a radiative transfer model, indicating both the validity
of the model and the sensitivity of the MTS channels to atmospheric t
emperature. The predicted and observed correlations exhibit the same g
eneral trends, although the predicted values are usually slightly grea
ter.