B. Schmid et al., COMPARISON OF MODELED AND EMPIRICAL APPROACHES FOR RETRIEVING COLUMNAR WATER-VAPOR FROM SOLAR TRANSMITTANCE MEASUREMENTS IN THE 0.94-MU-M REGION, JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES, 101(D5), 1996, pp. 9345-9358
Four atmospheric transmittance models, LOWTRAN 7, MODTRAN 3, FASCOD3P,
and the Thomason model, are investigated to quantify the relationship
between water vapor transmittance as function of water vapor amount,
T-W(U), for an instrument specific band pass in the 0.94-mu m region.
In a second step an empirical T-W(U) function is established using lon
g term measurements with our high-precision Sun photometer (SPM) in Be
rn, Switzerland along with 1300 simultaneous and collocated water vapo
r retrievals performed with a dual-channel microwave radiometer (MWR).
In order to avoid a possible bias in the empirical T-W, (Li) function
, the MWR data set is prescreened by comparing retrievals coincident w
ith radiosonde ascents. Over a 2 1/2-year period of common observation
s, radiosondes and SPM agreed to within 0.19 cm (13%) of columnar wate
r vapor (CWV) using the empirical T-W, (U) relationship. Completely in
dependent comparisons with an additional MWR and two Fourier transform
spectrometers yielded agreement within 13% and 9%, respectively. Comp
aring empirical and modeled results, we found that with respect to the
experimental data, LOWTRAN 7, MODTRAN 3, and FASCOD3P reported higher
water vapor transmittances over almost the entire range of realistic
absorber amounts. By relating these differences to differences in retr
ieved CWV for the case of two standard atmospheres, we found that usin
g T-W(U) predicted by LOWTRAN 7, MODTRAN 3, and FASCOD3P leads to an o
verestimate of CWV by about 18-30%, 7-20%, and 2-18%, respectively. Th
e Thomason model yields good agreement with respect to the experimenta
l data up to medium absorber amounts, whereas at slant path amounts la
rger than 10 cm, errors up to 60% in retrieved CWV occurred. We also s
how in this work that a misinterpretation of the LOWTRAN 7 water vapor
output counterbalances incorrectly predicted T-W, leading to results
that agree well with experimental ones.