FIRST GEOMETRICAL PATHLENGTHS PROBABILITY DENSITY-FUNCTION DERIVATIONOF THE SKYLIGHT FROM SPECTROSCOPICALLY HIGHLY RESOLVING OXYGEN A-BANDOBSERVATIONS - 1 - MEASUREMENT TECHNIQUE, ATMOSPHERIC OBSERVATIONS AND MODEL-CALCULATIONS
K. Pfeilsticker et al., FIRST GEOMETRICAL PATHLENGTHS PROBABILITY DENSITY-FUNCTION DERIVATIONOF THE SKYLIGHT FROM SPECTROSCOPICALLY HIGHLY RESOLVING OXYGEN A-BANDOBSERVATIONS - 1 - MEASUREMENT TECHNIQUE, ATMOSPHERIC OBSERVATIONS AND MODEL-CALCULATIONS, J GEO RES-A, 103(D10), 1998, pp. 11483-11504
First measurements of the probability density function of geometrical
pathlengths (PDF-GP) for the skylight transmitted from clear and cloud
y skies to the terrestrial surface are reported. The measurements are
performed using a novel technique, which includes spectroscopically hi
ghly resolving observations of the O-2 A-band (760-780 nm) and a Lapla
cian back transformation of the intensity ratios measured for the set
of individual O-2 A-band rotational absorption lines. The technique pr
ovides a new and powerful tool to study the radiative transfer (RT) of
the atmosphere for different kinds of atmospheric aerosol loadings an
d cloudiness. First measurements show, as expected, that the photons t
ransmitted by clouds to ground generally experience longer geometrical
paths inside the cloud than expected for a corresponding clear sky co
ndition. The comparison of the measured oxygen A-band absorption syste
m with plane-parallel discrete ordinate radiative transfer (DISORT) mo
del calculations shows a reasonable agreement for direct Sun and clear
sky zenith observations. For cloudy sky the same comparison shows a g
ood agreement between the observed and the modeled weak absorption Lin
es but systematically larger observed than modeled optical densities f
or strong absorption lines. Our data also suggest that this significan
t discrepancy is most likely caused by shortcomings in the physical de
scription of the RT by nonstatistical models, which do not properly ac
count for the fractal nature of the terrestrial cloud cover. In partic
ular our study reveals that the photon path statistics involved with t
he multiple Mie scattering inside clouds is not properly modeled using
the DISORT code. Closing this gap in our understanding of the cloud s
ky RT may possibly provide much of the explanation for the frequently
lower modeled than measured cloudy sky absorption of the solar radiati
on.