Determination of Parameters of the Atmosphere and the Surface in a Clear Atmosphere - Short-Wave Solar Radiation in the Earth's Atmosphere - page 198

evident in Fig. 5.8 and could be explained with the atmospheric aerosols
presence at this altitude. It corresponds to the most probable altitude of the
cloud formation of the lower level.
The examples of the spectral dependence of the volume aerosol scattering
and absorption coefficients are demonstrated in Fig. 5.9. We should mention
the essentially different character of the spectral dependence of the scattering
coefficient inquestion above the desert and above Ladoga Lake. In the first case,
there is no spectral dependence of the scattering coefficient or there is a weak
growth with wavelength. It might be explained by the rather high amount of
large particles in the atmospheric aerosols above the desert.
Figure 5.9 illustrates the results of the volume coefficient of the aerosol
absorption obtained from the sounding data above the desert under pure
atmospheric conditions (the weak absorption) and after a sand storm (the
strong absorption). The latter case demonstrates the evident absorption band
of the hematite, which appears even in the spectra of the solar radiative flux
divergence (Sect. 3.3). The second case illustrates the apparent decreasing of
the aerosol scattering coefficient with wavelength.
The examples of retrieving the spectral values of the surface albedo are pre-
sented in Fig. 5.10. The deviation of the spectrum of the snow surface from the
monotonic behavior (Fig. 5.10b) is likely caused by the surface inhomogeneity
(the snow was melting on 29th April). The surface inhomogeneity has been
smoothed during the second stage of the data processing, but the spectral
distortions of the upwelling irradiances have remained and they cause the sys-
tematic uncertainty of the retrieved albedo, which does not exceed the interval
of three SD and statistically can be assumed the insignificant one.
Note, that the spectral albedo is retrievedwith the relative uncertainty about
1-3% that is much more accurate than in the case of direct dividing the up-
welling irradiance by the downwelling (Sect. 3.4). In addition, the retrieved
albedo is exactly correspondent to the notion of albedo used in the radia-
tive transfer theory (Sect. 1.4) and it has no distortion connected with the
gases absorption bands. Thus, the airborne experiments accomplished with
thesoundingschemeandthefollowinginverseproblemsolvingcouldbeused
for obtaining the surface albedo values with high accuracy.
We should mention that the uncertainties of the retrieved atmospheric pa-
rameters are greatly affected by the information content of the results of solar
irradiance observations (Sect. 3.3) and by the spectral resolution within the
absorption bands of gases, while retrieving their content. Thus, the uncertain-
ties of the retrieval from different soundings data are essentially different. It is
seen from the presented figures, where the posterior SDs of the retrieved pa-
rameters are shown. The uncertainties of the retrieval in the lower troposphere
are 10-50% on the average for the volume aerosol scattering coefficient; are
50-100% for the volume aerosol absorption coefficient (that, however, is less
than the a priori uncertainty); are 20-30% for ozone content and are 20-50%
for H 2 O content. We should point out that the higher the aerosol content in the
atmosphere the higher the accuracy of the aerosol parameters is.
The discreteness of the registration during the measurements (Sect. 4.3)
is not accounted for in the formal scheme of the inverse problem solving