Qy. Han et al., The effects of aspect ratio and surface roughness on satellite retrievals of ice-cloud properties, J QUAN SPEC, 63(2-6), 1999, pp. 559-583
Citations number
21
Categorie Soggetti
Spectroscopy /Instrumentation/Analytical Sciences
Journal title
JOURNAL OF QUANTITATIVE SPECTROSCOPY & RADIATIVE TRANSFER
This study investigates the effects of non-sphericity on satellite retrieva
ls of ice-cloud properties including optical thickness and particle sizes.
Ray-tracing technique is used to calculate single scattering phase function
and single scattering albedo for both smooth and rough surfaces of hexagon
al columns and plates at visible and near-infrared wavelengths. Two paramet
ers, aspect ratio and distortion parameter, are used to simulate different,
randomly oriented, ice crystal shapes and surface roughnesses in the ray-t
racing process. A wide range of aspect ratio and distortion parameter is ex
plored in the calculations. The resultant phase functions and single scatte
ring albedos are used to compute bidirectional reflection functions by a ra
diative transfer model with adding-doubling technique. The results show tha
t in a direct backscattering regima (Theta > 150 degrees), if no informatio
n of particle shape is available, the uncertainties in the retrieved optica
l thickness (a factor of more than ten) would make the retrieval meaningles
s. For images with viewing geometry outside of this region, the typical ran
ge of uncertainty of retrieved optical thickness is less than a factor of a
bout two. Using averaged phase function will cut this uncertainty in half.
That is, the uncertainty in the retrieved optical thickness is about 40%. S
ensitivity tests show that aspect ratios are critical in reducing the uncer
tainties of the retrieved optical thickness using satellite data. The uncer
tainties in the retrieved ice particle sizes are also estimated in a simila
r way. It is found that using an averaged aspect ratio and roughness, the u
ncertainty of retrieved particle size is about 30% for small particles and
10% for large particles at lambda = 3.7 mu m. The corresponding uncertainti
es are about 70% for small particles and 35% for large particles if data at
lambda = 1.6 or 2.2 mu m are used in the retrieval. However, at lambda = 1
.6 and 2.2 mu m, retrieval of particle size may encounter even larger uncer
tainties for thin ice-clouds ( tau < 3). (C) 1999 Elsevier Science Ltd. All
rights reserved.