The effect of horizontal photon transport within real-world clouds can be o
f consequence to remote sensing problems based on plane-parallel cloud mode
ls. In this paper, analytic approximations for the root-mean-square horizon
tal displacement of reflected and transmitted photons, relative to the inci
dent cloud-top location, are derived for plane-parallel cloud layers. With
anisotropic scattering, separate approximations are needed depending on the
order of scattering. When sufficient numbers of photon scatterings occur,
an approximation based on random walk theory (photon diffusion) is applicab
le; when scattering numbers are relatively small, a modification to the dif
fusion result is used. The resulting formulae are a function of the average
number of photon scatterings, as well as particle asymmetry parameter and
single scattering albedo. In turn, the average number of scatterings from p
lane-parallel, vertically inhomogeneous cloud layers can be determined from
efficient adding/doubling radiative transfer procedures. The transport app
roximations are applied to liquid water clouds for typical remote sensing s
olar spectral bands, involving both conservative and non-conservative scatt
ering. Results compare well with Monte Carlo calculations. Though the empha
sis is on horizontal photon transport in terrestrial clouds, the derived ap
proximations are applicable to general anisotropic, multiple scattering, pl
ane-parallel radiative transfer problems. Approximations useful for three-d
imensional transport are also given. The complete horizontal transport prob
ability distribution can be described with an analytic distribution specifi
ed by the root-mean-square and average radial displacement values. However,
it is shown empirically that the average displacement can be reasonably in
ferred from the root-mean-square value. An estimate for the horizontal tran
sport distribution can then be made from the root-mean-square photon displa
cement alone. (C) 2000 Elsevier Science Ltd. All rights reserved.