1686
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 42, NO. 8, AUGUST 2004
with respect to
Fig. 7. Computation of
for various incidence
angles, after Oh et al. [16].
Fig. 6. (Top)
and (bottom) HH, VV correlation measured in the
field along a profile crossing the studied paleosoil.
hand. Moreover, Sarabandi [18] has shown that the probability
density function (pdf) of the copolarized phase difference could
be characterized by two parameters: the degree of correlation
, which is a measure of the width of the pdf, and the polarized
phase difference , which is the value of
at the
maximum of the pdf. Under this assumption, Oh et al. [16]
proposed an empirical expression for and of the following
Fig. 8. Geometry of the two-layer scattering problem.
form:
(1)
sented in Fig. 7 show that for a RAMSES local incidence angle
(2)
of 40 at the dune's surface, with a measured relative permit-
tivity for the wet paleosoil of 8.07 and an estimated paleosoil's
with
rms roughness height of 1 cm, we obtain a copolarized phase
difference of 22 , close to the phase difference observed on the
(3)
RAMSES L-band images of the Pyla dune.
(4)
V. TWO-LAYER IEM MODEL
where
is the Fresnel reflection coefficient at normal inci-
dence
In the section above, we applied the semiempirical model to
a two-layer problem that represents the Pyla dune geometry. In
order to understand the physical phenomenon involved here, we
(5)
now consider an analytical approach based on the IEM proposed
by Fung [26] to reproduce the phase difference between HH and
is the rms roughness height of the wet layer, is the local
VV signals.
is the wavenumber.
incidence angle in radians, and
Using the and parameters, we could model the
A. Schematic Description
distribution with respect to both surface parameters (i.e., relative
dielectric constants and rms height ) and radar parameters
Fig. 8 shows the two-layer geometry of the problem. The
and roughness characteristics
(i.e., frequency and incidence angle). Fig. 7 presents the com-
paleosoil of permittivity
with respect to for various incidence angles
(Fourier transform of the surface autocorrelation function),
puted
at 1.6 GHz. It reproduces the copolarized phase difference for a
(rms height), and
(correlation length) is covered by a sand
, i.e., 10% water content) covered by a
layer of thickness , albedo , extinction coefficient , and
wet paleosoil (
. Assuming a smooth airdune
permittivity . The sand layer surface is described by its rough-
low loss dry sand layer
, , and . The incident wave of wave-
interface, backscattering is mainly generated by the buried pa-
ness parameters
length illuminates the dune's surface at an incidence angle
leosoil layer that corresponds to the moisture front. Results pre-