1684
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 42, NO. 8, AUGUST 2004
Fig. 2. (Left) Outcrops of paleosoils and (right) cross section of the internal structure of the Pyla dune.
Soil moisture monitoring is a key challenge for several envi-
ronmental studies: 1) water resources management in arid and
semiarid regions; 2) water saturation and infiltration in soils for
flood risk management; 3) biomass evaluation for the global
monitoring of the carbon cycle.
In order to study the radar phase variation due to buried wet
layers, some experiments were conducted over a test site lo-
cated in France: the Pyla dune close to Bordeaux. It is a nat-
ural sand dune that contains subsurface paleosoils (fossil vege-
tation), which act as moisture tanks at depths from 0.5110 m.
Airborne L-band (1.6 GHz) SAR images of the Pyla dune re-
vealed a phase signal correlated to the buried wet structures: a
phase difference reaching 23 between the horizontal (HH) and
vertical (VV) channels was clearly observed. It has been shown
that this phase signature allows detection of the paleosoil down
to 5.2 m, which is almost 2 m deeper than when only consid-
ering HH and HV amplitude signals [14]. Furthermore, field
campaigns were performed on the Pyla dune, which led to the
The objective of this study was to model the capabilities of
L-band SAR systems to penetrate soils to retrieve subsurface
Fig. 3. (a) Aerial photography of the Pyla dune (source IGN), and RAMSES
moisture. For that purpose, we propose a two-layer integral
L-band image of the dune for (b) HH and (c) HV polarization. Arrows indicate
equation method (IEM) model that reproduces the phase dif-
the location of the wet paleosoil (bright feature in the white ellipse) appearing
through the sand.
ference between HH and VV channels and physically explains
the observed phenomenon.
First, we briefly describe the polarimetric analysis of the
a high signal penetration. The Pyla dune contains large wet sub-
airborne SAR images and present the field measurements that
surface structures called paleosoils (fossil vegetation, cf. Fig. 2),
confirmed the phase signature of the buried paleosoil. We then
which act as moisture tanks at varying depths (0.5110 m).
present results obtained from the semiempirical model of Oh
These buried paleosoils are detectable using radar because of
and Sarabandi applied to a two-interface problem representing
the high permittivity due to their water content.
the geometry of the Pyla dune. Next, we describe a two-layer
IEM scattering model that reproduces the phase difference
III. SAR POLARIMETRIC ANALYSIS
variation as a function of the thickness of the overlying sand
layer. A more accurate geometrical and electrical description
A. RAMSES Airborne L-Band SAR
of the Pyla dune is then used as input to a finite-difference
RAMSES is a French airborne SAR operated by the National
time-domain (FDTD) electromagnetic code in order to improve
Office for Aerospace Studies and Research (ONERA) [25]. It
the two-layer IEM model.
provides a wide range of frequencies and is fully polarimetric.
The spatial resolution for L-band (1.6 GHz) is 0.7 m in range
II. PYLA DUNE
and 0.9 m in azimuth. Fig. 3 shows a RAMSES SAR image
Results presented here concern the southern part of the Pyla
of the Pyla dune that was acquired in June 1998, with a mean
sand dune (cf. Fig. 1), which is a large sandy area (2.5 0.5 km)
incidence angle of 55 (that is a local incidence on the dune of
40 ) and a heading of 20 N. It allows unambiguous detection of
located on the edge of the Arcachon Basin close to Bordeaux
subsurface structures in the southern part of the dune [bottom of
(France). This area is particularly suitable for radar sounding be-
Fig. 3(c)].
cause it is mainly composed of low loss sandy material allowing