76
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
30 cm could penetrate through the plants (foliage etc.) and
reach to the objects underneath such as ground, and etc.
Because the soil moisture normally limits the penetration of
waves to the depths of a few centimetres, the surface wetness
conditions become apparent at longer wavelengths. The
penetration of L-band radar to the several meters provides the
observation of the moisture content under extremely dry soil
conditions. However in many studies, the potential of SAR data
for the retrieval of surface soil moisture was investigated not
only with longer wave lengths but also with shorter wave
lengths as C-band radars, and the microwave measurements
have shown their sensitivity to surface soil moisture (Ulaby et
al. 1978, Dobson and Ulaby, 1986, Dubois et al. 1995, Shi et al.
1997).
Following the evolution of SAR satellite technologies,
researchers have been investigated the effect of dielectric
features on backscattering. For instance, Peplinski et al. (1995)
investigated dielectric properties of different soil types and their
effects on backscattering. They discovered that soil texture and
volumetric moist content had an effect on the dielectric
coefficient. In addition, these researchers emphasized that
dielectric coefficients of soils vary with respect to clay types,
and as the clay specific surface area becomes wider dielectric
conductivities increase linearly. Romshoo et al. (2002) tried to
forecast soil moisture in Sukhothai area, using a time series of
space-borne ERS-2 SAR satellite data for the temporal
monitoring. In their study, it has been discovered that in the
study area, the backscatter coefficient of SAR data was
sensitive to volumetric soil moistures of 0-5 cm in depth. Shao
et al. (2003) have empirically investigated the variations in
dielectric features of moist and saline soils, the samples of
which were taken from a salt lake, and they observed an
increase in the backscatter as the saline content increased. Yang
et al (2006) demonstrated a technique to estimate the retrieval
of soil moisture change by using multi-temporal Radarsat
ScanSAR data. Their study had two parts. First part focused on
minimizing the effects of surface roughness by using two
microwave radar measurements with different incidence angles.
Second part dealt with to reduce the effects of vegetation cover
on radar measurements by using semi-empirical vegetation
model and measurements obtained from the sensors as Landsat
TM and AVHRR. Throughout the surveys in more than one
decade, it has been detected that there is a strong relation
between the backscatter coefficient and the soil moisture. The
researchers whether used data from only one sensor type (such
as ERS 1/2, RADRASAT-1, ENVISAT) to analyze the
sensitivity of SAR data to soil surface parameters at various
polarisations or incidence angles or they used two different
sensor data to make the comparisons (such as ERS 1/2 versus
RADARSAT-1 or ENVISAT versus RADARSAT-1) over
fields with different characteristics. (Baghdadi et al., 2002;
Baghdadi et al., 2006; Boisvert et al., 1997; Beaudoin et al.,
1990; Alvarez-Mazos et al., 2005; Holah et al., 2005; Kelly et
al., 2003; Oldak et al., 2003; Siegert and Ruecker, 2000;
Sahebi et al., 2003; Srivastava et al., 2003; Weimann et al, 1998;
Zribi et al., 2005a, 2005b). Researchers indicate that, the major
difficulties in retrieving soil moisture with SAR measurements
are due to the effects of surface roughness and vegetation cover.
The objective of the present study is to investigate the
behaviour of RADARSAT, ASAR and PALSAR images to
retrieve soil moistures for bare and just seeded soil. Besides,
tests have been carried out to obtain the cross correlation not
only between the different bands (C/ L) but also between the
different polarizations (VV/HH). This work will enable us to
perceive which sensor has the best potential for extracting soil
moisture in such an agricultural plain areas including the latest
satellite ALOS-PALSAR data.
2. IMPORTANCE OF SOIL CHARACTERISTICS IN
ACTIVE AND PASSIVE REMOTE SENSING
APPLICATIONS
In both passive microwave and active remote sensing, it is
important to know the soil characteristics. The soil is
constituted from 25 % air, 25% water, 45% and 5% inorganic
and organic substances respectively. Organic and inorganic
substances that are the solid parts of the soil form the structure
of the soil. Inorganic solid matter of soil is composed of various
rock decompositions and minerals in different sizes and
composition as well as rock pieces (Altinbas et al. 2004). The
texture of the soil is formed from various ratios of sand, silt,
and clay which are called inorganic substances. It is known that
diameters of particles range between 2 and 0,02 mm, of silt
particles between 0,02 and 0,002, and of clay particles which
have diameters smaller than 0,002 mm. In interpreting soil
reflection values for remote sensing applications, size of surface
soil particles, volume of the pores, ratio of the size of a pore,
and amount of water stored in these pores is very important.
The pores in sandy soil texture are called macrospores and the
pores in clay texture are called microspores. Although the size
of pores in sandy soil is large, total volume of pores is smaller
than that of clayey soil. In addition, reflection of
electromagnetic radiation from the surface of soil is dependent
on some features such as slope of the terrain, surface relief,
structure of the soil, organic matter content, size distribution of
the particles constituting the soil, stoniness, saltiness, iron oxide
content, and etc. Dielectric contents of the soil play important
role in microwave back scattering. Soil structure and moisture
are the main characteristics that determine the dielectric
contents of soils. For instance, while increased amount of sand
in soil enables the soil to become less absorbent and to have
low water holding capacity, increased amount of clay in soil
causes the soil to become more absorbent and to have more
water holding capacity. Increase in the amount of water causes
the dielectric content to be increased. Spectral characteristics of
the soil are mainly influenced by the organic matter content and
the moisture content (Stoner et al. 1980).
3. STUDY AREA
The study area is in the lands of Menemen (Izmir) Plain to the
west of Gediz Basin, and covers about 400 square km. The
Aegean Sea lies to the west of the study area, and Manisa
Province lies on the East. The area is also bordered by Bakircay
Basin on the North, and Izmir Bay on the South (Figure 1).