In: Wagner W., Sz£kely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, ¡APRS, Vol. XXXVIII, Part 7B 
Figure 1: Location of La Tejería experimental watershed 
mVmeas [vol%] 
50 r 
[dB] 
feb 27 mar 06 mar 23 mar 30 apr 02 
Date 
Figure 3: Normalized backscatter coefficients (cf) obtained at 
different acquisition dates in 2003 
40 
30 
20 
I 
10 
I 
i 
I 
feb 27 mar 06 
mar 23 mar 30 apr 02 
Date 
Figure 2: Soil moisture contents (mv mcas ) measured at different 
acquisition dates in 2003 
2.1 Study site and data 
The studied watershed, La Tejería, is situated in the north of Spain 
(Figure 1), has a humid, submediterranean climate and consists of 
clayey and silty clay loam textures. It is almost completely cul 
tivated, with an emerging cereal crop covering most of the fields 
during the experimental period (February - April 2003). A more 
detailed description of the study site is given by Alvarez-Mozos 
et al. (2006). 
For each acquisition day, field average soil moisture contents were 
calculated for fifteen seedbed fields based on measurements with 
a Time Domain Reflectometry (TDR) instrument with 11 cm 
probes (Figure 2). For a detailed description of the sampling 
method we refer to Álvarez-Mozos et al. (2006). 
Next, five C-band, HH polarized RADARS AT-1 SGF scenes were 
acquired over the experimental region during spring 2003, at low 
incidence angles (13°-29°). This configuration has proved to 
be particularly well suited for soil moisture research over ce 
real canopies (Ulaby et al., 1982b; Biftu and Gan, 1999; Mat- 
tia et al., 2003b). The images have a range resolution of 20 m or 
24 m and an azimuth resolution of 27 m, from which field average 
backscatter coefficients were calculated. Furthermore, in order to 
reduce the effect of the local incidence angle on the backscatter 
coefficients, these coefficients were normalized correspondent to 
a reference incidence angle, according to Lambert’s law for op 
tics (Ulaby et al., 1982b; Van Der Velde and Su, 2009): 
o 
0 
o-1 
COS 2 -#ref 
COS 2 9 
(1) 
ence incidence angle [°], in this case chosen to be 23° and 9 is the 
local incidence angle [°], The resulting field average backscatter 
values are shown for every acquisition date in Figure 3. 
2.2 Integral Eqation model 
The single scattering approximation of the Integral Equation 
Model (IEM) (Fung et al., 1992; Fung, 1994) is the most widely 
used scattering model for bare soil surfaces (Moran et al., 2004). 
It allows for the calculation of backscatter coefficients based on 
bare soil surface roughness parameters, soil dielectric constant, 
local incidence angle, wave polarisation and frequency. The IEM 
describes surface roughness by three complementary parameters: 
rms height (s), correlation length (l), and an autocorrelation func 
tion. Davidson et al. (2000) and Callens et al. (2006) demon 
strated that for smooth to medium rough agricultural bare fields 
this autocorrelation function is best represented by an exponential 
function. 
The conversion of the dielectric constant to the corresponding soil 
moisture content is performed by means of the four-component 
dielectric mixing model of Dobson et al. (1985), for which the 
residual and saturated soil moisture content used 
throughout this study are set to 3 vol% and 45 vol% respectively. 
It is expected that the emerging crops on the fields influence the 
results of the inversion of the IEM, since this was developed for 
bare soil conditions. However, the canopies were only weakly 
developed and the incidence angles were low, which are reasons 
to believe that the effect of the vegetation is minimal (Ulaby et 
al., 1982b; Maffia et al., 2003b). Furthermore, simulations by 
Lievens et al. (2010) using a water cloud model (Attema and 
Ulaby, 1978; Prévôt et al., 1993) indicated that the attenuation 
of the backscatter by the cereal canopy was to a large extent com 
pensated by a direct canopy contribution. This led to insignifi 
cant vegetation corrections within the relative radiometric accu 
racy of the RADARSAT observations, i.e. +/-1 dB (Srivastava et 
al., 1999). Therefore this study will not take into account a pos 
sible influence of the crop cover on the backscattered signal. 
2.3 Effective roughness 
The idea of using effective roughness parameters was first intro 
duced by Su et al. (1997). The effective roughness parameters are 
estimated using backscatter and soil moisture observations. They 
replace in situ measurements of soil surface roughness for the re 
trieval of soil moisture content from successive SAR images. 
where cr,° n is the linear normalized backscatter coefficient [-], of In case of the IEM, two effective roughness parameters need to 
is the linear measured backscatter coefficient [-], 6W is the refer- be defined: rms height (s) and correlation length (/). Lievens 
177