417 
3.2 Introducing the canopy properties in MIMICS 
One of the main problems encountered when trying to use a theoretical model such as Mimics is that of 
specifying the numerous input parameters and variables. Many of them are difficult to determine and it is 
usually not possible to measure them during an extensive experiment such as the Agriscatt’88 campaign. 
for the leaves, LAI, thickness, water content and dry matter density are not mutually independent) some 
“reasonable” choices were made when the variables were not measured, insuring the consistency with the 
measured variables (LAI, fresh and dry biomass of the leaves and stems, canopy height). 
3.3 MIMICS results 
The comparison of MIMICS outputs and Erasme measurements is given in figure 2 for C-band HH and in 
figure 3 for X-band W, for each of the extreme incidence angles: 20° and 40°. A fairly good agreement is 
found in C-band. In X-band, Mimics strongly underestimates the radar signal at 40° for the higher levels 
of the biomass (day 169). The agreement is reasonable for the other days and for all days at 20°, although 
more dispersion is observed than in C-band. This discrepancy in X-band may be due to the limits of a first 
order radiative transfer model. 
4 Simplified expressions for MIMICS components 
Instead of directly fitting a semi-empirical model on the total backscattering coefficients calculated by Mimics, 
we first tried to obtain simplified expressions for the different components of the backscattered signal. This 
avoids compensation effects between parameters, that may appear when fitting a Semi-empirical model on the 
total backscatter. After a sensitivity analysis, this was done by fitting simplified “water-cloud” like models 
to the Mimics components. All the 36 vegetation data points of the Agriscatt’88 experiment were used 
here. 
4.1 Canopy transmittivity 
The two-way canopy transmittivity can be expressed as a function of the total canopy water content m„: 
presents the evolution of the canopy transmittivity with m v . As expected, the attenuation by the canopy 
increases with increasing frequency. The polarization has a strong influence on the variation of B with the 
The canopy properties used in Mimics are given in Table 1. The leaves and the stems are respectively 
represented by fiat, disks and vertical cylinders. Keeping in mind the existing dependencies (as an example, 
< 2 = exp(-2J3m„/cos0) 
(4) 
where B is a unique parameter fitted over the 36 data points. Table 2 gives the B values obtained and figure 4 
incidence angle: in HH polarization, B is almost constant between 20° and 40°, whereas it strongly varies in 
VV polarization. This is probably due to the vertical structure of a winter canopy, as pointed out by [11]. 
The B values obtained here are in excellent agreement with those given by [12]. 
4.2 Canopy direct backscatter 
The direct backscatter component can be simplified by: 
where L is the LAI, t 2 is given by eq. (4) and A is a unique parameter fitted over the 36 data points. Table 3 
gives the A values obtained and figure 5 shows the fitting of eq. (5). The A parameter appears to be almost 
constant with the incidence angle. As expected, it strongly increases with increasing frequency. 
5 Coupling a simplified model to MIMICS 
The previous part showed that in C-band, HH polarization, the direct backscatter term is negligible and the 
ominant term is the attenuated soil direct backscatter. Thus, the following semi-empirical “water-cloud” 
model is proposed as a simplified form for Mimics: 
(6)