International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B1. Istanbul 2004 
  
  
  
  
  
f 0 a n o Q 
$ 5 | 
= | 
= | XHH 
$ -10 1 % m9 mu m" i 
= muv 
Es xau 
= OVV 
= 204 m 
= 25 
E 
= -30 
0 0.5 1 IS 2 25 3 
rms surface height (cm) 
5 -— - — we 
S 
uz 0 x x X x 
= | BHV/HH 
È i 
g -5 i OHV/VV 
s 8 | 
s . 99 X VV/HH 
= -10 8 
-15 |^ 10 er] 
0 0.5 1 1.5 2 2:3 3 
rms surface height (cm) 
1.0 E = 
= 0.8 | 
= 
Se | HV.HH 
SE 0.6 | « 
= 2 |OHV.VV 
26 | 
= 2 04 [X VV.HH 
= 5 | : 
& 02 o 2 0 a 
z x xx x 
0.0 + ——— —— - 
0 0.5 | 1.5 2 2.5 3 
rms surface height (cm) 
1.0 y : 
| 
0.8 i 
= 06: | 
2 00 a m mg a | 
= 04 = | 
0.2 
0.0 
0 0.5 1 1.5 2 2.5 3 
rms surface height (cm) 
90 — ry 
80 i 
70 
© 60 
= 30 
cr 
E 40 
3 30 S s m ! 
20 a | 
10 
0 
0 0.5 1 15 2 2:5 3 
rms surface height (em) 
1.0 
| 
0.8 | 
=" | 
= 0.6 | 
Ë | 
e | 
Z 04 m 
5 = s m m 
0.2 
0.0 
0 0.5 1 1.5 2 2.5 3 
rms surface height (cm) 
Figure 4. Variation of backscattering coefficients, 
copolarization and depolarization ratios, correlation coefficients 
and polarimetric parameters (entropy, c -angle and anisotropy) 
as a function of rms surface height. 
Our results show also that the radar signal remains virtually 
constant when the rms surface height increases from 0.85cm to 
2.4em (R1 to R4). We observe that the copolarization ratio 
reduces significantly the dependence of the radar signal on the 
roughness parameter. For the depolarization ratios, we observe 
a very weak correlation between the radar signal and the 
roughness parameter. As for the correlation coefficients, these 
were found to be much less dependent on the roughness 
parameter. Discrimination between the bare soils surface 
roughness in the H/@ and H/A planes is not possible (cf. 
Figures 3 and 4). However, the parameters H and @ increase 
slightly with the rms surface height. As for the anisotropy, we 
observe the opposite behaviour, this parameter decreases with 
rms surface height. In conclusion, the polarimetric parameters 
do not provide discrimination in X-band at low incidence angle 
between the roughness states. However, it has been observed 
(Baghdadi et al., 2002) that the influence of soil roughness on 
radar return at C-band is higher at high incidence angles (47?) 
than at low incidence angles (23? and 40?). This suggests 
further analysis of X-band data acquired at a high incidence 
angle is necessary. 
5. CONCLUSIONS 
The potential of polarimetric SAR data at X-band has been 
tested with data acquired with the ONERA's RAMSES system 
over an area of Avignon in Southern France. The polarimetric 
measurements provide a more complete description of targets 
than is possible with a single-channel radar system. Results 
obtained using the  backscattering coefficients and the 
polarimetric ^ parameters calculated from the eigen 
decompostion of the coherency matrix show moderate 
discrimination between classes at X-band. Certain classes, 
wheat, lawn and orchards for example, are difficult to classify. 
The radar signal at X-band acquired at an incidence angle of 
around 26? is weakly correlated to the surface roughness over 
bare soils. However, it is possible to observe the surface 
degradation due to the slaking process and to distinguish the 
freshly tilled fields. 
The entropy and a -angle plane indicate clearly that the class 
signatures are grouped in five main clusters. The introduction of 
the anisotropy parameter does not allow discrimination between 
different classes whose cluster centres are in the same zone of 
the entropy/ a plane. 
These results suggests that for a more complete analysis, X- 
band data at a high incidence angle must be studied. 
References 
Baghdadi N., Bernier M., Gauthier R., and Neeson 1., 2000. 
Evaluation of C-band SAR data for wetlands mapping. 
International Journal of Remote Sensing, vol. 22, no. 1, pp. 71- 
88. 
Baghdadi N., King C., Bourguignon A., and Remond A., 2002. 
Potential of ERS and RADARSAT data for surface roughness 
monitoring over bare agricultural fields: application to 
catchments in Northern France. International Journal of 
Remote Sensing, vol. 23, no. 17. pp. 3427-3442. 
Cloude S.R. and Pottier E. 1996. A review of target 
decomposition theorems in radar polarimetry. [EEE 
Transactions on Geoscience and Remote Sensing, vol. 34, no. 2, 
pp. 498-518. 
  
     
   
  
   
   
  
    
     
    
    
   
  
  
   
    
   
   
  
  
  
   
  
  
   
   
   
   
   
  
   
   
  
   
  
   
  
  
  
   
  
    
Internation 
Cloude 5. 
classificatic 
IEEE Tran: 
no. |, pp. 6 
Ferro-Fami 
SAR 
Télécommu 
Hajnsek I, 
1999. Soil 
Proceeding 
Le Toan T 
A., Nagid 
monitoring 
pp. 883-88. 
Scheuchl 1 
Automated 
SAR data 
Australia. 
Shi J., Do: 
Regions w 
Geoscience 
Acknowle 
The author 
the acquis 
was Suppc 
ONERA ( 
Centre for 
(French N 
(National |