Tal Svoray 
  
  
voray et | curves and the herbaceous formations present changes in backscatter along the growing season. Figure 4 is illustrating 
ovides a | the variations in backscatter from annual vegetation canopies at February, April and May 1997. 
kg m*] | 
and the 
roposed Figure 4 shows that the mean backscatter from 
herbaceous vegetation and crops is very similar at 
February and May. On the contrary, a significant 
difference (00.05) was found between the backscatter 
from the canopies of these two formations at April. 
Moreover, figure 4 shows that the backscatter from the 
two formations in February is similar to the backscatter 
from the two formations in May and the backscatter at 
these two months differs significantly from the 
backscatter at April. This two observations which to our 
assessment are related to the same cause may have the February April May 
following explanation: on February when the natural 
Backscatter [dB] 
  
herbaceous vegetation and crops (wheat) are still low, D 
and on May when the herbaceous vegetation is withered Figure 4. The relationship between the ERS-2 SAR 
and the wheat is harvested, a relatively high backscatter backscatter and plots of natural herbaceous 
is recorded in the ERS-2 SAR antenna. This may occur vegetation and crops. The points represent mean 
due to the strong effect of soil at these two phenological values and confidence intervals (00.05) of 50 
phases of the annuals. On April when the herbaceous 
vegetation is grown, it presents a high level of biomass 
and the contribution of lower constituents (soil moisture 
and soil roughness) is masked and thus the mean 
backscattering values are relatively low. This result may 
points to each of the formations per date. 
  
tric soil imply that the contribution of the soil to the total 
tions in backscatter is much stronger than the contribution of the m 
lots are vegetation canopy. Thus in April when the effect of the 3 
vegetation prevails, the backscatter from the different = 
canopies differs, but when the soil contribution prevails £ 
1 strong the soil underneath the two formations presents similar d 
proven backscatter. This phenomenon has a strong influence on 
noisture radar modeling of annuals and it should be considered | 
ld, have when trying to determine quantitative biophysical February April May 
attributes of from variations in backscatter. On the Date 
contrary to the case of the herbaceous vegetation, the 
ERS-2 SAR backscatter, recorded from homogenous Figure 5. The ERS-2 SAR backscatter from plots of 
plots of woody vegetation formations (forests, woody formations. The points represent mean values 
shrublands and dwarf shrublands) appeared as relatively of 50 points to each of the formations per date. 
stable along the growing season. 
Thus the backscattering from these formations did not vary significantly (0=0.05) between the three dates of 
observation. This result is very well illustrated in figure 5. The backscattering curves of shrubs and dwarf shrubs are 
very similar and the curve of the trees keeps the same trend but with much higher values of backscatter. The reason 
for this may be the much more dense layer of the trees at the upper part of the canopy than the upper part layer of the 
lower formations. This result is in line with an earlier observation of Quegan et al (1998) who measured the 
multi-temporal backscatter from the King’s forest in the U. K. their results indicate that mature pine trees perform a 
relatively stable (between —8 and —12.5) backscatter curve between January and December 1997. The explanation for 
the stability of the backscatter from the woody formations along the vegetative season may be the relative stability of 
the moisture content in the canopies of these formations along the year, especially when it is in comparison to the 
water content in the herbaceous vegetation. 
  
40 
> ERS-2 
ric soil The observation of the strong effect of soil moisture on the herbaceous vegetation but not on the woody formations 
have lead us to the understanding that a combined model should be developed and include the following rule: the 
areas covered by the herbaceous vegetation should include the contribution of soil moisture and the areas covered by 
Woody formations should include only the effect of the upper part of the canopy. The difference between the trees and 
t (trees, shrubs / dwarf shrubs may imply on the effect of leaf density at the upper part of the canopy layer on backscatter. At 
rves of the first stage, the water cloud model was applied to areas of homogenous herbaceous vegetation cover. The use of an 
baceous inversion model enabled to predict herbaceous vegetation biomass from the ERS-2 SAR backscatter. Comparison of 
emporal measured and predicted herbaceous green biomass is illustrated in figure 6. 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part Bl. Amsterdam 2000. 325