exposed previously (Figure 4 and tablel). We still observe for most plots except for the white backgrounds an
underestimation of LAI that compensates for a general overestimation of leaf chlorohyll or water contents.
Figure 6 and table 1 show that the estimates of the canopy chlorophyll and water contents are even slightly
better with the 6 TM broad bands as compared to the use of the 188 AV1RIS narrow bands. This surprising
result could be explained by the high degree of redundancy characterizing the high spectral resolution data,
particularly if small absorption features (such as the one potentially induced by lignin, nitrogen, cellulose . .. in
the middle infrared domain) are not accounted for. Further, the addition of non independant wavebands could
add some significant noise or bias that are fitted through the inversion process by tuning the canopy biophysical
variables values to a wro ng or unstab l e value.
!
® 188Bands I
j D 6 Bands
Figure 5. Regression coefficient ( r- = i _ v( X - if /v(*_ xf where x is the measured value of the canopy variable.
x is the estimated value, and 3c is the mean measured value) associated to the relationships between measured
and estimated values of leaf area index. LAI. canopy chlorophyll (LAi.Cab) and water (LAI.Cw) contents. The
inversion was performed both on the 188 AVIRIS like bands and the 6 TM like bands on all the plots except
those with a white background. The negative values of the R2 associated with leaf chlorophyll (Cab) or water
(Cw) contents are not presented. The negative values are due to the variance of the residuals that was greater
than the variance of the measured values.
Figure 6 . Comparison between measured canopy chlorophyll ( LAI.C^ and water (LAI. CJ) contents and the
values estimated through model inversion using the 188 AVIRIS bands. The inversion was performed with
[ C ab , Cyy, LAI] to be retrieved. Dots, stars and crosses correspond respectively to natural soils, white and black
backgrounds. The symbols for plots with LAJ> 0.5 are inscripted in a circle.