273
whereas the others corresponding to the senescence of vegetation are shown in open symbols (post-L max data).
The relationship between NDVI and LAI for these two distributions of points has been approximated by (Baret et
al„ 1989; Wiegand et al„ 1992):
Y = A + B exp (-CX) (3)
where Y refers to NDVI and X to the leaf area index. For the first stage, A is the NDVLo , that is, the limiting
value of vegetation index at large NDVI; B is the difference between NDVI for the soil background and the
NDVIoo , and C is a coefficient related to the extinction of radiation through the canopy. For the second stage,
the meaning of the coefficient is still under keen analysis. The mathematical expressions obtained for the fits are
summarized in table 2 (the curves are also represented in figure 2).
TABLE 2
Coefficients of the adjustment between NDVI and LAI data according to the model shown in expression (3). The
meaning of A, B and C is discussed in the text, and r refers to the correlation coefficient.
P r e-Lmax data
post-Lmax data
A
0.956
0.965
B
-0.666
-30.7
c
1.13
1.70
r
0.99
0.98
Nevertheless, when the NDVI is represented versus biomass (figure 3), the distribution of data clearly
shows a different shape since after reaching the NDVI its maximum value, biomass still shows increasing values
and after it remains more or less constant until the end of the cycle. By this reason, the hysteresis behaviour is
not appreciated in this case. The logarithmic model, as shown in expression 4, has resulted to be the better way
to adjust the relationship (only for the pre-L max data since after this biomass remains more or less constant):
Y = A + B log X (4)
where Y refers to NDVI, X to biomass, and A and B are the parameters of the adjustment: A=0.098, B=0.288,
obtained with a correlation coefficient equal to 0.98.
Figure 3. NDVI values versus biomass for the first period of the phenological cycle of the crop. The logarithmic
curve obtained from the data adjustment has also been represented.
From figures 2 y 3, it could be concluded that the correlation LAI/NDVI seems to be a better indicator
for the whole phenological evolution of the com given that for the senescence period biomass remains practically