271
TABLE 1
Summary of the reflectance measurements concerning the date and the phenological stage of the com.
DATE
PHENOLOGICAL STAGE
19 June
Four leaves out
1 July
Six leaves out
14 July
Eight leaves out
6 August
Fourteen leaves out
15 August
Sixteen leaves out (spike tip showing)
15 Sept.
Pasty grain
NOTE: Another series of measurements were carried out on 27 August but they were lost
by technical reasons.
On the other hand, a red edge analysis was performed throughout the phenologycal cycle of the com. For
this reason, the reflectance spectra were firstly smoothed (in order to eliminate the possibility of choosing
maxima which are created by random noise) and after this, the first derivative curves were calculated using the
GER software package. The spectral zones of the curves compressed between 600 and 800 nm were analyzed in
terms of the wavelength corresponding to the maximum, X re , and the value of the derivative at this point,
(dR/dA.) re . This latest parameter has not been incorporated in the study because it was found to be very correlated
with the NDVI.
In next section, the relationships between the biophysical parameters and all these spectral parameters
will be analyzed.
3 - RESULTS
A preliminary analysis (Gilabert et al., 1993) was carried out to study the dispersion among the reflectance
values as due to the spatial variation (when we move along a row) and to the temporal variation (along the period
of the day when the measurements were taken). It could be noted that:
(i) The dispersion among the values, for a given solar angle, which seems to be a consequence of the gaps
between vegetation in a row, is greater in the first period of the plant development.
(ii) The dispersion among the mean reflectance values for different solar angles (evolution of the reflectance along
the day) is normalized when vegetation index values are used. Therefore, it is recommended to use vegetation
index values to spectrally characterize the canopy in each date.
In order to compare the temporal evolution of LAI, biomass, NDVI and the red edge, 7 re , Figure 1
shows their values versus the Julian day; the biomass is represented in a logarithmic scale, for the better
presentation of its highly increasing behaviour. As can be observed, the LAI exhibits a maximum around the day
229 (17th August), just the period when the com presents its maximum foliar surface and its maximum
photosynthetic activity. After this period LAI decreases as due to the senescence of the leaves, although the
photosynthetic activity and the production of biomass remains a little more. Therefore, positive correlations
between spectral parameters and LAI must be expected since they present a maximum situated around the same
date.
In order to study the correlations between biophysical and spectral parameters, and given that for these
latest we do not have measurements coincident in time with LAI and biomass, a fit of the NDVI and red edge
values has been carried out as a function of the Julian day, D, by means of the quadratic law shown in
expressions 1 and 2:
NDVI = -8.68 + 8.45x 10" 2 D - 1.85xlO' 4 D 2 r 2 =0.98 (1)
X re = 76.5 + 5.69D - 1.21x 10' 2 D 2 r 2 =0.96 (2)
The equations of the fits let us estimate the values of NDVI and corresponding to the dates where LAI and/or
biomass measurements were available.
3.1. Broad band analysis
Figure 2 shows the NDVI values as a function of the LAI. As can be observed, there is an obvious tendency for
the NDVI to reach a plateau at very high LAI levels (middle of August). The trend indicates a temporary
saturation of reflectance which disappears with the subsequent development of the vegetative structure. This
asymptotic behaviour has been reported by several authors (Baret et al., 1989; Sellers, 1989; Wiegand et al.,
1992). As can be observed, the data show a steep initial response, or good sensitivity; that is the relation is
therefore of most value when NDVI is small, which occurs with low amount of vegetation.
