270
provide a quantitative method for specific remote identification of plant stress induced by sulphur dioxide in the
field. They found that the spectral effects of sulphur dioxide on leaves could be quantified by means of the
changes in curve-slope of spectral first derivatives.
Baret et al. (1992) analyzed the spectral shifts characterized by the wavelength of the inflexion point in
the red-edge region by using model simulations. They found that the information provided by the high spectral
resolution index appears to be equivalent to that obtained from the red and near-infrared broad band reflectances at
leaf level. But a canopy level the red-edge inflexion point is more sensitive to leaf area index and chlorophyll
content.
Miller et al. (1991) studied the reflectance of leaves from ten tree species sampled at weekly intervals for
a period of 150 days. The red-edge position was found to be useful to describe phenological events characteristic
of deciduous trees.
In this study, the monitoring of the growth of a com canopy is carried out by means of the correlations
between spectral and agronomical parameters. In the first group, the NDVI, derived from a broad band analysis,
and the red edge wavelength, derived from a high resolution analysis, have been selected; in the second group the
LAI and the biomass were used to characterize the phenological evolution of the crop.
2 - MATERIALS AND METHODS
The field experiment took place in a corn pivot from the EFEDA site (Barrax, Spain). The canopy spectral and
biophysical measurements were acquired on a regular basis from June to October during the 1992 growing
season.
2.1. Biophysical measurements
Biophysical measurements (leaf area index, LAI, and biomass) were regularly acquired approximately every seven
days throughout the growing season by the University of Castilla-La Mancha group from the EFEDA project
team.
Leaf area index measurements were obtained by using a LI-COR LAI 2000 Plant Canopy Analyzer.
This instrument estimates the LAI through the radiation transmittance values obtained ratioing the radiation on
the top and on the bottom of the canopy (LI-COR, 1992). To obtain a representative value for the LAI in each
date, different transects were carried out on the canopy and afterwards the mean value was calculated.
The biomass in each date was estimated by multiplying the sowing density of the plot by the average
weight of plants. To determine the sowing density, the number of rows per 10 meters was counted. This
operation was replicated 50 times in 50 different places of the plot randomly chosen. From this, it is easy to
obtain the plants per meter, which is constant for all the studied period. To determine the weight per plant, ten
plants were picked up from the crop and afterwards they were carried to the laboratory in order to determine the
fresh and dry weights by desiccating the plants in fluent air heater at 70°C temperature during 24 hours with their
shoots, leaves and fruits separated on different trays.
2.2. Radiometric measurements
Spectral radiance measurements (from 400 to 2500 nm), used to determine reflectance factors, were acquired at
approximately 2-weeks intervals with a GER SIRIS spectroradiometer, from planting (middle of June) till the
middle of September, just before harvesting (see table 1). The optical head of the instrument was attached to a
boom mounted on a pickup truck, elevated about one meter above the canopy, and levelled for a nadir view. An
Spectrally panel was used as the ideal diffuse reference surface, mounted on a tripod with a height similar to that
of the canopy. Reference panel measurements were collected immediately before the radiance measurements from
the targets. The spectral reflectance of the targets relative to the reference were calculated by using the GER
software package. The data for each wavelength were calibrated for the correction factor of the panel according to
the method proposed by Jackson et al. (1992), which has been previously used by the authors (Gilabert and
Melia, 1993).
The measurements were taken within 1 hour either side of solar noon. To ensure comparability among
data recorded on different dates, the radiance data were acquired for a longer period in two days to capture a wider
range of possible solar zenith angles. These two days were the 1st July (developing stage of the com) and the 6th
August (period of intense photosynthetic activity), corresponding to two different stages of the crop, presenting
different vegetation density. All the data presented in this study were collected under essentially cloud-free skies.
Radiance data were always obtained by scanning a row of the canopy, since the field of view of our
instrument (14°x4°) prevented us from getting a representative value for the global spectral response of the crop.
Two different kind of treatments were applied to the reflectance spectra. On the one hand, filters
reproducing the relative spectral response of the Thematic Mapper bands have been applied to each spectrum, in
order to reproduce the reflectance in TM3 (red) and TM4 (near infrared) bands. These values have been used to
calculate the normalized difference vegetation index, NDVI, selected in this study to spectrally characterize the
crop.
