×

You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

Title
Mesures physiques et signatures en télédétection

269
HIGH SPECTRAL RESOLUTION SPECTROSCOPY FOR MONITORING CORN
GROWTH AT EFEDA SITE (SPAIN)
M.A. GILABERT & J. MELIA
Departament de Termodinämica, Facultat de Fisica, Universität de Valencia
46100-Burjassot, Valencia, Spain.
ABSTRACT
The empirical study reported in this paper has focused on the relationships between vegetation properties and the
upwelling spectral radiances observed above a com canopy throughout its phenological evolution. NDVI data and
red edge shifts have been analyzed as a function of LAI and biomass in order to document the phenological
evolution of the com. One of the results shows that there is an obvious tendency for the aforementioned spectral
data to reach a plateau at very high LAI levels (middle of August). This temporary saturation disappears with the
subsequent development of the vegetative structure, looking almost like an hysteresis cycle for the NDVI as a
function of the LAI. The study shows that LAI seems to be a better indicator of the whole cycle of com growth
when spectral measurements are involved, meanwhile biomass is more appropriate at the first stage of the
phenological cycle.
KEY WORDS: High spectral resolution, Field Radiometry, Corn Canopy, Phenological Evolution, EFEDA
Project.
1 - INTRODUCTION
Assessment of biophysical characteristics of vegetation, such as leaf area index (LAI) and biomass, has been a
major goal of remote sensing of agriculture. This assessment is usually made possible because of the contrast
between the spectral reflectance of vegetation and the soil background. The wavebands of reflected radiation that
have been of most value in these studies are the red, R, and near-infrared, NIR. Reflected red radiation is
negatively correlated with chlorophyll concentration and thus to leaf area, whereas reflected near-infrared radiation
is positively correlated with the amount of multiple scattering at the interfaces between cells and the air, and
therefore to the area of leaf also (Knipling, 1970). Vegetation indices computed from these two bands have been
related to various vegetation canopy properties, including green LAI (Price, 1992; Wiegand et al., 1992), canopy
biomass (Anderson and Hanson, 1992), absorbed photosynthetically active radiation ( Daughtry et ah, 1992),
grain yield (Ashcroft et ah, 1990) and total nitrogen content (Ashcroft et ah, 1990; Jensen et ah, 1990). The
most common of these indices are the NIR to red ratio, the normalized difference vegetation index (NDVI) or
linear combinations of red and NIR reflectances (Guyot and Baret, 1990). Due to the limitations that these
indices still present various efforts have been made by researchers to design new ones (Huete, 1988; Guyot and
Baret, 1990).
As recent investigations have shown, broad band spectral data are of limited value for the description of
plant properties. Considerable improvements may be expected from the extension of the spectral resolution down
to bandwidths of a few nanometers. New techniques such as the derivative spectroscopy, which is an established
method in analytical chemistry, are also used to define spectral parameters to be correlated to biophysical data.
Measurements with such a high spectral resolution open up new opportunities to find characteristic spectral
phenomena correlating to the status of crops.
In the context of remote sensing, the relationship between leaf chlorophyll content and the wavelength
of maximum slope on the 'red edge' is a well known application of derivative spectroscopy (Horler et al., 1983).
The red edge is the sharp change in leaf reflectance between 650 and 800 nm, approximately, and has been
measured on leaves of a variety of species by first derivative spectrophotometry. Horler et al. (1983) defined a
red-edge parameter, >^ e , as the wavelength of maximum slope and found that it was dependent on chlorophyll
concentration. They interpreted the results in terms of Beer's Law and Kubclka-Munk theory. In particular, rededge
rededge shifts associated with phenological crop development were demonstrated by sampling flag leaves of winter
wheat and spring barley on several occasions. As chlorophyll concentration increased, A re moved to longer
wavelengths, and then reverted as senescence began.
Demetriades-Shah et al. (1990) made a complete review of derivative techniques applied to remote
sensing, showing an example of the use of derivatives for monitoring vegetation subjected to chlorosis inducing
stress, for a sugar beet canopy. Danson et al. (1992) found that the first derivative of the reflectance spectrum at
wavelengths corresponding to the slopes on the edges of the water absorption bands was highly correlated with
leaf water content and insensitive to differences in leaf structure. On the other hand, Li et al. (1993) clearly show
the advantages of the derivative technique over the conventional vegetation indices (ratio and NDVI). Their
results show the insensitivity of the second derivative of canopy spectral reflectance to the soil background. The
study of Gemmell and Colls (1993) also is an example about how the analysis of spectral derivatives may