2.2. Empirical approach 
Leaf surface and mesophyll structure induce confounding effects when chlorophyll or carotenoid are to be 
retrieved from reflectance measurements. They act linearly (offset and multiplicative factor) on leaf reflectance as 
demonstrated by equation (1). For this reason, ratios of reflectances observed in the blue domain where carotenoid 
and chlorophyll absorb, and in the red domain where only chlorophyll absorb lead to a First order approximation 
of C c /C a )when reflexion by the leaf surface is small. However, derivation of an analytical relationship from 
equation (1) is not straightforward, except when C c is strongly linked to C a . For limnology and oceanography 
studies, Margalef (1974) used the ratio of absorbances in the blue (A b ) and the red (A r ):A b /A r to empirically 
estimate the CjC a ratio. Analogously, we defined the simple ratio pigment index ( SRP1) as the ratio between 
blue (R b ) and red (R r ) reflectances: SRPI=R b /R r . Penuelas et al. (1993a, and b) proposed in the same way the 
normalized difference pigment index, (NDPI=(R r -R b )l(R r +R b )) to evaluate the ratio of total pigments to 
chlorophyll a. Note that both indices are functionally linked since NDPI=(1-SRPI)/(1+SRPI). These two former 
indices are empirically derived and might suffer from some limitations. In the following, we will derive semi 
empirical approaches to derive C c /C a ratio from equation (1). 
2.3. Semi-empirical approach 
If the specular component is supposed to be dominant in the R s term, equation (1) could be written as follows in 
3 spectral domains : 
blue domain: R b =R s +S.exp(-k a b C a -k c b C c ) 
red domain R T -=R s +S.exp(-k a T C a ) (2) 
near infrared domain: R n -R s +S 
Thus, we have: 
R n -R b =S(l-exp(-k a h C a -k c h C c ), 
and R n -R r =S(l-exp(-k a T .C a )) (3) 
We define the structure independent pigment index ( SIPI ) as: 
SIPI =(R n -R r )l(R n -R b )=(\-exp{-kJ.C a ))l{ 1 -exp(.-k b >C a -k b >C c )) (4) 
Thus, the SIPI index becomes a function of C a and C c and the effect of the structural components should be 
eliminated. However, the relationship between SIPI and C ( JC a ratio could not be derived analytically, except if 
carotenoid and chlorophyll pigment concentration are closely linked together. 
3. MATERIAL AND METHODS 
3.1. Plant material 
Several plant species and conditions were chosen to get a wide range of variation in leaf pigmentation. The 
species studied were maize (Z ea mays L.), wheat ( Triticum aestivum L.), tomato ( Lycopersicon esculentum 
Mill.), soybean ( Glycine max L.), and sunflower ( Helianthus annuus L.), grown in a greenhouse, and sugar beet 
(Beta vulgaris L.), oak (Quercus robur L.), maple (Acer negundo L.), and a succulent plant (Othonopssis 
cheiriifolia L.), grown outdoor. Several leaves of each species were collected, and their spectral reflectances were 
measured. Precise description of the leaves could be found in Baret et al. (1988), Malthus (1989) and Jacquemoud 
and Baret (1990). 
3.2. Reflectance measurement 
Directional-hemispherical reflectance of the adaxial faces of the leaves was measured in the laboratory using a 
Varian Cary 17 DI spectrophotometer. It is equipped with an integrating sphere coated with Barium sulfate. 
Complementary bidirectional reflectance of the upper face of sugar beet leaves were measured in the laboratory 
with the GER IRIS spectrophotometer as described in Malthus (1989). For both instruments, the spectral 
resolution was better than 2 nm in the 400 nm-800 nm domain studied. A 5 nm sampling interval was used for 
both instruments. The wavelength calibration was checked. Reflectance data were calibrated using a known 
barium sulfate reference sample. The accuracy of the measurements was about 1% reflectance. 
343