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

Mesures physiques et signatures en télédétection

Environmental Information Centre
Institute of Terrestrial Ecology
Monks Wood Abbots Ripton
Cambridgeshire PE 17 4LS
United Kingdom
Tel: 44 487 3381 Fax: 44 487 3277
Simple semi-empirical models can be used to relate leaf area index (LAI)and the fraction of photosynthetically
active radiation absorbed (P) by plant canopies to vegetation indices (VI).These models do not explicitly account
for the effect of leaf optical properties, average leaf angle (ALA), soil properties or solar-target-view geometry
on the VI-P and VI-LAI relationships, Baret (1988). Therefore the estimates of LAI and P derived from such
semi-empirical models contain noise due to the uncontrolled variability in these other controlling factors. A
further and hitherto unquantified source of noise is due to uncertainty in the vertical distribution of chlorotic
material within the canopy. A canopy reflectance model was used to simulate the vegetation indices (NDVI and
TSAVI) of canopies with three different distributions of chlorotic material. A range of ALA, soil reflectance and
LAI were used in the simulations. Semi-empirical models were fitted to the simulated LAI-VI and P-VI data.
The relative equivalent noise (REN) of both the LAI-VI and P-VI relationships were calculated. The signal to
noise ratio of VI for LAI prediction fell below unity when LAI>2. Chlorosis related noise exceeded both the soil
reflectance and the ALA related noise when LAI>2. The signal to noise ratio of VI for P prediction always
exceeded unity for NDVI and is always greater than 10 for TSAVI. Chlorosis distribution had only a minimal
effect on the noise in the P-VI relationships. When P>0.75 the noise was greater for unknown than known
chlorosis distributions.The TSAVI noise was dominated by uncertainty in the ALA throughout the range of P,
but NDVI noise was dominated by soil reflectance at low P.
KEY WORDS: Chlorosis distributions, semi-empirical models, LAI, APAR, relative equivalent noise 1
The relationships between LAI and vegetation indices, and between the fraction of absorbed photosynthetically
active radiation (P) and vegetation indices have been investigated empirically for several crops (Daughtry et al,
1983; Asrar et al, 1984; and Daughtry et al, 1992); these studies have shown that simple semi-empirical
relationships, analogous to Beer’s Law, provide a good fit to the field data. However there is "noise" or error
associated with spectral estimates of LAI and P. This noise is caused by the uncertainty in other factors which
affect the canopy reflectance and hence vegetation indices. Soil reflectance, average leaf angle (ALA), solartarget-view
solartarget-view geometry and leaf optical properties have been shown to be implicitly related to the parameters in
these semi-empirical models, Baret (1988).
An implicit assumption often made when semi-empirical and physically based canopy reflectance models
are used to extract biophysical information from radiometric data is that chlorotic material if present is randomly
dispersed throughout the canopy. This paper sets out to examine the effect the distribution of chlorotic material
within the canopy has on the noise in the LAI and P estimates predicted from semi-empirical models. There are
two specific objectives to this study. The first is to examine the effect of changing the chlorosis distribution on
the noise associated with ALA and soil reflectance will be examined. The second is to examine the noise arising
from uncertainty in the chlorosis distribution will be examined in conjunction with the effects of uncertainty in
ALA and soil reflectance.