×

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

ABSTRACT
A regional sahelian grassland ecosystem model had been developed within a methodology to integrate remote
sensing data into an explicit formulation of the main processes occuring in the ecosystem. A validation of the
model, irrespective of remote sensing data, had been carried out with biomass measurements acquired on two
different regions of the Sahel. In order to assess how sensitive vegetation indices are to varying vegetation
conditions, a canopy reflectance and a soil reflectance model fed with structural and other parameters simulated
by the ecosystem model were used to compute reflectances of the green and dry vegetation, and of the soil. A
simple linear mixture model using the simulated vegetation cover fractions were applied to weight the
contributions of the different components of the landscape reflectances. The latter were then processed through
an atmosphere model to yield observable TOA reflectances. Confrontation with real satellite data showed that
observed NDV1 profiles could be well reproduced by model simulations, in phase as well as in amplitude.
Key Words : ecosystem model, grassland, Sahel, remote sensing, reflectance model
INTRODUCTION
It is nowadays generally accepted that vegetation monitoring at a regional scale should necessarily be carried
out with the contribution of remote sensing techniques. This is particularly true for the sahelian zone where
direct measurements are difficult to perform due to both the high diversity and heterogeneity of vegetation
formations and the logistic problems associated with field measurements. Since 1979, satellite data delivered by
the AVHRR sensor on board the NOAA series has been widely used for this purpose, as they contain repetitive
observations of the whole Sahel region on a daily basis and at a scale compatible with the regional processes
under consideration. At present, the NDVI, the Normalized Difference Vegetation Index of the bidirectionnal
reflectances in the red (0.58-0.68 pm) and near-infrared (0.72-1.10 pm) is commonly used to monitor the
sahelian vegetation development during the growing season as it is known to be highly correlated to vegetation
parameters like the Leaf Area Index LAI and the Photosynthetically Active Radiation (PAR) interception
efficiency Ej (Curran, 1980). In addition, the NDVI integrated over a growing season (or END VI) is used to
estimate the aboveground season biomass through a simple linear relationship (Tucker et al, 1983; Justice and
Hiemaux, 1986 ; Diallo et al, 1991). Such a relationship represents, in fact, the extension of models that have
been developed for crops. Indeed, for homogeneous crop canopies, there exists a quasi-linear relationship
between NDVI and Ej (Kumar and Monteith, 1982; Asrar et al, 1984) which once integrated in the Monteith's
productivity model (Monteith, 1972) leads to a linear relationship between aboveground biomass BM produced
during the time interval t and E,NDV1 under the form : BM = e E, (aNDVI.PAR), where BM is expressed in kg
dry material per ha (kgDM ha’ 1 ), e is the growth efficiency (gDM MJ* 1 ) and a is a coefficient of proportionality
relating NDVI to Ej (Tucker et at, 1981; Kumar and Monteith, 1982). This relation has more or less been
successfully applied to semi-arid countries and, in particular, to the sahelian zone (Prince, 1991a). On the
whole, this relationship is valid for a specific site and at the scale of a growing season but the same relation
does not hold for different sites and for successive years, thus strongly restraining its application. Furthermore,
such a model does not allow the primaty productivity and associated physiological processes to be
quantitatively monitored during the growing season. Beside the problems of uncorrected atmospheric effects,
limitations of the applicability of this method that can be identified are the following : firstly, the bare soil and