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

♦Laboratoire dEtudes et de Recherches en Télédétection Spatiale (CNES/CNRS UMC00010)
18, avenue E. Belin (BPI 2801), 31055 Toulouse Cedex, France,
tel : (33) 61 28 14 18, fax : (33) 61 28 14 10, e_mail : moulin@lerts.cnes.fr,
**INRA Station de Bioclimatologie,
BP 91, 84143 Montfavet, France.
This paper presents the modelling of the time profile of satellite radiometric signal by the use of a vegetation
growth model, and the assimilation of remotely sensed observations into canopy model. We combine satellite
measurements and a functional model to monitor the development of the cover. In direct way, the linkage
between a crop growth model and 4 different reflectance models allows to reproduce well reflectances
observed on SPOT/HRV scenes over test-sites. Then, over any wheat crop, we assume that correction on initial
state can be obtained by the minimization of the difference between observed and simulated reflectances.
Consequently, after adjustment of initial conditions, we can be confident on the simulation of different
processes by this crop model, in particular daily carbon budget of the green cover.
KEYWORDS : Assimilation, Satellite data, Vegetation growth model, Radiative transfer model.
In the frame of this study, we attempt to assimilate remotely sensed data into a crop model in order to retrieve
some pertinent parameters which characterize the cover and which cannot be observed by remote sensing. So,
we need both radiative measurements to monitor the cover seasonal evolution and vegetation models to
simulate biophysical processes. The coupling of data and models is performed through a state variable (e.g.
leaf area index) which appears in the radiative budget and can characterize the vegetation : remotely sensed
data are used to monitor the green cover development and adjust crop parameters which are required to
simulate cover growth. We particularly expect to retrieve the sowing date of a crop cover, which governs the
plant development.
Many previous studies have used satellite measurements in complementarity with vegetation model.
The different techniques have been summarized by Deldcolle el al. (1992) and Seguin et al. (1991). Most of
these studies concern the inversion strategy : biophysical variables are computed from remote sensing data and
used as inputs into the cover model through a forcing method. This approach leads to the estimation of
important variables. However, these do not appear directly in the processes modelling. On the other hand, if
we expect a vegetation model to describe well the biophysical mechanisms, satellite data are used according to
the direct way strategy, through assimilation techniques. In this study, we focus our attention on : first, the
simulation of radiometric time evolution at the top of the canopy and second, the assimilation of satellite
measurements (corrected for atmospheric effects) into vegetation model. Here, the radiometric time profile is
completely simulated in the direct way : a radiative transfer model is linked to the functional model to process
reflectances that can be remotely observed. Then, the modelling is considered reliable when it agrees with
Different types of model have been used in complementarity with satellite measurements. Fust, an
empirical relationship can be used : e.g. between biomass production and vegetation index (Tucker et al.,
1986). Next, semiempirical model can describe the seasonal vegetation index (Fischer, 1993). Models can also
be a simply retrieval of the green cover fraction (Monteith, 1977). In each case of empiricism and
parameterization, models do not describe biophysical processes. A more descriptive model like process
vegetation model is necessary to understand the meaning of remotely sensed observations and thus to
assimilate them.