1128
The first step displays the methodology of simulation tested at a high spatial resolution. We simulate
the temporal profile of the canopy reflectances in the SPOT/HRV wavelengths. Radiative variables are
obtained through the linkage of a production model with a reflectance model. Here we compare results that
can be computed with several radiative transfer models. For each available acquisition day, we compare SPOT
observations with predicted reflectances over the study site. It is a validation of the methodology applied to
predict radiometric signal. The second step concerns the assimilation of satellite measurements into crop
growth model. SPOT observations are used as a reference. Then, the sowing date is adjusted to obtain
modelled reflectances as close as possible to remotely sensed measures. Retrieved sowing date is then
introduced in growth model to estimate carbon fluxes. Here again, we compare results obtained with 4
radiative transfer models.
2 - MATERIALS AND METHODS
We particularly consider the agricultural region of the Beauce, where wheat and barley are largely represented.
Over this well-known crop region, growth models relative to many crops are available. Moreover, as many
environmental factors can be controlled, the dependence of green cover growth and development with external
parameters is limited. The study area concerns a 40km by 40km surface in the south-west of Paris.
2.1. Satellite data : High spatial resolution
High spatial resolution data acquired by SPOT/HRV are used for this field scale study (Table 1). SPOT/HRV
observations are available for 5 clear days during the activity period of winter wheat Satellite measurements
have been corrected to take into account calibration and atmospheric effects. Climatologies have been used for
water vapour and ozone atmospheric contents. A constant value of optical depth is chosen to compute the
aerosol effects.
acquisition date
view zenith angle
view azimuth angle
solar zenith angle
solar azimuth angle
09 April
-6.6
104.0
42.2
162.6
13 May
-16.8
105.3
30.9
163.0
18 May
-22.7
106.2
29.6
164.7
19 May
4.3
282.6
30.5
155.4
19 July
30.3
100.8
29.8
162.6
Table 1 : SPOT/HRV data acquisition angles, in degrees, over the Beauce region, during activity period of
winter cereals.
As shown on Table 1, we have got 5 acquisitions dates, and we note that observations have been acquired in
different angular configurations. This geometry has been taken into account to simulate reflectances. The
observation dates are distributed over the winter cereals activity period. On April 9, it is the beginning of the
crops growth period. In May (13, 18 and 19) winter vegetation reaches its maturity stage. On July 19, cereals
have already been harvested.
12. Land use occupation
Many information about our study site can be available through a pilot project applying remote sensing to
agricultural statistics (Sharman and Boissezon, 1992). We first have a land use classification performed over
the 40km by 40km studied area which gives all the represented vegetal species and their respective surfaces.
Then, a sampling has been carried out over 16 test-sites of 700m by 700m uniformly distributed over the site.
Here, for most fields of the test-sites, we have got additional information like plant variety, sowing date and
crop yield (estimated by farmers).
23. Vegetation and radiative transfer models
2.3.1 The crop growth model AFRCWHEAT2
This functional model (Porter et al., 1992) can predict daily net assimilation of carbon dioxide by the
vegetation, respiration, assimilates distribution in different plant organs, phenological development, and
organic dry matter net production. The input parameters are : the meteorological data (minimum and
maximum daily temperatures, solar irradiance, precipitation, wet bulb temperature, mean wind speed), the
