near-infrared reflectances, corresponding to the AVHRR spectral bands. This widely used model takes soil and 
directional effects into account. The main assumption here is that the LAI is the strongest driver of the 
reflectance signal. Satellite data have to be filtered, to avoid contamination by clouds and occurrences of 
turbid atmosphere, and this filtering reduces the number of available data. Our 'observations' have therefore 
been sampled to weekly data, to give a realistic frequency of measurements. Resulting reflectances are shown 
on figure 3. Canopy development leads to an increase in the near infra-red band, and a decrease in the visible 
band. The soil is assumed to be lambertian, while the directional properties of the vegetation result in high 
frequency variability, caused by the orbital cycle of the satellite and sampling process. 
33. Assimilation of synthetic reflectances 
3 32.1st guess simulation. We assumed that all the parameters of the model are correct except , Iq and a. 
The values prescribed to Ej (1000 °C instead of 600 °C for 'ref) and (3700 °C instead of 3200 °Q roughly 
correspond to a 35 days shift in the start of the growing season and the senescence period. These errors are 
'reasonable', because global scale phenology modelling is not yet well established. The allocation coefficient a 
was decreased from 0.48 to 0.45. The LAI time profile generated with this parameter set is displayed on figure 
1. 
Fig. 1 Seasonal profile of Leaf Area Index, obtained with the reference parameter set (ref) and the first guess 
(1st guess), for which GPP is decreased by 26% and NPP by 37%. 
This figure also illustrates the sensitivity of the model to the chosen parameters. GPP and NPP values drop 
from 1011 to 740 g C m-2 year-1 (-26%) and from 625 to 392 g C m-2 year-1 (-37%) respectively, mainly 
because photosynthesis, for first guess simulation, occurs when temperature and daily irradiance are depleted. 
Fig. 2 top) Daily Net CO 2 exchange between vegetation and atmosphere for the reference simulation. Positive 
values correspond to assimilation of CO 2 by the vegetation, when photosynthesis is the dominant process, 
while negative values outside the growing season are due to stem maintenance respiration, bottom) Difference 
in Net CO 2 between the reference and the first guess simulations. The reference canopy develops earlier and 
allows high photosynthesis thanks to spring warm temperature and high irradiance leading to high positive 
values. In contrast, after day 300, although the '1st guess' canopy is still developed, photosynthetic rate is 
rather low and the difference of Net CO 2 fluxes take small negative values. 
1088