1120 
2 - MATERIALS AND METHODS 
2.1 Terrestrial Biosphere Model 
The terrestrial biosphere model we used, is composed of a Net Primary productivity and a Soil Respiration 
module, which are both presented in the following paragraph. 
2.1.1 Net Primary Productivity NPP 
The NPP model is based on the Monteith's approach (1972, 1977), which considers NPP as the product of 
incoming solar energy by various yields, or efficiencies of conversion. In the original model for crop growth 
remote sensing (Kumar and Monteith, 1981), NPP is the product of 3 efficiencies by the incident global 
radiation: 
NPP(t) = e . f (t). c. Sf/t) (1) 
with : t : time 
c : 'climatic efficiency*, i.e. the ratio between incident Photosynthetically Active Radiation (PAR) and 
incident global radiation. In this study it is fixed to 0.48 (Me Cree 1972) 
/: 'absorption efficiency*, i.e. the ratio (PAR absorbed by the canopy)/(incident PAR) 
e : 'conversion efficiency*, i.e. the ratio (dry matter produced)/(absorbed PAR) 
The absorption efficiency / is assumed to be linearly related to the remotely sensed Normalized Difference 
Vegetation Index or NDVI (e.g. Asrar et al. 1984): 
fit) =A + B.NDVl(t) (2) 
A and B were calibrated using minim um and maximum NDVI values for all the years of study (0 <, fit) <, 1). In 
a first step we use the same couple of coefficient for the six considered years, but as we focused on 1989, we 
used two different sets of coefficients, a first one for the "top atmosphere" NDVfl° A , a second one for the 
surfaces vegetation indexes, corrected from the atmospheric effects. In the rest of the text, TOA and SURF 
upperscript will refer to variables respectively without and with atmospheric correction. 
As e is function of the considered biomes (Ruimy et al. 1993-a), we consider that, at the first 
order, e depends on autotrophic respiration. We have therefore the following model: 
NPP(t) = [1 - r(t)J.GPP(t)= [1-r (()]■ e'. f(t). c. Sf/t) (3) 
with : e': *photosynthetic efficiency*, i.e. the ratio (assimilates produced by net assimilation)/(absoibed PAR) 
r : 'respiratory losses': fraction of assimilates photosynthesized lost by autotrophic respiration. 
The parameterization of e' was done by plotting daily CO2 fluxes measured above canopies against absorbed 
PAR for several vegetation types (Ruimy et al., 1993-b)): e' is the slope of the linear regression between these 
variable (e*=4.45 g COj assimilated per MJ of absorbed PAR ). In this preliminary study, we used the 
parameterization of Goward and Dye (1987), where r, the respiratory losses, depend on temperature only: 
r(t) = (7.825 + 1.145. T(t))/100 (4) 
with: T: air temperature (°C) 
2.1.2 Soil Respiration SR 
The CO2 flux from the soil to the atmosphere depends on several environmental and soil-related parameters: 
carbon content of the soil, temperature, humidity,... At the first order, it depends on temperature. In this model, 
we use an exponential dependency on temperature: 
SR=K.Q W 1VW (5) 
with : K : Heterotrophic respiration coefficient 
Qjq : Relative increase of respiration flux for an 10° increase in temperature T (Qjo “ 1.5 in this 
study). 
In order to parameterize the K coefficient, we have considered the biosphere in equilibrium over a year on each 
pixel (approach of Heimann and Keeling 1989 for instance). Our approach should be improved in the future by 
using a simple parametrization of SR, independent of NPP over short time scales. 
2.2 Input data sets