LAI rinn*, profile can be not correlated. Thus, if crop production results essentially from the absorbed 
photosynthe.fically active radiation by a LAI “big- leaf’ in most of the production models, they are in fact, more 
successful for the crops without vegetative period (winter dormancy when tillering occurs) like: maize, 
sugarbeets, spring wheat... For the winter wheat, which can show important differences in the canopy 
development regarding to the farming practices. Porter (1984) and his colleagues found that it was a fundamental 
point to simulate correctly the shoot population and the growth of the leaves at the individual level in oder to 
have a good description of the variables relevant for the yield estimation. 
Because radiosity approach seems to be possible due to the architecture description made by 
AFRCWHEAT, we explain here the main considerations used in the canopy development submodel. Driving 
variables of the canopy development part are the daily average temperature, the time, and the daylength. 
Leaf appearance: The rule used came from ground observations showing an empirical relation between the rate 
of leaf appearance per degree day and the rate of change of daylength at emergence. So, in a practical way, the 
phyllochron interval (thermal time required between the appearance of successive leaves), computed at the day of 
emergence, is bigger for early sowing fields than for late sowed fields, because emergence day is closer to the 
equinoxe date in the first case than in the second case. A mean value of the phyllochron interval is 100 - 110 
degres.days. 
Tiller production and survival : When the stage “4 leaves an the main shoot” is reached, tillers are produced until 
the double-ridge stage. A time step of 7 days is used for die simulation: the number of tillers initiated in a week 
depends on the temperature of the previous week, and on a shoot production rate. Tiller production stops at 
double ridge stage. The survival of each tiller group is calculated from the accumulated thermal time since double 
ridge, taking into account the existing shoot density at the birth of each tiller group. Higher is this density, more 
reduced is die proportion of surviving tillers. In a practical way. it means that the first bom tillers have a chance to 
survive of about 0.9 or 0.95, when the tillers bom near the double ridge stage generally do not survive. At 
anthesis stage, tiller death stops, and all live shoots are assumed to cany ears. 
Leaf growt h and senescence- The maximum dimensions of the leaves are defined and depends on the position of 
the leaves an the stem. Each leaf reaches its final size in 1.8 phyllochrons. The model simulates the leaf growth 
by describing the daily length and width growth of both laminae and sheaths. Because various observations 
during the main period of vegetative growth showed that there are generally between 3 and 4 active green leaves 
on a shoot, the thermal time required from the attainment of maximum size to zero active area is estimated to be 
3.5 time the phyllochron interval far most of the leaves. Each leaf remains totally active before the b eginning of 
the senescence during approximately the 2 thirds of that time. 
23 Radiosity experiment 
We want to examine the results of a radiosity computation at early stage of the growing season of the winter 
wheat The program of the functional runs with a daily time step. When the day processed is a day chosen for the 
computation of the radiosity, all the variables related to the description of die structure are written in an output 
file. It means: a) per square meter main shoot population, population of the first category of tillers (on the first 
leave of the main shoot), of the second category, etc b) dimension of all the existing leaves: width and length 
of lamina and sheat areas, c) percent of senescent area on the various leaves. 
To keep computation time (which is very hi gh with the radiosity technique) w ithin reasonable 
limits, we restrict our study area to a quater of square meter. Then, with an original sowing density of 250 shoots/ 
square meter, we worked with 63 main shoots on our study area, which allows an easy representation in 7 rows of 
9 shoots. The space between the rows is approximately 7 cm, and die space between 2 shoots in the same row is 
between 5 and 6 cm. The x y positions of the main shoots are then defined. The x y position of any tiller is the x y 
position the main shoot where this tiller appeared, plus a deviation of 1 or 2 cm following an «ximnth direction 
which is randomly chosen. 
Some other information, which are not simulated by the model are needed to use the radiosity 
techniques: the height and the orientation of each element. Many graphic representations or pictures of a wheat 
shoot at various phenological stages have been consulted in general agronomic literature (e.g. Soltner, 1978), as 
well as some reported values of height or leaf mr.linatirm which were found in published studies. The values used 
in that work are arbitrary, even if they are carefully chosen. In a future study, the use of empirical relation should 
be used to add to the canopy development submodel, the simulation of the position (especially the height) and the 
orientation of the leaves. For same particular days, we present now the structure of the canopy and our way to