SEASONAL BEHAVIOR OF CROP CANOPIES: 
LINKAGE OF FUNCTIONAL MODELS AND RADIOSITY. 
A .FIS CHER and F. CABOT 
LESTS (CNES/CNRS - UM C00010) bpi 2801 
18, av. E. Belin, F - 31055 Toulouse Cedex - France 
ABSTRACT 
The first objective of a crop production model is the correct simulation of the mechanisms which drive the yield. 
In order to compute the absorbed photosynthetically active radiation, the LAI time profile is described to 
characterize the development of the canopy generally considered as a big leaf. The linkage with radiometric 
measurements in short wavelengths, which allow to monitor crop growth and canopy development during the 
growing period, is classically realised by the mean of a radiative transfer model. In such models, the canopy is 
assumed to be a semi-infinite medium with little particules of known optical properties homogeneously 
distributed. A crop canopy, especially at the beginning of growth, cannot be correctly represented in that way. 
The AFRCWHEAT model simulates the canopy development with description of the elements at the individual 
level (population and size of each element). It is then possible to test the radiosity method, in order to compute 
bidirectional reflectances over the young canopy taking into account the light interception for each element. 
Results show that the mean features of the visible reflectances are retrieved, with a clear evidence of the hot spot 
phenomenum. 
KEY-WORDS: Functional model. Canopy structure. Reflectances, Radiosity. 
1 - INTRODUCTION 
Bidirectionnal effects on the surface reflectance are an important limiting factor far the accuracy when using 
remotely sensed data, mainly for multitemporal studies. Many attempts have been made to modelize these 
effects, through physical or empirical expressions, in order to take them into account when analyzing the 
variations of the surface reflectance. If some success have been acquired in this topic, the temporal behavior of 
the surface due to vegetation activity infers on the variations of the parameters in a way that is hardly known 
especially when considering heterogeneous structures. This study uses a deterministic description of the surface, 
based on the radiosity method, with inputs (LAI, geometry, ...) from a crop production model and should be 
considered as a preliminary step on the path to model confidently a virtual canopy, upon which we would have a 
total control, well designed far testing existing models and parameter retrieval procedures. 
The radiosity method is briefly described in the next section. The simulation of the development 
of a wheat canopy gives some information about the architecture, which must be completed by empirical 
estimations of the position and the orientation of the elements. In the third section, die radiosity method will be 
especially tested at the beginning of growth, over a sparse canopy with low value of LAL 
2 • MATERIALS AND METHODS 
2.1. The radiosity method 
The radiation transfer regime inside a given kind of canopy has been die main topic of many studies. Still, it 
appears very complicated to account far all the heterogeneities that occur in natural cover types. Most of the 
solutions proposed up to now simply assume homogeneity in one or more dimension in order to make possible to 
solve analytically the radiative transfer equations. An alternative method has been described by Borel et al. 
(1991). The basic principle comes from thermal computations but can be applied to short wavelengths. This 
method independantly considers all the elements of a canopy, and therefore needs a very accurate description of 
the architecture of the plants and soils (figure 1). The idea is to compute, for each element a radiation budget and 
to solve the obtained set of equations. To achieve this computation, one needs to know three parameters: F the