A Model of Light Scattering in Three-Dimensional Plant Canopies: A Monte Carlo Ray Tracing Approach

The global quantitative monitoring of terrestrial environments from optical remote sensing involves five main steps: (1) identification of the spatial and temporal scales and resolutions at which information on vegetated surfaces are required; (2) characterization of the nature of relevant radiometric signatures; (3) implementation of an instrument designed to observe these signatures; (4) development of mathematical tools capable of interpreting the radiative measurements in terms of the environmental variables of interest; (5) evaluation of the suitability and accuracy of the delivered products. To this end, physically-based models are developed to represent the radiative processes involved in the different steps. The physical processes that control the reflectance of a structured medium (such as a canopy) occur at a variety of spatial scales: the milli-scale which represents the smallest scattering element from a remote sensing point of view, typically a leaf; the meso-scale which concerns the arrangement of scatterers and finally, the macro-scale which includes topography and large landscape features such as the distribution of vegetation types. Clearly, a model to be used in remote sensing should represent those effects that are relevant to the scale of observation. This thesis focuses on the development of a new radiative transfer model, called Raytran, designed to investigate solar radiation transfer problems in terrestrial environments over a variety of spatial scales. The model is based on Monte Carlo procedures and the latest computer graphics ray tracing and parallelization techniques. The canopy reflectance is estimated on a ray-by-ray basis. This model may be used as a virtual laboratory, to generate reflectances and absorption profiles of three-dimensional complex targets where all geometrical and physical quantities can be controlled explicitly. The accuracy of the model has been established by comparisons with another published Monte Carlo model and with laboratory measurements. First, we explored the optical properties of a single plant leaf. The three-dimensional cell structure of the leaf is explicitly described and represents the different biological constituents of the leaf. Thanks to the very complex scene which can be described with Raytran and the availability of significant new parallel computing resources, this is the first time that the radiative transfer in a leaf is computed taking into account its three-dimensional structure. The accurate representation of the spatial organization of the canopy is one of the main advantages of the Monte Carlo ray tracing technique. We reviewed different approaches to parameterize the structural organization of the canopy, from simple statistical representations up to more sophisticated grammar-based methods such as the L-systems. We then explored the potential use of laser altimeter echo recovery method to infer the spatial organization of the canopy, discussed the advantages and drawbacks of the method and compared it with classical bidirectional reflectance measurements. Finally, we illustrated the potential of Raytran to evaluate