Abstract : Plant growth modelling can be considered at different temporal and spatial scales. In the growth model GreenLab, spatial units are phytomers and temporal steps are growth cycles. Such scales are relevant for most crops but regarding branching plants with complex structures, description at these scales is difficult to achieve. The thesis proposes a multi-scale analysis of GreenLab according to different levels of simplification for parameter identification.
We first identified the conditions under which, in the current formulation of the model, biomass production and allocation in the plant are independent of its architecture. By contrast, we showed that some processes are strongly related to plant architecture, such as radial growth for trees. Two kinds of approaches were discussed to address the problem: (i) development of simplified models with different levels of variable aggregation and (ii) parameter identification for the complete model using simplified targets. We introduced three levels of model simplification according to the feasibility of the associated experimental protocols and to the classical set of data usually collected for forestry models. For each of these three levels, we theoretically studied the conservation of some key-variables of the model (including the ratio of biomass production to demand of the plant) from one simplification level to another one. On this basis, we proposed simplified equations based on aggregate variables and consistent with the GreenLab approach.
In practice, data concerning the plant architecture are often available, even if not with the same level of details as in simulations. These data may be obtained from sample measurements or from preliminary botanical analysis. Therefore we developed methods to identify not only functional but also topological parameters of plants, depending on the version of the model (deterministic, stochastic or deterministic with feedback of the plant trophic state on its development) and depending on the nature of available data. The applications presented in this thesis mainly concern trees but also other kinds of branched plants.