The aim is to propose a dynamic model of forest growth and biomass suitable to varied ecosystems with different species, soil types, climate conditions and forest managements. This model is combining different approaches (growth and yield, process-based and biogeochemical cycles) to take into account carbon, water and nutrient cycles and to include several processes such as wood production, transpiration, litterfall, litter decomposition or losses of nutrients by drainage. Such a model is necessary to anticipate and adapt forest management under different environmental and management scenarii (global changes). Considering the whole forest ecosystem, the seasonality of canopy dynamics and litterfall production is involved in key processes: photosynthesis and carbon production, stand transpiration and water cycle, litter decomposition and nutrient cycling. A dynamical probabilistic model for leaves demography has been created. This model is strongly constrained by environmental factors and is able to rank their influences. Model adjustment can highlight relationships between different scales of processes involved, from cellular-scale to stand-scale. The aim is to provide: i) a predicting model suitable to a large range of ecosystems, ii) hierarchical analyses of the environmental processes driving canopy dynamics.