Développement Et Évaluation Du Modèle de Surface ORCHIDEE

Résumé du livre

This thesis is to evaluate and develop a land surface model ORCHIDEE (Organizing Carbon and Hydrology In Dynamic EcosystEms). In the first part, ORCHIDEE on multiple timescales is evaluated by a novel methodology linking Artificial Neural Networks to Singular System Analysis. The joint analysis of observations and simulations uncovers the characteristics of model bias at and across timescales in different plant function types and climate groups, which provide references for future ORCHIDEE developments. To have a throughout understanding of snow effects on present and future carbon fluxes, ORCHIDEE with a decent snow model should be developed. In the second part, a site-synthesis analysis of winter ecosystem respiration and its controls across eddy covariance sites in mid- and high-latitude regions has been conducted. The result corroborates the role of winter respiration in annual carbon budget, and snow effects on winter respiration could be indirectly observed by its insulating effect on soil. The standard ORCHIDEE snow model is a simple bucket model and has been shown to bias snow simulations. In the third part, internal snow processes (e.g. snow melting/refreezing; water infiltration between snow layers) were thus developed in ORCHIDEE. Moreover, a new forest albedo parameterization was also implemented. In the third and final parts, this newly developed snow model has been validated based on both site and continental levels, and a significant improvement has been seen in terms of snow pack properties examined. This new snow model coupled with permafrost will be used to explore high-latitude water and carbon dynamics in the future.