Process level sensitivity analysis for complex ecological models

I propose a methodology of sensitivity analysis applicable to complex process-based models. In application and discussion of this methodology, a model is conceived as a mechanism that is composed by multiple interacting processes. In this approach, model structure is explicitly investigated by analyzing the relationship that exists between process interactions and the emergent properties of the model. The methodology focuses on sensitivity analysis at the process level; and studies how process level uncertainty propagates through the network of modeled processes, affecting the dynamics generated through the interacting processes as well as the final model outcome. The information provided by this analysis can be used to ensure that the way in which the model operates resembles the phenomena being modeled, when uncertainties in processes are considered. It allows evaluation of the quality of a model and increases the confidence in model results. Some of the advantages of this approach are that it allows handling high levels of complexity by abstracting information at the process level, it facilitates the interpretation of model behavior and it provides information that allows exploration of how uncertainty in processes might affect model output. I illustrate the method through an example using the vegetation model Biome-BGC.