Sensitivity analysis to help individual plant model parameterization for winter oilseed rape

摘要

Functional Structural Plant models can help to understand plant plasticity. Such a model was developed for winter oilseed rape, based on the GreenLab model. However, the model had a large number of parameters, which made its calibration difficult. In order to simplify the model, we applied sensitivity analysis method to select the most important parameters. Analyses were performed on the following outputs: biomasses of the whole plant, of leaves, internodes and pods. First, the model linearity was computed with the Standardized Regression Method, revealing two phases of high nonlinearity. Second, total sensitivity indices were computed for each parameter with the Sobol method. We deduced that five parameters out of 46 had a prevalent impact on model outputs, while the other had a limited influence and could be fixed to reference values.