Analyse de sensibilité du modèle mathématique

Abstract

Sensitivity analysis of the mathematical model “Erosion Productivity Impact calculator” (EPIC) by approach One-Factor-At-A-Time (OAT). Nowadays, mathematical models are very complex and have many input parameters. Moreover, all do not have the same level of influence on the outputs. In order to identify the most significant parameters of EPIC that will guide the user on the field and in the process of calibration of the model, a sensitivity analysis by the approach One- Factor-At-A-Time was carried out on two data files: one coming from the Programme-Action-Hesbaye” (PAH, 1996) a n d the other of the INCO project (2000). The method used consists in modifying each input parameter of the model by -10% and +10% around its initial value. The effect of each operated modification is analysed on 6 outputs of the model (nitrate leached, water percolation, water consumption, biomass, yield and mineralised humus) of which their sensitivity was quantified by calculation of an index called “Sensitivity Index” and by a percentage of variation. The outputs considered showed particularly sensitive to the parameters “Potential Heat Units” (PHU), “Fraction of nitrogen in plant at 50% of maturity” (BN2) but especially to the “Second point of the evolution of the LAI curve” (DLAP2). A variation of 10% of this parameter causes average changes of approximately 14% and 85% in the six outputs, respectively for the data files of the PAH and the Inco project.

Keywords

• One-Factor-At-A-Time
• sensivity analysis
• sensivity index
• percentage of variation
• erosion
• productivity