A global sensitivity analysis approach for marine biogeochemical modeling

Clémentine Prieur (1, 2) , Laurence Viry (1) , Eric Blayo (1) , Jean-Michel Brankart (3)

1 AIRSEA - Mathematics and computing applied to oceanic and atmospheric flows
2 GdR MASCOT-NUM - Méthodes d'Analyse Stochastique des Codes et Traitements Numériques
3 IGE - Institut des Géosciences de l’Environnement

This paper introduces the Sobol' indices approach for global sensitivity analysis, in the context of marine biogeochemistry. Such an approach is particularly well suited for ocean biogeochemical models, which make use of numerous parameters within large sets of differential equations with complex dependencies. This sensitivity analysis allows for a detailed study of the relative influence of a large number of input parameters on output quantities of interest to be chosen. It is able to distinguish between direct effects of these parameters and effects due to interaction between two or more parameters. Although demanding in terms of computation, such a tool is now becoming affordable, thanks to the development of distributed computing environments. An applicative example is presented with the Modecogel bio-geochemical model.

Mots clés: marine biogeochemistry; sensitivity analysis; Sobol' indices

Domaines: Equations aux dérivées partielles [math.AP]; Statistiques [math.ST]; Océan, Atmosphère

Clémentine Prieur, Laurence Viry, Eric Blayo, Jean-Michel Brankart. A global sensitivity analysis approach for marine biogeochemical modeling. 2018. ⟨hal-01952797v1⟩