An effective strategy for combining variance- and distribution-based global sensitivity analysis. (December 2020)

• Record Type: Journal Article
• Authors: Baroni, Gabriele; Francke, Till
• Is Part Of: Environmental modelling & software. Volume 134(2020)
• Publication Date: 2020-12
• Subjects: Global sensitivity analysis -- Variance -- Distribution -- Generic sampling design; Environmental monitoring -- Computer programs -- Periodicals; Ecology -- Computer simulation -- Periodicals; Digital computer simulation -- Periodicals; Computer software -- Periodicals; Environmental Monitoring -- Periodicals; Computer Simulation -- Periodicals; Environnement -- Surveillance -- Logiciels -- Périodiques; Écologie -- Simulation, Méthodes de -- Périodiques; Simulation par ordinateur -- Périodiques; Logiciels -- Périodiques; Computer software; Digital computer simulation; Ecology -- Computer simulation; Environmental monitoring -- Computer programs; Periodicals; Electronic journals; 363.70015118
• DOI: 10.1016/j.envsoft.2020.104851
• Languages: English
• ISSNs: 1364-8152

Abstract

Abstract: We present a new strategy for performing global sensitivity analysis capable to estimate main and interaction effects from a generic sampling design. The new strategy is based on a meaningful combination of variance- and distribution-based approaches. The strategy is tested on four analytic functions and on a hydrological model. Results show that the analysis is consistent with the state-of-the-art Saltelli/Jansen formula but to better quantify the interaction effect between the input factors when the output distribution is skewed. Moreover, the estimation of the sensitivity indices is much more robust requiring a smaller number of simulations runs. Specific settings and alternative methods that can be integrated in the new strategy are also discussed. Overall, the strategy is considered as a new simple and effective tool for performing global sensitivity analysis that can be easily integrated in any environmental modelling framework. Highlights: A new strategy to perform global sensitivity analysis is developed. Variance- and distribution-based approaches are effectively combined. Main and interactions effect are estimated from a generic sampling design. The new strategy converges faster than Saltelli/Jansen formula.