A web application to facilitate crop model comparison in ensemble studies. (November 2017)

• Record Type: Journal Article
• Authors: Hossard, Laure; Bregaglio, Simone; Philibert, Aurore; Ruget, Françoise; Resmond, Rémi; Cappelli, Giovanni; Delmotte, Sylvestre
• Is Part Of: Environmental modelling & software. Volume 97(2017)
• Page Start: 259
• Page End: 270
• Publication Date: 2017-11
• Subjects: Crop model behaviour -- Plant processes -- Simulation experiment -- Non-parametric statistical methods -- Model clustering; 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.2017.08.008
• Languages: English
• ISSNs: 1364-8152

Abstract

Abstract: Crop models are reference tools that can be used to evaluate the performances of cropping systems under current and future agro-climatic scenarios. A recent trend is the adoption of multi-model ensembles, as crop model responses vary across pedoclimatic contexts. We present the web application MOBEDIS, aimed at investigating the causes of differences in crop models' behaviour. MOBEDIS combines non-parametric statistical methods (Spearman correlation, Random Forest, Hierarchical clustering, Mantel statistics) to analyze and cluster crop models according to the relationship between final outputs (e.g., yield) and a set of intermediate outputs related to plant processes. We applied MOBEDIS in three case studies to (1) discuss its capability to facilitate the understanding of the behaviour of two crop models in a simulation experiment, and (2) prove its applicability for model ensemble studies. MOBEDIS is freely available and ready-to-use for understanding single model responses and identifying groups of crop models sharing similar behaviour. Highlights: The application evaluates the relationship between crop models' final and intermediate outputs. It can analyze single model responses and cluster a set of models. Four non-parametric methods are used to avoid assumptions on variable distributions. Three cases studies are presented to demonstrate the application capabilities. Final results provide insights into differences between models' behaviours.