Séminaire FDD-FiME // Z. M. Debaly

Chargement Évènements

« Tous les Évènements

  • Cet évènement est passé

Séminaire FDD-FiME // Z. M. Debaly

10 mars @ 14 h 00 min - 15 h 00 min

Zinsou-Max Debaly (CREST-ENSAI, EcoDep ANR Project) 

Title :  Does climate affect the intensity of defoliation in Québec ?  Adjacent-Category Autoregressive modeling approach

 Abstract : Due to its ecological and economical damages, spruce budworm outbreaks (SBW) become one of major research topics in forestry in Canada. There is a growing consensus among scientists that SBW outbreaks depend on climate conditions. In this paper, we investigate the effect of climate on the intensity of defoliation of white spruce in two regions of Québec : Témiscamingue and Matawinie. Autoregressive model for adjacent-category time series is then proposed. The model’s parameters are estimated via the maximum likelihood approach. We show that this estimator is consistent and asymptotically Gaussian distributed. Portemanteau test for goodness-of-fit is proposed. The quality of the proposed method is shown with small numerical experimentation for finite samples. Our method is then applied to real data consisting of defoliation levels in Témiscamingue and Matawinie recorded as categorical data with four levels and climate data. In Témiscamingue, precipitations reduce the intensities of defoliation. However, in Matawinie, when the variation of range of daily minimum temperatures in summer increases, the level of defoliation is high. However, it appears that there is no statistical evidence to conclude a difference between corresponding defoliation processes.

Keywords: climate variations, defoliation data, goodness-of-fit, multivariate analysis, ordinal categorical time series, spruce budworm outbreak.

 A Joint work with Olaloudé Judicaël, Franck Osse, Philippe Marchand and Miguel Montoro Girona.

 This work was funded by project ”EcoDep” PSI-AAP2020-0000000013 

Download slides


Date :
10 mars
Heure :
14 h 00 min - 15 h 00 min


Institut Henri Poincaré