Spatio-temporal modelling of extreme storms

• Published in: The Annals of Applied Statistics
• Main Authors: Economou, Theodoros, Stephenson, David B., Ferro, Christopher A. T.
• Format: Text
• Language: English
• Published: The Institute of Mathematical Statistics 2014
• Subjects: Bayesian hierarchical model; spatial random effects; natural hazards; extra-tropical cyclones; extreme value distribution; European windstorms; North Atlantic; North Atlantic oscillation
• Online Access: http://projecteuclid.org/euclid.aoas/1419001741; https://doi.org/10.1214/14-AOAS766

A flexible spatio-temporal model is implemented to analyse extreme extra-tropical cyclones objectively identified over the Atlantic and Europe in 6-hourly re-analyses from 1979–2009. Spatial variation in the extremal properties of the cyclones is captured using a 150 cell spatial regularisation, latitude as a covariate, and spatial random effects. The North Atlantic Oscillation (NAO) is also used as a covariate and is found to have a significant effect on intensifying extremal storm behaviour, especially over Northern Europe and the Iberian peninsula. Estimates of lower bounds on minimum sea-level pressure are typically 10–50 hPa below the minimum values observed for historical storms with largest differences occurring when the NAO index is positive.