Colloquium: Philippe Naveau

HOW TO REVISE  RETURN PERIODS FOR RECORD EVENTS IN A NON-STATIONARY CONTEXT

Philippe Naveau

Laboratoire des Sciences du Climat et de Environnement

Date and time: 

Friday, October 16, 2015 - 3:00pm

Location: 

ECCR 245

Abstract: 

Breaking a record simply means that the current observation exceeds all past measurements. Such a type of an event is regularly followed to a media frenzy and, in such instances, climatologists are often asked if the frequency of this record is different from previous ones. 

This leads to the question of Detection and Attribution (D&A)   ("Detection" is the process of demonstrating that climate has changed in some defined statistical sense, without providing a reason for that change and "Attribution"  is the process of establishing the most likely causes for the detected change with some defined level of confidence, see the IPCC definition). 

The field of statistics has become one of the mathematical foundations in D&A studies because computing uncertainties represent   difficult inferential challenges when analyzing small probabilities.  In our case, we leverage methods from probability theory, in particular the molding of records and extreme events. 

In this context, we will give a brief overview on the main statistical concepts underpinning the D&A and proposes new methodological approaches to  revise return periods for record events in a changing climate.  We will show the advantages of our method throughout theoretical results and simulation studies.