CMStatistics 2016: Start Registration
View Submission - CMStatistics
B1491
Title: On the hourly temperature data behaviour and the daily extreme temperature tail dynamics Authors:  Debbie Dupuis - HEC Montreal (Canada) [presenting]
Abstract: It is reasonable to think that changes in hourly temperature could hold information on daily extreme temperatures, both in terms of their frequency and their size. Investigating the extent to which this holds, for both daily maximum and daily minimum temperatures, is the purpose. The seasonal nature of temperatures, along with their evolution in time due to climate change, make their analysis challenging. We take a conditional peaks over threshold approach and establish time-varying thresholds to maintain stationarity. Our analyses show that hourly changes in temperature can be better or worse predictors of extreme temperature than daily measures, depending on the location and local climate dynamics. This is illustrated by the analysis of U.S. data over the 1973 to 2015 period.