Resampling methods for meteorological and climatological data analysis
Alexander Gluhovsky, Purdue University, West Lafayette, IN; and E. Agee
Conventional statistical methods are frequently based on unrealistic assumptions, regarding data sets under study. This includes the assumption of a linear model for the observed time series, as well as the assumption that observations follow a normal distribution. In reality, however, some meteorological variables have a strong evidence for non-normality and many observed time series exhibit nonlinear features. It will be discussed in the talk as to how modern resampling methods become instrumental in obtaining reliable inference from meteorological and climatological time series without making questionable assumptions about the data generating mechanism.
Extended Abstract (88K)
Session 9, Statistical Climatology
Thursday, 2 February 2006, 8:45 AM-12:30 PM, A304
Browse or search entire meeting
AMS Home Page