|
Activity Number:
|
410
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Wednesday, August 5, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #304859 |
|
Title:
|
A Hierarchical Spatio-Temporal Zero-Inflated Model for Correlated Tornado Reports in the United States
|
|
Author(s):
|
Ali Arab*+ and Christopher Wikle and Scott Holan and Christopher J. Anderson
|
|
Companies:
|
Georgetown University and University of Missouri-Columbia and University of Missouri-Columbia and Iowa State University
|
|
Address:
|
32 and O Streets, Washington, DC, 20057,
|
|
Keywords:
|
hierarchical models ; environment ; climate ; Bayesian inference ; count data ; spatio-temporal models
|
|
Abstract:
|
Environmental count processes are often characterized by zero-inflation and complex spatio-temporal variability. Recently, there is increasing interest in correlating the count processes of severe events such as tornadoes to climate indices for inferential and prediction purposes. In this paper, a hierarchical model is considered to incorporate spatially and/or temporally-varying climatological effects and other covariate data on the underlying count process. We categorize tornado reports into "less damaging" and "more damaging" tornadoes, resulting in zero-inflated bivariate counts. Thus, a bivariate structure is considered to jointly model these correlated counts with non-linear effects. This method enables us to evaluate the effect of climatological processes on development of tornadoes and how the nature of such associations may differ for the categories of tornadoes under study.
|
