|
Activity Number:
|
484
|
|
Type:
|
Invited
|
|
Date/Time:
|
Thursday, August 10, 2006 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
International Indian Statistical Association
|
| Abstract - #304944 |
|
Title:
|
Bayesian Melding: an Application and Critical Assessment
|
|
Author(s):
|
James Zidek*+ and Zhong Liu and Nhu Le
|
|
Companies:
|
The University of British Columbia and The University of British Columbia and BC Cancer Agency/University of British Columbia
|
|
Address:
|
Department of Statistics, Vancouver, BC, V6T 1Z2, Canada
|
|
Keywords:
|
ozone ; Bayesian melding ; physical modeling ; spatial prediction
|
|
Abstract:
|
Adrian Raftery and his coinvestigators developed Bayesian melding as a method for combining grid cell outputs from a deterministic model with point measurements from a random spatial field made at monitoring sites. This paper will describe the method and present the results obtained by applying it to the hourly and weekly ozone field over the eastern United States. The data involves hourly grid cell outputs from the MAQSIP model and monitoring data from the AIRS database. Through simulation studies, we investigate topics such as how well spatial predictors based on the method work and whether the grid cell data helps to make the predictor more accurate (as well as better calibrated). We demonstrate the method on the data and describe extensions of the method to, for example, space-time fields.
|
