|
Activity Number:
|
274
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 4, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics and the Environment
|
| Abstract - #305381 |
|
Title:
|
Bayesian Spatio-Temporal Downscaling Models for Local Climate Prediction
|
|
Author(s):
|
Paul D. Baines*+ and Xiao-Li Meng
|
|
Companies:
|
Harvard University and Harvard University
|
|
Address:
|
Dept of Statistics, Cambridge, MA, 02138,
|
|
Keywords:
|
Spatial statistics ; Downscaling ; Bayesian ; Climate change
|
|
Abstract:
|
We present a class of Bayesian models for making local level predictions of future climate, utilizing both point-source data such as monitoring stations and lower-resolution information such as regional climate models (RCMs) and global climate models (GCMs). While deterministic RCMs/GCMs can be successful at capturing large/mid-level trends, higher-resolution analyses are often of much more importance for regional planning. The question of how to combine information at different spatial resolutions has received much attention in the recent literature. By using a latent process representation we implement a spatial-temporal down-scaling model in the spirit of Fuentes and Raftery (2005) and Berrocal, Raftery and Gneiting (2008). Predictive properties of the model are investigated via a large-scale simulation, together with the results of an application to recent climate data.
|
