|
Activity Number:
|
235
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics and the Environment
|
| Abstract - #306921 |
|
Title:
|
Spatial and Temporal Models for Evaluating IPCC Climate Model Outputs
|
|
Author(s):
|
Mikyoung Jun*+ and Douglas W. Nychka and Reto Knutti
|
|
Companies:
|
Texas A&M University and National Center for Atmospheric Research and National Center for Atmospheric Research
|
|
Address:
|
Department of Statistics, College station, TX, 77843,
|
|
Keywords:
|
climate models ; model bias ; spatio-temporal process ; spherical process ; multivariate processes
|
|
Abstract:
|
There are extensive efforts to develop climate models to study climate change. We have about 20 climate models from the Intergovernmental Panel on Climate Change. The previous works with climate model outputs commonly assume climate model outputs are random samples from a symmetric distribution centered around the true climate. One of the most interesting problems to climate scientists and modelers is verifying the bias of climate models and how the biases of different models are correlated. We propose modeling the climate model outputs as spatio-temporal processes on sphere x time and focusing on the spatial and temporal covariance structure. We propose cov models as well as cross-cov models for pairs of climate model outputs. Then, we quantify model biases and classify climate models with common biases.
|
