Spatial Anova Modeling of High-Resolution Regional Climate Model Outputs in NARCCAP

• Activity Number: 523
• Type: Contributed
• Date/Time: Wednesday, August 3, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Emily L. Kang*+ and Noel Cressie
• Companies: Statistical and Applied Mathematical Sciences Institute and The Ohio State University
• Keywords: ANOVA ; atmosphere-ocean general circulation model (GCM) ; North American Regional Climate Change Assessment Program (NARCCAP) ; regional climate model (RCM) ; Spatial Random Effects (SRE) model

Abstract

We consider the current (1971-2000) and the future (2041-2070) average seasonal air surface temperature fields from two regional climate models (RCMs) driven by the same atmosphere-ocean general circulation model (GCM) in North American Regional Climate Change Assessment Program (NARCCAP). We analyze the differences between the future and current temperature fields and incorporate the factor of season and the factor of RCM. Noticing that classical ANOVA doesn't include "spatial" modeling and can't account for the spatial variability across the domain, we propose to use the Spatial Random Effects (SRE) model and carry out ANOVA in the reduced space. Using the SRE model enables us to model the spatial dependence through the spatial basis functions, and our analysis indicates that the interaction between season and RCM is very important. Additionally, with the SRE model, the computation associated with analyzing these high-resolution outputs from RCMs can be carried out efficiently, due to the fixed number of spatial basis functions and the resulting dimension reduction.