Assessing NARCCAP climate model effects using spatial confidence regions.

Seth McGinnis,Joshua P. French,Armin Schwartzman

doi:10.5194/ascmo-3-67-2017

Abstract

We assess similarities and differences between model effects for the North American Regional Climate Change Assessment Program (NARCCAP) climate models using varying classes of linear regression models. Specifically, we consider how the average temperature effect differs for the various global and regional climate model combinations, including assessment of possible interaction between the effects of global and regional climate models. We use both pointwise and simultaneous inference procedures to identify regions where global and regional climate model effects differ. We also show conclusively that results from pointwise inference are misleading, and that accounting for multiple comparisons is important for making proper inference.

Highlights

The behavior of future climate is of great interest because of its potential impacts on health, finance, government, and many other arenas
We have utilized four classes of ANOVA-related regression models to compare the effects of the regional climate models (RCMs) and general circulation models (GCMs) used by the North American Regional Climate Change Assessment Program (NARCCAP) on average summer temperature
The APL model assumes an additive effect for the RCMs and GCMs, but assumes the rate of temperature change is constant for all combinations of models

Summary

Introduction

The behavior of future climate is of great interest because of its potential impacts on health, finance, government, and many other arenas. One of the ways that these uncertainties have been explored is via large-scale atmosphere–ocean general circulation models (GCMs) These models seek to understand the relationship between various environmental factors and use the modeled dynamics to generate various responses at certain times in the future. Local inference and decision making is made difficult because locations in each grid cell can have very different local climate systems. In response to this difficulty, regional climate models (RCMs) have been used to make predictions on a much finer scale (≈ 50 km spatial resolution; Mearns et al, 2009). The coarse-scale GCMs are used to provide the environmental conditions at the boundary of the study area for the RCMs, and the RCMs are used to dynamically downscale and model climate behavior within the study area on a finer scale

Objectives

Methods

Results

Conclusion