Functional ANOVA of Precipitation Extremes from Regional Climate Model Output

• Activity Number: 241
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Ernst Linder*+ and Yibin Pan and Eric Laflamme
• Companies: University of New Hampshire and University of New Hampshire and University of New Hampshire
• Keywords: Generalized Pareto Model ; NARCCAP ; Climate Change ; Daily Precipitation ; Extreme Values ; Bayesian Hierarchical Modeling

Abstract

Phase 2 of the North American Regional Climate Change Assessment Program (NARCCAP) provides an ensemble of outputs from 6 regional climate models (RCMs) whose boundary conditions originate from 4 global circulation models (GCMs) in an arrangement of a balanced incomplete block design. We exploit this block design for regional analysis of extreme daily precipitation in the case of New England. In particular we seek a variance stabilizing transformation of projected extreme precipitation characteristics, such as m-year return levels for m=10,25, 50, etc., which are derived via the Xcdft downscaling method (see Kallache et al. 2011) at NCDC station locations. The block design enables us to quantify main and interaction effects of RCM's and GCM's, as well as effects of geographic variables such as orography and nearness to the ocean. Further it enables us to derive a spatial optimal weighting scheme of RCM outputs over the region, which then provides minimum error estimates of future extreme precipitation characteristics. These, in turn, will help to improve impacts assessment as well as climate-change resilient infrastructure design.