Downscaling Precipitation Extremes from Regional Climate Model Outputs

• 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): Eric Laflamme*+ and Ernst Linder and Yibin Pan
• Companies: University of New Hampshire and University of New Hampshire and University of New Hampshire
• Keywords: Extreme value distribution ; Bayesian estimation ; NARCCAP ; return levels ; New England ; Precipitation

Abstract

There is a great societal interest in assessing the impacts of projected climate change on infrastructure design. Such impact assessment typically requires future projections of high-resolution time series of relevant climate variables. We extend a statistical downscaling method based on the work of Kallache et al. (2011) which applies a CDF transfer function to local-level daily precipitation extremes (from NCDC station data) and corresponding NARCCAP regional climate model (RCM) outputs to derive local-scale projections. Our extension consists of a semi-parametric mixture model for precipitation, where extremes are modeled parametrically using generalized Pareto distributions, and non-extremes are modeled non-parametrically using quantiles. This downscaling method is performed on 58 locations throughout New England to derive local level 25 year return levels. We implement the downscaling in a Bayesian estimation framework to obtain uncertainty estimates of future return levels. Results are generally larger than those obtained from a previously applied parametric Bootstrap procedure, indicating that projected trends to be less significant than is hinted at in many studies.