Projected changes to precipitation extremes for northeast Canadian watersheds using a multi‐RCM ensemble

Abstract

This study focuses on projected changes to seasonal (May–October) single‐ and multiday (i.e., 1‐, 2‐, 3‐, 5‐, 7‐, and 10‐day) precipitation extremes for 21 Northeast Canadian watersheds using a multi‐Regional Climate Model (RCM) ensemble available through the North American Regional Climate Change Assessment Program (NARCCAP). The set of simulations considered in this study includes simulations performed by six RCMs for the 1980–2004 period driven by National Centre for Environmental Prediction reanalysis II and those driven by four Atmosphere‐Ocean General Circulation Models (AOGCMs) for the current 1971–2000 and future 2041–2070 periods. Regional frequency analysis approach is used to develop projected changes to selected 10‐, 30‐ and 50‐yr return levels of precipitation extremes. The performance errors due to internal dynamics and physics of the RCMs and those due to the lateral boundary data from driving AOGCMs are studied. The use of a multi‐RCM ensemble enabled a simple quantification of RCMs' structural and AOGCM related uncertainties in terms of the coefficient of variation. In general, the structural uncertainty appears to be larger than that associated with the choice of the driving AOGCM for majority of the precipitation characteristics and watersheds considered. Analyses of ensemble‐averaged projected changes to various return levels show an increase for most of the watersheds, with smaller changes and higher uncertainties over the southeasternmost watersheds compared to the rest. It is expected that increases in return levels of precipitation extremes will have important implications for water resources related activities such as hydropower generation in this region of Canada.