Effects of Nonstationarity of Extreme Wind Speeds and Ground Snow Loads in a Future Canadian Changing Climate

Abstract

Modern structural design codes assume a statistically stationary climate. However, it is now known that the climate is changing at a rate that calls into question the stationarity assumption. This study was undertaken to explore various methods of accounting for climate nonstationarity. The geographical domain of this study covers most of Canada, intending to explore how best to update future climatic variables in the National Building Code of Canada (NBC). Outputs from future climate simulations for North America by Environment and Climate Change Canada (ECCC) and the Pacific Climate Impacts Consortium (PCIC) were used, focusing on a representative concentration pathway (RCP) scenario, namely the RCP8.5 emissions scenario. Surface hourly mean wind speeds and ground snow amounts were extracted from the simulations and used to detect and quantify trends in the extreme climatic variables from 1950 through 2100. This study reviewed different concepts for evaluating climatic design variables under stationary and nonstationary climates. It was found that within the range of linear nonstationarity detected for grid points within Canada, the results predicted using these nonstationary methods are close. Therefore, a simpler approach, namely Minimax design life level, was adopted to estimate the design wind speeds for 1/50 and 1/500 annual probability of exceedance and the design ground snow loads for 1/50 and 1/1,000 annual probability of exceedance. Design values derived based on nonstationary approaches were compared with the results estimated using the conventional return period concept between 1986 and 2016 and a moving-time-window method between 2060 and 2090 centered at 2075.