On extreme ground snow depth and ground snow load at Canadian sites

Abstract

The roof snow load in the National Building Code of Canada is expressed as the sum of the snow and rain components. A probabilistic analysis of the components was carried out by using the recorded historical meteorological data. Probabilistic characterizations of the annual maximum ground snow depth S<sub>A</sub>, ground snow load S<sub>L</sub>, and snowpack bulk density were given. By considering the commonly adopted probabilistic models for S<sub>A</sub>, it was found that the use of the lognormal, Gumbel and generalized extreme value distributions for S<sub>A</sub> are preferred for 46%, 35%, and 19% of the considered meteorological stations. A similar observation was made for S<sub>L</sub>. Snow hazard maps in terms of return period values of S<sub>A</sub> and S<sub>L</sub> were developed. Also, the hazard maps of (winter) rain load, S<sub>R</sub>, were developed. It was found that the correlation coefficient between S<sub>A</sub> and S<sub>R</sub> is negligible; the implication of this is discussed.