A Probabilistic Approach to the Prediction of Snow Loads

Abstract

The magnitudes of loads imposed by snow depend upon a number of climatological and meteorological variables and as a result exhibit marked variations geographically, due to local effects within a particular region, and with time. The snowload formation process, which depends both on the macro- and microclimates of such meteorological variables as the depth of the snowfall, the snowfall density, wind speed, air temperature etc., as well as, the size and geometry of particular roofs and the influence of their immediate environment, is discussed.A model of the snow load formation process based on a mass balance approach, which takes into account the deposition of snow by individual snowfalls and the depletion of the snow load by wind action and thermal effects, is introduced. The use of this approach requires the establishment of statistical descriptions of the various meteorological variables, as well as a knowledge of the physical process of snow accumulation and depletion for a particular roof. The statistical properties of some of the more important meteorological variables are discussed. Also presented are some model derived data of snow accumulation and depletion for particular roofs located in different terrain.It is shown that even relatively simple statistical descriptions of the relevant meteorological data and snow accumulation and depletion mechanisms can lead to realistic predictions of roof snow loads. Snow loads on a flat roof are generated by a digital simulation technique and compared with full scale observations. Annual extreme values of the simulated snow load process are presented and compared with currently specified design values. Comments are made regarding the practicability of this approach.