CGRG Bibliography of Canadian Geomorphology
Search Results
Author : Floyer, J.; and Jamieson, B,
Date : 2008.
Title : Avalanche weak layer prediction from penetrometer force-resistance—depth profiles.
Publication : 4th Canadian Conference on Geohazards: From Cause to Management. May 20-24, 2008. University Laval, Quebec City, Quebec.
Issue :
Page(s) :
Abstract
Weak layer identification is an important part of avalanche forecasting. Digital, high-resolution snow penetrometers can aid weak layer identification by giving a rapid and detailed depth profile for certain snow properties, such as force-resistance. Presently, weak layer identification is done by a skilled human interpreter; this process is slow if many profiles need to be interpreted. This paper presents a framework for automated weak layer identification from profiles taken with a force-resistance penetrometer. The technique of spiking deconvolution is applied to the depth profile using one or more characteristic weak layer signatures as the input wavelet. First, the energy delay characteristic of the spiking filter is matched to that of the input wavelet. Then the deconvolved output is analysed for spikes that correspond to locations where similar weak layer shapes exist in the depth profile. This is done by relating the relative amplitude of the filtered signal at each point along the deconvolved output to a probability density function built using a training set of responses from known weak layers and non-weak layers. This gives a likelihood of a point within the depth profile being a weak layer and this can be combined using Bayes law with an estimate of the a-priori probability of finding a weak layer for cases in which the weak layer is assumed, or not assumed, to be present. Assumptions regarding the likely depth of the weak layer, the hardness of the weak layer and the hardness of adjacent layers can improve the posterior probability distribution.
Bibliography of Canadian Geomorphology