A STATISTICAL FRAMEWORK FOR TREE-RING RECONSTRUCTION OF HIGH-MAGNITUDE MASS MOVEMENTS: CASE STUDY OF SNOW AVALANCHES IN EASTERN CANADA

This paper investigates the application of the peaks-over-threshold method in combination with fitting of the generalized Pareto distribution for estimating the frequency of high-magnitude geomorphic events, on the basis of tree-ring data. In particular, attention is focused on extreme value distribution of tree-ring responses and in the minimum threshold or index number required to assure the past occurrence of high-magnitude events. The objective is to set a threshold, adapted to the sample distribution, which will make the distinction between high-magnitude years and remaining years, where the response of sampled trees is too weak to be attributed to a high-magnitude geomorphic event. The statistical framework proposed is based on the statistics of extremes. Through the use of simple, strong and effective mathematics, this model should strengthen the position of dendrogeomorphology in the evaluation and management of natural hazards.