A sensitivity analysis on main factors involved in the landslide dam phenomena

Abstract

This paper is about a statistical analysis and sensitivity analysis on the main factors directly involved in the landslide dam phenomena, meaning, for “sensitivity analysis”, “... the study of how the variation (uncertainty) in the output of a mathematical model can be apportioned, qualitatively or quantitatively, to different sources of variation in the input of the model ...” (SALTELLI et alii, 2008). With this aim these parameters have been linked each other with a dimensionless index which describes the degree of correlation through the landslides and the river-alluvial flat. This index has been first purposed by CENCETTI et alii (2001) and it is based on the definition of geometrical and morphological characteristics of the two interacting processes (landslide and river processes). The software used to simulate the evolution of the two processes in time is a Cellular Automata model (CAESAR - COULTHARD et alii, 2007). This model is able to predict the morphological evolution of the river (at reach or basin scale) resulting from sediment supply variations and it implements a sediment transport model able to repro- duce the morphological changes. Recently the code has been updated (DE ROSA, 2008) to consider the eventual sediment supply from a bank (e.g. landslide). Whit this last update it is possible to model the landslide dam phenomena. Considering the parameters showed by CASAGLI & ERMINI (1999), let’s choose to analyse the factors: landslide volume, landslide duration (to evince the velocity parameter) and liquid discharge of the intercepted river. From the series of simulation done using CAESAR, a significant events sample has been extracted. This sample has been processed in statistical mode to calculate the sensitivity of each parameter on the others; so the weight of each single parameter in land- slide dam phenomena has been evaluated. This specific paper aims to dispose a method useful to establish the probability of occurrence of landslide phenomena which can produce a natural dam accurately examining all the parameters involved in this process; this way is opposite to the great part of the methods now present which aim to predict the probability of occurrence of dam break (evaluating its stability). So it’s clear that this method works at a prediction level different from the others.