Credal Networks for Hazard Assessment of Debris Flows

Debris flows are among the most dangerous and destructive natural hazards that affect human life, buildings, and infrastructures. Worldwide, these phenomena claim hundreds of lives and millions of dollars in property losses every year. Starting from the ’70s significant scientific and engineering advances in the understanding of the processes and in the recognition of debris flow potential have been achieved (for a review see Costa and Wiekzorek 1987; Iverson et al. 1997). Several experiments have been conducted in many laboratories with the aim to investigate the processes of generation, motion and deposition of debris flows. Similarly, field investigation focused on geologic and topographic characteristics of the source areas and, finally, the triggering variables (e.g., rainfall depth and intensity) have been analyzed in order to derive semi-empirical threshold values able to statistically describe debris flow initiation. Takahashi (1991) was the first to propose a comprehensive theory behind the mechanism of debris flow dynamics (Sec. 1.2.1). Despite these advances, it is widely recognized that human expertise is still a fundamental building block of hazard identification and hazard map construction. This is so in part because many aspects of the whole process are still poorly understood, and because the identification of model parameters is exposed in practice to high uncertainty making it difficult to apply models. In order to close the gap between theory and practice it seems necessary to merge theoretical models with qualitative human knowledge, and to recognize that uncertainty is a pervasive com-