Assessing landslide susceptibility by applying fuzzy sets, possibility evidence-based theories

Abstract

A landslide susceptibility model was developed for the city of Manizales, Colombia; landslides have been the city’s main environmental problem. Fuzzy sets and possibility and evidence-based theories were used to construct the mo-del due to the set of circumstances and uncertainty involved in the modelling; uncertainty particularly concerned the lack of representative data and the need for systematically coordinating subjective information. Susceptibility and the uncertainty were estimated via data processing; the model contained data concerning mass vulnerability and uncer-tainty. Output data was expressed on a map defined by linguistic categories or uncertain labels as having low, me-dium, high and very high susceptibility; this was considered appropriate for representing susceptibility. A fuzzy spec-trum was developed for classifying susceptibility levels according to perception and expert opinion. The model sho-wed levels of susceptibility in the study area, ranging from low to high susceptibility (medium susceptibility being mo-re frequent). This article shows the details concerning systematic data processing by presenting theories and tools regarding uncertainty. The concept of fuzzy parameters is introduced; this is useful in modelling phenomena regar-ding uncertainty, complexity and nonlinear performance, showing that susceptibility modelling can be feasible. The paper also shows the great convenience of incorporating uncertainty into modelling and decision-making. However, quantifying susceptibility is not suitable when modelling identified uncertainty because incorporating model output information cannot be reduced into exact or real numerical quantities when the nature of the variables is particularly uncertain. The latter concept is applicable to risk assessment.

Keywords

• landslide
• susceptibility
• uncertainty
• possibility theory
• evidence-based theory