Fuzzy Risk Assessment of Flash Floods Using a Cloud-Based Information Diffusion Approach

Abstract

The assessment of flash flood risks is the basis on which flash flood prevention measures and early-warning systems can be proposed and reasonably implemented. However, inadequate monitoring of flash flood data which mainly occurs in remote areas restricts quantitative risk assessment. This study aimed to address the problems of limited data availability. A new model that integrating the Cloud-based Information Diffusion (CID) model and Analytic Hierarchy Process (AHP) approach is developed to assess flash flood risks. The proposed method improves the information diffusion function by introducing cloud model, while converting single-valued observations with incomplete information into fuzzy set-valued samples. The new approach can perform comprehensive risk assessment tasks with various indexes that are composed of incomplete information. Another advantage of this approach is that it can illustrate fuzzy uncertainty, fuzziness, and randomness of risk indexes by coupling with the cloud model, the risk magnitude is quantified with membership degree. A mountainous watershed in Southern China named Yangshan County was selected as the study area. Compared with information diffusion (ID) and entropy-based information diffusion (EID), the assessment results of the three models are, in general, consistent with one another, and agree with the reported values. According to the principle of maximum membership degree, Yangshan County is determined as a high-risk region when confronting with flash flood hazards. This model provides a feasible and effective method on fuzzy risk assessment of flash flooding, and is expected to be applied in other multi-criteria decision with limited available information or small sample problems.