Due to limitations of currently used disaster warning systems, such as accuracy, time constraints, and many uncertainties, the timely evacuation of inhabitants and the reduction of the risk of casualties have become difficult issues for local governments to manage during typhoons or heavy rainfalls. While some researchers have investigated the evacuation decisions made by local governments or inhabitants, most previous studies have not explored the relative weights between each evacuation decision-making factor, and have lacked the data required to create a hierarchical structure necessary for evacuation decision-making. The present study establishes an evacuation decision-making model based on pairwise comparisons and an analytical hierarchy process model for local governments and inhabitants. The results not only offer a strategy for improved disaster prevention, but also provide the foundation for bettering existing disaster warning systems. The findings suggest that evacuation decisions made by different levels of local governments are significantly diverse. Evacuation decisions are also location dependent. Thus, the findings indicate that using only a single disaster warning system is insufficient, and support the establishment of an evacuation decision support system as one of the first priorities of future disaster prevention actions.
Rainfall-based warning systems are widely used as a means of evacuating inhabitants to prevent sediment disasters. However, considering that only 2.2% of Japanese local governments carried out evacuation orders after sediment disaster alerts were issued in 2008, it appears that the existing rainfall-based warning systems are neither effective nor taken seriously. Furthermore, the case study of Shiaolin village in Taiwan indicates that the existing rainfall-based warning systems may not be sufficient. In addition, most assessments of the effectiveness of warning systems have merely examined whether alerts were issued before a disaster occurred; the appropriate timing of alerts has not been thoroughly studied. Here, we examine the characteristics of the warning models and warning-issuing systems in Japan and Taiwan. We propose evaluation indices of the effectiveness of warnings, and focus on the shortcomings of rainfall-based warning models through case studies of disasters and several years of statistical data. Finally, we recommend improvements to disaster prevention strategies.
It is widely recognized that the effects of a phase shift of fine sediment in large-scale debris flows are likely to be large. Therefore, in numerical simulations, it is essential to describe fine sediments in the fluid phase, and not in the solid phase. Recently, the “Kanako” numerical simulator has been widely used for a variety of objectives, particularly because it has a graphical user interface. However, to date, there is no widely available numerical simulation model for large-scale debris flows that includes the effects of phase shifts. Here, we present a modified version of Kanako to describe this phase shift for fine sediment. In the new numerical simulator, which we refer to as Kanako-LS, we assume that sediments can be classified into two groups in terms of sediment diameter (fine and coarse), and define the critical diameter of the sediment (Dc) as the smallest diameter at which sediments behave as a solid phase. Then, we test the applicability of Kanako-LS using an example of debris flows triggered by a deep-seated rapid (catastrophic) landslide in Japan. Our results suggest that Kanako-LS may be useful for a variety of types of large-scale debris flow, particularly if the amount of fine sediment and the magnitude of the interstitial fluid turbulence are sufficient.