Emergency Shelter Geospatial Location Optimization for Flood Disaster Condition: A Review

Abstract

Today, the world is experiencing a tremendous catastrophic disaster that can lead to potential environmental damage. However, awareness of how to deal with this catastrophic situation still remains very low. One of the most critical issues in disaster response is assigning disaster victims to the best emergency shelter location. This article reviews various existing studies to develop a new approach to determining emergency shelter locations. There are four evaluation criteria that are reviewed: optimization objective, decision variable, methodology, and victim identification. From the investigation, there are two major evaluations that can be further developed. In terms of decision variables, most of the previous research applies direct distance (Euclidean Distance) in the analysis process. However, the application of travel distance can represent a real evacuation process. Another interesting point is the victim identification process. Recent research applies grid-based partitioning and administrative-based partitioning. However, this method leads to a bias in the assignment process. This article recommends the application of K-Means clustering method as one of the unsupervised machine learning methods that is rapidly developing in many engineering fields. For better understanding, an example of K-Means clustering application is also provided in this article. Finally, the combination of travel distance and K-Means clustering will be proposed method for any further research.