Abstract:

Spatial configuration of urban emergency shelters is one of the hotspot issues in disaster prevention and urban emergency management. The purpose of this study is to conduct spatial configuration of urban shelters which can help urban residents have access to the emergency shelters as soon as possible with less congestion. The spatial configuration model is built on the basis of the agent-based model and multi-criteria decision-making method. Remote sensing image data, high-precision population data, road network data and expert knowledge are integrated in the model. Three types of agents involved in emergency evacuation are designed, which include the government agent, shelter agents and resident agents, to conduct evacuation simulation. A government agent can delimitate the service areas of shelters in accordance with the administrative boundaries and road distances between the positions of residents and the locations of the shelters. Shelter agents can select specified land uses as potentially available shelters for different disasters, generate the service areas of shelters, record the information of the residents in their service areas and do relative statistical work of evacuation processes. Resident agents have a series of attributes, such as ages, positions, and walking speeds. They also have several behaviors, such as reducing speed when walking in the crowd, and helping old people and children. Integrating these three types of agents which are correlated with each other, we can simulate evacuation procedures. The simulation results are utilized to support location-allocation and configuration of emergency shelters. The location-allocation method of this study is based on multi-criteria decision making and weight sensitivity analysis, so that the locations of new shelters can be selected in highly-suitable regions for location-allocation. When the new shelters are allocated, a new round of emergency evacuation simulation will be executed to realize loop optimization of location-allocation based on the new spatial distribution of shelters to generate the final spatial configuration plan. A case study in Jing'an District, Shanghai, China, was conducted to demonstrate the feasibility of the model. The simulation results convinced that the new model can provide detailed planning for spatial configuration of urban shelters, which can help the residents evacuate to nearby shelters as quickly as possible with less congestion risks. The model provides a new methodology to conduct high-quality location-allocation of urban emergency shelters. It can also be extended to conduct similar spatial configuration work in other urban regions for different kinds of emergency shelters.

Key words: emergency evacuation, agent, shelter, spatial location-allocation, GIS, Jing'an District, Shanghai