2017 Volume 82 Issue 739 Pages 2451-2459
When a large earthquake occurs, rescue operations and fire-fighting are obstructed by street-blockages. In order to reduce property/human damages caused by the delay in arrival at the disaster site, it is important to quickly collect, share and utilize disaster information among multiple users. In this paper, we developed a system that can collect and share information acquired by users in real time in the event of a disaster. The advantages of the system are summarized as follows: (1) Users can access to the system by using various kinds of information terminals such as mobile phones (Android/iOS) and personal computers because it was implemented as a Web application; (2) The system can run on a cloud server in several countries with high disaster resistance; (3) A simulation to predict the property damage by fire-spreading can be performed based on collected information on the location of building fires as an example of secondary usage of disaster information; (4) Persons in charge of collecting disaster information and the location with high possibility of damages are recommended to support users for effective information collecting.
Using this system, we conducted a demonstration experiment on the assumption that local volunteers collect information immediately after an earthquake occurs in Setagaya Ward, Tokyo. The participants were asked to post a virtual disaster site to the system which they discovered while walking around freely. The locations of virtual disasters were expressed with markers, which vary according to the type of disaster (building-collapse/street-blockage/fire-outbreak) and the distance from a user (i.e. invisible beyond 60 m, visible 30 m or more and less than 60 m, detail view less than 30 m). As a result of the experiment, about 0.7% of disaster information distributed in the whole of Setagaya Ward could be clloected by just a dozen people for 15 minutes. The collection rate can be further improved by increasing the number of participants in collecting disaster information. Additionally, we clarified factors related to the efficiency of collecting information and discussed policies for improving the system based on the rate of correct information out of all posted information, the movement log of participants, etc.
Next, we evaluated the effects by using the function to support collecting disaster information through another demonstration experiment. More specifically, recommended collection zones and persons in charge were calculated by the system and shown on a user's screen on the basis of the damage estimation by simulation. Comparing with the case that the support function was not used, the number of posted information increased by 2.2 times per person, and the travel distance required for discovering a disaster site decreased by 0.47 times. The result suggests that it is possible to perform more effective information collecting by reducing the duplication of traveling routes or improving the method for recommending collection zones.
Furthermore, we conducted an evaluation experiment to verify the system availability under the bandwidth limitation after a large earthquake occurs. Under the condition of a certain transmission speed (1 Mbps/15 Mbps), we measured the time required for synchronization of sharing disaster information among all users in 100 times. As a result, the time increased to less than 50 milliseconds while virtually increasing the number of users to 3,000 step by step. Therefore, it may be possible to support the activities requiring immediate response such as emergency vehicles. However, the connection state was sometimes unstable in case the number of users was more than 3,000 persons. For higher reliability of sharing information in real-time, it is necessary to improve the system from the viewpoint of load distribution.
