Abstract: Are estimation algorithms applicable for disaster managements? - This is a serious question for the authors who conducted a national research project named ‘G-space platform project’. Objectives of G-space platform project are planned to develop an innovative disaster management systems for Japanese government. This paper presents the experimental demonstration for the final evaluation of this project.

Many governmental stakeholders (i.e. firefighters, police officers, city government officers and so on) would like to start their actions (rescue, support citizens, etc.) immediately when a serious disaster like a huge earthquake occurred. However it is difficult due to lack of information to understand the disastrous situation.

The proposed innovative disaster management systems consolidates highly various and real-time data to share such information in many stakeholders. Moreover, the system will provide estimation data using data fusion technologies to provide more information. A serious problem however exists: is such estimated and uncertain data useful even for the disastrous situation?

The authors actually developed an integrated demonstration system to confirm usability. The system implements following technologies: people number estimation using cell-phone connection logs and simulation search using few real-time data. The authors set a virtual disastrous scenario for the demonstration: huge earthquake (M7.3) comes to Tokyo. By the huge earthquake, many buildings will collapse, public transportation will be stopped, and most lifeline will be broken. The city governments can obtain a few information of such situation, so their first task is collection of information. The information collection takes over 30 minutes – thus the officer will not be able to start their action for at least 30 minutes.

The authors assume that the estimated data is useful for the decision making in such poor-information situation. The demonstration system estimates the location of the evacuating people. Moreover, the demonstration system find out most suitable “spread of fire” simulation scenario for the current situation, from a huge spatio-temporal database.

The demonstration system was actually presented to 39 governmental officers to review availability of the demonstration system, on Jan. 27^th -28^th, 2016. The authors classified their comments into 6 categories (applications, rules, technologies, usability, data and promotion). The category of the most comments (5 comments) was “data” category. The most comments were relevant to variation of data. Such system should handle very various data, so it was confirmed that functions to import various data are highly required. The authors additionally conducted a questionnaire for each organization (9 organizations gave their answers). The answer to the question about whether the technologies applicable for disaster management or not were ‘Yes’ by 7 organizations. This result showed that the availability was confirmed.

Acknowledgement: This work is supported by consignment research and development of techniques about use and application of real-time information in G-space platform from the Ministry of Internal Affairs and Communications, Japan.