Disasters cause damages estimated in many billions of dollars and many lives lost every year. Disaster management is a challenging task due to the seemingly unpredictable alterations of the environment and impact on people. With the primary focus on the application of information technology and Big Data, this award establishes a "virtual institute" for Global Research on Applying Information Technology to Support Effective Disaster Management (GRAIT-DM). It will foster research collaborations and community activities, with the goal of improving our preparedness for, response to, and recovery from disasters. Led by partners at the Georgia Institute of Technology and the University of Tokyo, the virtual institute is a U.S.-Japanese cooperative effort that should grow to become a global collaboration in the near future.
Information technology has transformed modern disaster management, as demonstrated by Twitter which was a valuable information source during the Tohoku Earthquake. A Big Data-based approach to disaster management research can be both transformative and challenging, at both human and social scales. Reflecting this, the GRAIT-DM project supports the collection of large data sets from environmental sensors and information networks shared by many researchers working on various aspects of disaster management. This virtual institute promotes global research on the application of information technology by engaging the big data producers (e.g., sensor networks researchers), big data consumers (e.g., disaster management researchers), and big data managers (e.g., data analytics researchers) who connect the big data producers to consumers. Concrete activities include community-building workshops in the U.S. and Japan, outreach and publication of research reports, educational activities such as summer schools for graduate students and junior researcher exchanges, and a web portal to provide access to data and support for software tools for community use. Broader impacts include beneficial leveraging of international research and infrastructure investments, enhancement of on-going projects through cross-fertilization, an accelerated rate of innovation relevant to disaster management, and the development of a work-force with specialized talent, capable of excelling in a new, highly interconnected world that must cope with disasters.
This award has been designated as a Science Across Virtual Institutes (SAVI) award and is being co-funded by NSF's Office of International Science and Engineering.
Disaster management is a difficult task due to the unpredictable alterations of the environment by a disaster such as an earthquake and tsunami. Although some physical sensor networks can provide detailed information on specific kinds of disasters (e.g., USGS Global Seismographic Network on earthquakes), there are many disasters for which very little physical information is available. For example, landslides often are triggered by other disasters such as earthquakes and heavy rainfall, but there are few physical landslide detectors for landslides. A promising approach to obtaining sufficient information to handle unpredictable environmental alterations is the integration of social network information (e.g., Twitter) with physical sensor data. This integration of disparate big data sources presents significant systems level and data analytics research and development challenges. The project created the SAVI (Science Across Virtual Institutes) called GRAIT-DM (Global Research to Apply Information Technology to Disaster Management). The first achievement of the SAVI is the building of an international collaborative research community to apply information technology, in particular, big data technologies and analytics to improve the disaster management globally. The first collaborative effort is with Japan. As part of this effort, SAVI organized the Joint NSF/JST Workshop on Big Data for Disaster Management, which was held in Arlington, VA, May 23-24, 2013. The workshop brought together about 15 participants from each country. Afterwards, several workshop participants contributed to a report entitled "Big Data and Disaster Management: A Report from the NSF/JST Joint Workshop". This SAVI effort resulted in the US-Japan Big Data and Disaster Research program (BDD, NSF14-575), which was announced in June 2014. The report was cited in the synopsis of the BDD program announcement (NSF 14-575) and it is accessible online through the SAVI web portal [http://grait-dm.org]. The SAVI web portal is the main channel for publication of collaboration activities, sharing of research results, and dissemination of relevant information such as disaster-related big data sets, pointers to software toolkits, and papers. The second achievement of the SAVI is the development of a system level software support toolkit to enable the initial integration of physical and social sensor networks and allow the evolution of the participating sources in the integrated big data service. Our team developed the prototype LITMUS landslide information service demo program. LITMUS is collecting live information on landslides around the world and available for researchers and interested parties with near-real-time information on landslides. LITMUS is available on the SAVI web site for experimentation by our collaborators and the public. It currently combines near real-time information from the USGS earthquake monitoring system, the NASA TRMM rainfall monitoring system, and social networks such as Twitter, Instagram, and YouTube. Preliminary evaluation (based on December 2013 data) shows that LITMUS detected 25 out of 27 landslides reported by the USGS authoritative list of landslides, and in addition LITMUS found 40 other landslides that were not listed by USGS. The relevant information (tweets, Instagram photos, and YouTube videos) on all the landslides is stored and displayed on user command. The LITMUS landslide information service maintains a live demo on the SAVI web portal.
