Project Proposed: This project, proposing research activities related to Gnome model predictions of the Gulf oil spill (within the CHMPR I/UCRC, to be located at UMBC, UCSD, and GaTech), aims to develop an instrument that acquires servers to collect, extract, locate, and process Gulf oil spill data from the existing social media sources such as Flickr, You Tube, Twitter and integrates them into a cloud. The data is collected from mobile devices and satellite sensors. As a result, the project provides instantaneous spatial distributions and temporal frequencies of oil slicks, tar balls, distressed and dead animals, along the complete coastline of Gulf Border States. The instrument is used to perform a 2-D VAR data assimilation using the Gnome oil spill forecast model as a first guess and the social media data as boundary conditions. The project will present the forecast oil slick dispersion products on the very large LCD tiled wall at UCSD for broadcast to thousands of viewers. System delivery is expected within 15 days. Improvement upon the NOAA operational Gnome model predictions of the Gulf is also expected. Moreover, the "human sensor web" data is likely to lead to more accurate operational oil dispersion forecasts for dissemination to decision makers through the 2-D VAR data assimilation; Broader Impacts: The work prototypes what in the future should become part of the responses to future event situations, either natural or anthropogenic. The "human sensor web" data is likely to lead to more accurate operational oil dispersion forecasts for dissemination to decision makers through the 2-D VAR data assimilation; hence the social aspects are strongly evident. This project involves students, especially minorities (at UMBC). Similar outreach activities are planned at Georgia Tech. Furthermore, UCSD will organize an internship program at a middle and high school.
In this work, we developed approaches to searching databases of large amounts of social media data to find relevant information for dynamic situations. We developed approaches to ingesting the above data into virtual world simulations. We continued to develop highly interactive, data rich, multi-user simulations that are able to be updated during runtime by data coming from external databases. We demonstrated that geophysical data can be automatically extracted from Social Media sources. We also demonstrated how this Geophysical data extracted from Social Media outlets can be ingested into an oil spill trajectory model. Then we showed how it can be used to improve the trajectory model forecasts and therefore deems very valuable for emergency responders in cases of natural and man-made disasters such as oil spills. Georgia Tech's team, in particular, worked to analyze hardware in mobile telephones that could be used to gather the social media data, and to off-load the processing of the data from large "heavy iron" servers in the back room to user handsets distributed throughout the areas of concern. Intellectual merit: It is both a new and quite promising approach to use social network data in order to track the course of a disaster such as the Gulf Oil Spill. We developed distributed algorithms that could run on anyone's smartphone to help in this. Broader impact: Helpling avert a disaster or, in the specific case of this project, track and contain a disaster, using technology and data to extract information from social networks helps all of us contribute to reducing the size of a disaster.
