Relating big data to local natural hazards: lessons learned from data mining the Twitter API for volunteered geographic information on earthquakes, wildfires, and prescribed fires in the contiguous United States

Document Type

Conference Paper/Presentation

Publication Date



New media are increasingly used to capture ambient and volunteered geographic information in multiple contexts, from mapping the evolution of the social movements to tracking infectious disease. The social media platform Twitter is popular for these applications; it boasts over 500 million messages (‘tweets’) generated every day from as many total users at an average rate of 5,700 messages per second. In the United States, Japan, and Chile to name a few, Twitter is officially and unofficially used as an emergency notification and response system in the event of earthquakes, wildfires, and prescribed fires. A prototype for operational emergency detections from social media, specifically Twitter, was created using natural language processing and information retrieval techniques. The intent is to identify and locate emergency situations in the contiguous United States, namely prescribed fires, wildfires, and earthquakes, that are often missed by satellite detections. The authors present their methodologies and an evaluation of performance in collecting relevant tweets, extracting metrics such as area affected and geo-locating the events. Lessons learned from data mining Twitter for spatiotemporally-explicit information are included to inform future data mining research and applications.

Publisher's Statement

© SIBERIAN STATE UNIVERSITY OF GEOSYSTEMS AND TECHNOLOGIES, 2015 © CZECH TECHNICAL UNIVERSITY IN PRAGUE, 2015 Publisher's version of record: http://www.isprs.org/proceedings/2015/2015-WG-IV-2/102_2015-WG-IV-2.pdf

Publication Title

Workshop Global Geospatial Information and High Resolution Global Land Cover/Land Use Mapping, 2015