Twitter: An Early Disaster-Warning System

IEEE report makes a case for use of the social network in emergencies

11 September 2013

When a natural disaster strikes, time is of the essence. Sometimes all you have are a few seconds to make a life-saving decision—perhaps the amount of time it takes to read a tweet.

In Indonesia, government agencies are looking to Twitter to help inform the public of weather-related emergencies. Because earthquakes can often lead to the formation of tsunamis in Southeast Asia, announcements of quakes and their activity can provide the time necessary for an evacuation.

The country’s early tsunami warning systems are designed to notify government agencies first, which in turn informs the public. In the report “Twitter Early Tsunami Warning System: A Case Study in Indonesia’s Natural Disaster Management” published in the IEEE Xplore Digital Library, Twitter was shown to be just as valuable a tool to alert citizens as traditional communication channels like radio and television. Two researchers from Australia—Akemi Takeoka Chatfield and Uuf Brajawidagda from the School of Information Systems and Technology in Wollongong—analyzed data from Twitter to understand how the agencies are using the platform to warn citizens and the impact it had during the last three earthquakes that struck the coast of Sumatra in 2010, 2011, and 2012.

Indonesia is the fifth most active country on the social network. To track and analyze the enormous amount of data, the researchers used an application called Tweepy, which calculated how often government agencies informed the public through Twitter and the messages’ reach. For example, when an earthquake struck the region on 11 April 2012, Indonesia’s Meteorological, Climatological, and Geophysical Agency posted this tweet: “Tsunami Early Warning in Bengkulu, Lampung, Aceh, West Sumatra, North Sumatra, Earthquake Mag: 8.9 SR, April 11 2012 15:38:29 West Indonesian Time, Location 2.31 LU,9.” The 14 warning messages that were sent out on the platform were retweeted (or forwarded) 2132 times, with a maximum retweet of 8291 for just one of them. In other words, one message had the possibility of reaching tens of thousands of locals. Despite the somewhat fleeting nature of a tweet, all it takes is a few words to understand it’s time to act. 

A downside of the current early-detection warning system is that it relies on alarms, that sometimes do not work, to ring when a natural disaster hits the area. That alarm also notifies an agency official, who might not be in the office during the emergency to send warning tweet to the public. The report suggests that even if a more advanced early disaster-warning system could be built in-house by the government, the complexity of the system and lack of expert users would reduce its usefulness. Using existing social media channels, the researchers say, can provide government agencies with the benefits of “financial affordance and ease of use during extreme disaster events.” OK  

However, Twitter has its drawbacks. Due to its short messaging system of 140 characters, some tweets can be unclear. For example, the wording of one warning had confused residents so much that they were unsure whether a tsunami was to occur. Another shortcoming is that with the social network there is no systematic way of sharing information among all agencies, which can increase the amount of time a warning is posted. After the last earthquake in the region, it took nearly 8 minutes to release a warning on Twitter, which could be too late in some instances.

The report concluded that Twitter is a valuable and complementary tool to use as an early warning system; however, its impact could be increased if agencies worked closer together to improve the accuracy of messages, were in real time, and location-based. The researchers also suggest that agencies should investigate the users’ experience to improve how messages are worded and their timing.

What role should governments play in relaying information about emergencies on social networks? What would be some best practices for agencies to implement to ensure an effective warning system through these platforms?

Photo: iStockphoto

Learn More