An all-red aurora captured in Independence, Mo., on October 24, 2011. Image Courtesy of Tobias Billings (nasa.gov)
On October 24, 2011, a geomagnetic storm raged across Europe and the United States and caught scientists by surprise. Arrival time forecasts were off by eight hours, predicted a much weaker storm, and did not anticipate the visible red aurora emission. The inaccuracy of this forecast is fairly typical in space physics today, as scientists do not have enough space weather observations to accurately forecast between the Sun and Earth.
During that same 2011 storm, citizens across the eastern United States and from Chicago to Alabama tweeted about the red aurora in real-time, making this an early example of large-scale documentation of a geomagnetic storm. The response on social media suggested a new way for scientists—like the Aurorasaurus team– to significantly improve aurora observations and forecasting, while engaging the public in advancing space weather science. Our team—Aurorasaurus— is capitalizing on the immediacy, convenience, and popularity of Twitter to improve forecasting of auroral displays, perhaps unleashing Twitter’s power as a forecasting tool for space weather events.
Twitter is an amazing medium for disseminating information quickly and in massive quantities from live commentary of ABC’s The Bachelor to funny #hashtag wars to celebrating the New Year. During New Year’s Eve in 2013, there were around 33,000 tweets sent per second as the midnight hour approached.
And it’s not farfetched to think these tweets could be used in a scientific context. Earlier in 2011, an earthquake in Virginia affected a long stretch of the East Coast— damaging the Washington Monument and the National Cathedral. The earthquake—the first earthquake in Virginia that was above 5.0 on the Richter Scale in over 100 years—also made its mark in the Twitter world. Word about earthquakes seems to spread more quickly via social media than official detection devices can report them.
The graphic shows the rapid proliferation of earthquake related tweets during the 2011 earthquake in Virginia that damaged the Washington Monument and the National Cathedral
While there have been earthquakes there (including several in 2003), this was the first substantial earthquake that occurred on the East Coast after the proliferation of smart phones and social media.
Our Aurorasaurus team finds itself in a similar situation as we are experiencing the first solar maximum during the social media boom. The solar maximum is a period of increased solar activity (e.g. sun spots) that occurs every eleven years. The solar activity can send charged particles towards Earth that can sometimes disrupt satellite and power systems, but sometimes they also create beautiful auroral displays. We are at the tail-end of the solar maximum, but continue to see many reports about beautiful auroral displays.
The Pros and Woes of Forecasting with Twitter
When the Aurorasaurus team began to think about how to use social media to improve forecasting the aurora more accurate and timely, Twitter was a natural choice. Each time a tweet is sent, it includes the exact time and exact location (when enabled) the user sending the tweet the send button. In addition, the tweet can also include a
picture or brief video— a tremendous bonus. Our system also tries to identify the place a tweet originated by looking for common place names in the text of the tweet, so you can help us out and tweet with your town’s name when reporting aurora.
After gathering the tweets, or data, savvy researchers can use the information to illuminate trends. Researchers at Hewlett-Packard Labs have discovered that you can actually use Twitter mentions to predict how well a movie will do in its first couple weekends of release.
The hard part of using Twitter as a forecasting tool is the incredible amount of data but also the incredibly varied quality of that data. Computer Scientists are currently evaluating all manner of ways to automatically evaluate Tweet content. While that research is still years away from providing anything fruitful for crisis response, there is another way to increase the accuracy of tweets – training users to identify higher quality, relevant tweets about large-scale events. Thankfully, our citizen scientists—all of you—are helping us achieve this at a record rate! So far over 25,000 tweets have been evaluated on Aurorasaurus to help find the “verified” sightings!
The Future of Space Weather Science and Twitter
As research about Twitter use during large-scale events has continued, one researcher has found that people using Twitter have become more aware that their tweets could be pulled by emergency responders. This has lead us to start to test our own abilities to share information during large-scale events like aurora sightings—and the possibility of developing an evaluation of Early-Warning-Systems in coming years.
For example, our system gathers tweets based on signal words (e.g “aurora”). Once the system notices a large number of “verified” tweets in a given area, it sends an email alert to registered Aurorasaurus users who are near that geographic location. Through this framework, space scientists will learn more about current auroral event, and emergency responders may also gain valuable insights into using social media to alert people about a pending event faster than official systems can.
What do you think?
- How would information you tweet about (or would tweet about) be used to predict a trend?
- What is another subject area (scientific or not) where the immediacy and wide breadth of Twitter may be useful?
- How could other social media platforms (e.g. Facebook, Instagram, Vine, FourSquare) be used to share real-time information about events, such as an aurora or earthquake? Would it be as effective as Twitter?