The U.S. earthquake early warning system has a surprising new data source: tweets.
According to a Twitter “#DataStories” blogpost released Wednesday, the U.S. Geological Survey’s National Earthquake Information Center — equipped with nearly 2,000 real-time earthquake sensors capable of detecting quakes and aftershocks around the globe — is turning to Twitter to seek confirmation of large-scale earthquakes.
Responsible for issuing alerts and confirmations of earthquakes that affect humans, NEIC records an average of 70 quakes per day. Many of these occur in isolated areas or are too insignificant to be felt, however, and seismologists found themselves in need of a verification system.
Twitter caught the attention of USGS in the aftermath of the catastrophic magnitude-8.0 earthquake that ravaged China’s Sichuan province in 2008, leaving 69,197 dead and 18,222 missing, and causing $146.5 billion in infrastructural damages. Researchers were shocked to discover that Twitter users were faster to report the earthquake than the USGS systems were.
Hesitantly, USGS began to monitor Twitter, cross-checking user reports of earthquakes with their real-time seismological data. What they discovered was surprising: The Sichuan trend was consistent. Twitter can typically trigger alerts in under two minutes. In one case, USGS issued an alert based on tweets only 29 seconds after an earthquake in Napa, California, began.
This technique, according to the blogpost, has numerous advantages. Since the USGS uses open-source software, Twitter monitoring comes at no extra cost. It also enables the NEIC to incorporate numerous languages into their data-sensing algorithms, bettering their ability to detect quakes in Asia or Oceania, far away from the mainly U.S.-based sensor array.
“…words used can be a clue as to the magnitude and location of the earthquake,” the post says. “Chile has two words for earthquakes: terremoto and temblor; terremoto is used to indicate a bigger quake.”
Twitter also helps to regulate false alarms: If the sensors are picking up major movement around a populated area, but no tweets are coming out of it, it’s likely that there isn’t actually an earthquake.
The next step for the USGS, according to the post, is to drop Twitter-data-based detections directly into seismic algorithms, creating near-instant quake alerts.