On 2019-07-04 17:33:49 (UTC) a big earthquake (Magnitude 6.4) happened at 12 km SW of Searles Valley, CA. A second and stronger earthquake (Magnitude 7.1) happened in the same area, at 18 km W of Searles Valley, CA, on 2019-07-06 03:19:53 (UTC). We wondered about the spatio-temporal patterns of people reactions on social media, specifically Twitter.
This work has been done in collaboration with Jason Baumgartner at PushShift.
Left: density map of earthquakes in the area (USGS data) between 3rd and 8th July. Right: density map of tweets in the area (PushShit data) between 4th July and 7th July.
Spatio-temporal information about earthquakes is obtained through the IRIS catalog of the United States Geological Survey (USGS).
4,509 confirmed earthquakes with magnitude larger than zero have been identified in the area (longitude between -127.71 and -109.21; latitude between 30.73 and 40.73) between 3rd and 8th July.
Tweets were collected over 10 hours using Twitter's search API for terms associated with earthquakes (earth,quake,terremoto,sisma). Collection began after the 6.4 quake but before the 7.1 quake. Additional tweets were collected after the second quake. A total of 2,684,419 tweets satisfied our criteria.
Tweets were further filtered by their location: tweets without explicit geotagging were geocoded if some (less accurate) information about location was present. A total of 1,029,017 tweets with geolocation is used.
This collection has been further analyzed by extracting the sentiment of each tweet. This allows us to build a geographic map of emotions before, during and after earthquakes in the area.
Top-Left: density map of all tweets in Los Angeles. Top-Right: same for Las Vegas. Bottom-Left: same for Phoenix. Bottom-Right: same for San Francisco.
Density map of quakes around the epicenter of main activities.
Twittershakes: emotional response on social media to earthquakes