Case Study: U.S. Geological Survey achieves near-real-time earthquake detection and rapid characterization with Elastic (Elasticsearch & Kibana)

U.S. Geological Survey - Customer Case Study

The National Earthquake Information Center (NEIC) team, led by Michelle Guy and Paul Earle, built a social-media driven monitoring capability to address gaps in traditional seismic systems: instrument data can be delayed by wave travel times and sparse station coverage leaves many felt events undetected or slow to characterize. The challenge was to provide faster, global awareness of felt earthquakes—especially in poorly instrumented regions—using noisy, multilingual human reports on Twitter.

Their solution ingests tweets matching a multilingual keyword list, filters and geocodes posts, and generates time-series detectors and locations, with outputs pushed as texts, emails, maps and Kibana dashboards. Over July 2013–March 2014 the system produced ~3 tweet-detection events per day, detected 90% of events within 2 minutes and located 90% within 200 km, handled magnitudes from ~M1.4–M8.2, and maintained a low false-trigger rate (~9.5%), proving valuable as an independent, rapid indicator and for rapid impact characterization (e.g., Napa, Chile, Indonesia) while being integrated with Elasticsearch/Kibana for richer situational awareness.

U.S. Geological Survey

Michelle Guy

Computer Scientist