Smartphone App Earthquake Detections: How Combined Analysis of Crowdsourced and Seismic Data Can Improve Performance of Existing Seismic Networks

Date: 4/26/2019

Time: 09:15 AM

Room: Elliott Bay

App earthquake detection is the latest crowdsourced-earthquake-detection method implemented at the EMSC to automatically identify felt events. It simply monitors concomitant LastQuake smartphone app launches. Compared to the 2 other methods in operation - Twitter earthquake detection and website traffic monitoring - it offers much more accurate geographical information and faster detections respectively. In practice, detections can be as fast as 20 sec and felt area be automatically mapped within tens of seconds of the propagation time to reach the most distant eyewitnesses, independently of any seismic data. Similarly to seismic networks, detection times are shown to depend on the number of users at close epicentral distances. App earthquake detections are exploited to reduce the latency of publishable seismic locations by combining their analysis with seismic data, a method named CrowdSeeded earthquake Location (CSLoc). Picks are automatically retrieved through a messaging system from GEOFON stations in a time and space window centered on the app detection, picks are then selected for consistency with seismic wave propagation before a seismic location is performed using the barycenter of app launches as seed. The process iterates until location is reliable. Reliable locations are available at global scale within 100s (median time), with a pick extraction delay of 30s, performance depending on the actual station coverage in the epicentral region. Beyond the presentation of the methods and its performances, we will outline future work in collaboration with the EarthquakeNetwork, an app that detects shaking on the internal sensors of smartphones.

Presenting Author: Remy Bossu

Authors