← Back to Sessions

Caution! Dangerous Intersections Ahead: Understanding Human Behavior, Traffic Patterns, Protective Actions and Accidents

Description: 

While earthquake early warning systems are technologically advanced, there is a gap in empirical evidence regarding how vehicle operators physically react to sudden shaking and to Earthquake Early Warning (EEW) alerts prior to the onset of shaking. The U.S. Geological Survey ShakeAlert Project, in partnership with the U.S. Department of Transportation Volpe Center and the University of Oregon, developed a collaboration to answer critical questions about individual human driving behavior in relation to protective actions. The project included an analysis of traffic conducted in a shaking environment where EEW alerts were distributed to drivers. Understanding driver reactions to EEW alerts and earthquake shaking is critical for reducing secondary accidents during and immediately following an earthquake. The project used a collection of videos collected from publicly available social media platforms depicting earthquake-related human behavior. Several hundred real-world videos were captured during an active earthquake, with the majority being from the April 3, 2024, M7.4 Hualien, Taiwan, earthquake. A subset of these video recordings provides rare, firsthand visual evidence of operator behaviors during earthquakes and EEW alerts.
The research team then analyzed and classified driver behavior during seismic events by using an object-detection algorithm to efficiently review hours of footage and identify "transportation-related activities," focusing specifically on vehicle cabins, road-facing views, and operator interactions. Another round of manual review was conducted to categorize specific reactions:
o Response to physical shaking.
o Response to audible/text alerts via mobile devices (EEW).
o Latency between alert/shaking and defensive driving maneuvers (braking, pulling over, etc.).
Findings from this study aim to determine whether in-vehicle cellphone alerts trigger immediate safety actions or cause temporary distraction, as well as develop recommendations for the timing and "voice" of audible alerts to optimize driver compliance and safety in relation to earthquake events.

Session: Advancing Earthquake Early Warning: Science, Technology and Engagement in the U.S. and Beyond - III

Type: Oral

Date: 4/16/2026

Presentation Time: 02:45 PM (local time)

Presenting Author: Emma J. Fox

Student Presenter: No

Invited Presentation: 

Poster Number:

Authors