Group A, Poster #021, Seismology

Setting Expectations for Earthquake Early Warning in a Large California Earthquake Using Replays of the February 2023 Türkiye M7.8 Earthquake

Savvas Marcou, Angela I. Lux, Andrei Akimov, & Richard M. Allen
Poster Image: 

Poster Presentation

2023 SCEC Annual Meeting, Poster #021, SCEC Contribution #12820 VIEW PDF
Earthquake early warning systems (EEWS) aim to deliver alerts to users prior to the arrival of strong ground shaking. They have been shown to have the potential to reduce deaths, injuries, and material damage to infrastructure and property, especially in large, rare earthquake events involving large ground motions. However, the rarity of such events means that the benefits to the general public and other societal stakeholders with access to EEWS are not often explicitly demonstrated. Here, we simulate what would be expected had a real-world existing EEWS alert delivery pipeline been operational in the region affected by the February 6, 2023 M7.8 Pazarcik earthquake. We use the waveform data ...from local networks to replay the event through the point-source algorithm EPIC, currently used in the US ShakeAlert EEWS. We then use the alerts produced by EPIC as inputs to a statistical simulation of real-time alert delivery performance by the MyShake platform, a smartphone app that delivers ShakeAlert messages to users on the US West Coast.

Despite a peak EPIC magnitude estimate of M6.4, we illustrate how EPIC’s rapid reporting (first alert at 5.1 s from origin time) combined with MyShake’s parallelized processing of alert delivery would result in long warning times for users in zones of strong (MMI V+) shaking. We show median expected warning times of >35 s in zones of MMI VIII+ shaking and ~40 s in the areas of most intense shaking (MMI IX) near Antakya, towards the southern end of the rupture.

We then use the M7.8 event replay results as a proxy for a similar San Andreas-Hayward rupture to set realistic expectations for EEW in California. Our replay illustrates that rapidly reporting point source algorithms allow useful alerting times in the areas of strongest shaking, with warning times >10 s for the San Francisco Bay Area. Unsurprisingly, when only using EPIC’s underestimated magnitude the system would underalert, underscoring the need to combine the rapid EPIC point source algorithm with a finite source algorithm as is done in ShakeAlert. These results provide a mechanism to set expectations for public alerting in California and to increase public awareness of the value of EEW.