Potential of real- time high-rate scientific, consumer, and crowd- sourced GPS for earthquake early warning and rapid response

Sarah MinsonCalifornia Institute of Technology

Real-time finite fault slip models for earthquake early warning

Sarah E. MinsonCalifornia Institute of Technology

Jessica R. Murray, John O. Langbein, Joan S. GombergU.S. Geological Survey

What happens after an earthquake?Real-time:

EEWLocationMw

Shortly thereafter:Network response

Better locationBetter MwSource mechanism

Much, much later:Research/source studies

Fault geometryFinite fault slip modelKinematic rupture model

Why not just do this in real-time?

Moment saturation

0 20 40 60 80 100 120 1404

5

6

7

8

9

Tohoku EEW

Time (sec)

Mag

nitu

de

Mw 9.0

Final

Gavin Hayes and David Wald

Let’s Play a Math TrickUse Bayesian inference with a conjugate prior to derive:1. Analytical expression for PDF describing all

possible slip distributions2. Analytical expression for the probability

associated with any fault geometry

Inversion DesignAssume an epicenter location from seismic EEWConsider the set of faults centered at the epicenter with any strike and dip

Build discretized fault with total extent large enough to encompass rupture Only include offsets from stations within expected P-wave range

No moment saturation with GPS

Seismic EEW

GPS

PTWC

True Mw Evolution

Tsunami early warningBest Slip Model Real-time Uniform

Seaflooruplift

Faultslip

ConclusionsReal-time finite fault source models require only existing information:

EEW locationReal-time high-rate GPS streams

These models are not limited by computational expense. They are limited by the time it takes for the rupture to grow in size and information about the rupture to propagate

Take-Home MessageFor regions with both dense seismic networks and real-time high-rate GPS, we could have non-saturating magnitudes and finite fault slip distributions (and thus accurate warnings and shaking forecasts) in real-time starting today.