Sismologia

Demonstration of Earthquake Early Warning Using Total Displacement Waveforms from Real-time GPS Networks
Brendan W. Crowell, Yehuda Bock, and Melinda B. Squibb

Brendan W. Crowell, Yehuda Bock, and Melinda B. Squibb
Scripps Institution of Oceanography, La Jolla, California

INTRODUCTION
Earthquake early warning (EEW) systems have historically been based on traditional seismicinstrumentation to provide rapid location of the earthquake source and its magnitude.

ern U.S. (see http://sopac.ucsd.edu/maps/ for a list of these

http://pboweb.unavco.org/) and its predecessors PGGA,

these networks has been daily download of GPS phase and now being upgraded to provide even higher-frequency deforGPS receivers, with a latency of less than 1 s (Genrich and Bock complementary to seismicnetworks and can be applied to earth-

Romania (Wenzel Instrumentation includes broadband velocity and strong motion acceleration sensors, which measure dynamic ground motions with great accuracy (e.g. seismic early warning of events requires continuous monitoring of a suite of stations looking for changes in key parameters. Most algorithms use displacement when trying to estimate the potentialmagnitude of an event from the initial few seconds. e magnitude estimates are made on the ratio of the short-term average displacement to the long-term average and on the apparent frequency of the initial energy. With seismic data, displacement has to be obtained by an integration of the broadband sensors or a double integration of the strong motion sensors. Due to the bandwidth and the dynamicrange limits of seismometers, the accuracy of displacements so derived is poor. Seismic instruments are also a ected by sensor tilts, instruments. While the seismic measurements provide a powerful constraint on the much noisier GPS measurements, the GPS receiver, unlike the seismometer, measures displacement directly and never clips, making it particularly sensitive to large earthquakes, which arethe main focus of EEW. dynamic displacements (“GPS Seismology”— Nikolaidis ments (“Earthquake Geodesy”—Bock -

We follow the instantaneous positioning approach of Bock (2000), which was applied to GPS seismology by Nikolaidis (2001). In this method, dual-frequency integer-cycle phase ambiguities and station positions are estimated independently at each epoch for a network of GPS receivers. Werefer to the time series of changes in position at a single station, i.e., displacements observed during an earthquake, as a “total displacement waveform” (Figure 1). “Total” refers to the contributions of both dynamic and static displacements. In instantaneous positioning, the displacements are computed an event the displacements are referenced to a station in the network that is not deforming dueto the event, or that is being precise point positioned by some other method (e.g., Blewitt , 2009). Either way provides “absolute” total displacement waveforms for all deforming stations in the network with respect to a global reference frame. hours of 1-Hz instantaneous positions, with an order of magnitude less precision in the vertical. Genrich and Bock (2006)

in seismically active regions,e.g., Japan’s GEONET (Miyazaki

772 Seismological Research Letters Volume 80, Number 5 September/October 2009

doi: 10.1785/gssrl.80.5.772

"#%" "#", "#"+

-./0123456789:;:

"#"* "#") "#"( "#"' "#"& "#"$ "#"% "#"" !"#"% !"#"$ $""'#(" $""'#*( !"#"$ "#"$ "#"" "#") "#"'

$""(#""

$""(#$(

$""(#("

! Figure 1. Time series of 1-Hz total displacement waveforms of station LAND in theperiod before and after the 2004 Mw 6.0 Parkfield earthquake, computed in real time on a PC workstation located at the USGS facility at Carr Hill near Parkfield, using the method of instantaneous positioning. The inset shows dynamic and static displacements during the interval from 07:14 to 07:16 on Sept. 28. The vertical axis is the same in both graphs.

sion for a single displacement for...