Self‐contained local broadband seismogeodetic early warning system: Detection and location

Abstract

Earthquake and local tsunami early warning is critical to mitigating adverse impacts of large‐magnitude earthquakes. An optimal system must rely on near‐source data to maximize warning time. To this end, we have developed a self‐contained seismogeodetic early warning system employing an optimal combination of high‐frequency information from strong‐motion accelerometers and low‐frequency information from collocated Global Navigation Satellite Systems (GNSS) instruments to estimate real‐time displacements and velocities. Like GNSS, and unlike broadband seismometers, seismogeodetic stations record the full waveform, including static offset, without clipping in the near‐field or saturating for large magnitude earthquakes. However, GNSS alone cannot provide a self‐contained system and requires an external seismic trigger. Seismogeodetic stations detect P wave arrivals with the same sensitivity as strong‐motion accelerometers and thus provide a stand‐alone system. We demonstrate the utility of near‐source seismogeodesy for event detection and location with analysis of the 2010 Mw7.2 El Mayor‐Cucapah, Baja, California and 2014 Mw6.0 Napa, California strike‐slip events, and the 2014 Mw8.2 Iquique, Chile subduction zone earthquake using observatory‐grade accelerometers and GPS data. We present lessons from the 2014 Mw4.0 Piedmont, California and 2016 Mw5.2 Borrego Springs, California earthquakes, recorded by our seismogeodetic system with Micro‐Electro Mechanical System (MEMS) accelerometers and GPS data and reanalyzed retrospectively. We conclude that our self‐contained seismogeodetic system is suitable for early warning for earthquakes of significance (>M5) using either observatory‐grade or MEMS accelerometers. Finally, we discuss the effect of network design on hypocenter location and suggest the deployment of additional seismogeodetic stations for the western U.S.

Defining the Coseismic Phase of the Crustal Deformation Cycle With Seismogeodesy

“…Therefore, a seismic trigger is required to determine the onset of the coseismic phase. Ideally, this can be obtained by collocation (within ∼3–4 km) of strong motion accelerometers and GNSS instruments (Emore et al., 2007; Goldberg & Bock, 2017). A seismogeodetic combination of GNSS and accelerometer data (Bock et al., 2011) allows P wave detection from seismic velocities, while improving the precision of the displacement waveforms during seismic motion (Bock et al., 2004; Saunders et al., 2016).…”

Development of a Geodetic Component for the U.S. West Coast Earthquake Early Warning System

“…Where real-time GNSS receivers and accelerometers are collocated (defined as ≤ 1:5 km apart), these data can be synthesized to produce a seismogeodetic displacement time series that records the P-wave arrival, dynamic motions, and static offsets (Bock et al, 2011) with substantially lower noise than displacements calculated from GNSS data alone. Using data from accelerometers collocated with drilled-braced GNSS monuments, Goldberg and Bock (2017) resolved displacements sufficiently small to detect M w ∼ 5 events. Crowell et al (2013) found that a magnitude scaling relationship that used the peak displacement amplitude (Pd) measured during the first 5 s of seismogeodetic time series showed less magnitude saturation with similar latency compared with the one based on displacement from doubly integrated, high-pass-filtered accelerometer data alone.…”

SCALODEEP: A Highly Generalized Deep Learning Framework for Real‐Time Earthquake Detection

“…Earthquake detection is a fundamental step in a broad scope of seismological tasks (Aiken et al., 2018; Beaucé et al., 2019; Frohlich et al., 1982; Goldberg & Bock, 2017; Hauksson et al., 2017; Reverso et al., 2015; Simons et al., 2006; Skoumal et al., 2016; ten Brink et al., 2020; Yoon et al., 2017). Despite major improvements in seismic instrumentation, conventional detection approaches become increasingly challenging in view of (1) a significant increase in the volume of seismogram databases, (2) the low signal‐to‐noise ratios (S/Ns) associated with low‐magnitude earthquakes, which are often beyond the visual detectability, and (3) the occurrence of a massive number of earthquakes with small magnitudes.…”