ALBERT

Description: Two hypotheses were tested for the origin of the Isabella tomographic anomaly, which has been interpreted as either a lithospheric drip ( Jones et al. , 1994 ; Ducea and Saleeby, 1998 ) or a remnant of the Farallon plate, possibly attached to the Monterey microplate ( Wang et al. , 2013 ). P -wave receiver functions and travel-time residuals based on teleseismic events recorded by 41 stations were used to construct simple geometric tomography models of the Isabella anomaly and to test whether the Monterey microplate, or a similar fragment of the Farallon slab, can be connected with the location of the anomaly. The travel-time residual pattern was modeled with a slab geometry composed of a high-velocity rectangular block, and a lithospheric drip was modeled as a dipping cylindrical anomaly. Both gave statistically similar fits, but the cylindrical block appears less well constrained than the rectangular block. The top surface of the best-fit rectangular block was located at 50 km below the Great Valley and dips 65° northeast toward the Sierra Nevada, with 100 km thickness. West of the Isabella anomaly, receiver-function P -to- S converted phases from the top and bottom surfaces of the oceanic crust of the Monterey microplate were modeled as late-arriving negative–positive dipole signals. Such dipoles are observed at times that place their origin at 10 km below the Moho at the Coast Ranges. These converted phases from the oceanic crust could also be traced from the Coast Ranges to the Great Valley, where the top surface of the slab model is located. The combined result of the receiver functions and a geometric tomography model is consistent with the fossil slab hypothesis, which initiated as a flat subduction system that later went through a delamination process. Online Material: Figures of receiver functions, sorted by back azimuth and ray parameter at 31 stations.

Description: Seismic network optimization for Earthquake Early Warning (EEW) differs from the more general mission of regional monitoring. A network designed for EEW must prioritize speed of detection, the magnitude and rate of occurrence of likely earthquakes, and the location of seismic sources relative to population centers. A recent analysis by Böse et al. combines these elements in an end-to-end analysis for EEW monitoring in Switzerland. They sample from the combined instrumental and historical catalog for the region to evaluate hazard and losses. We have developed a similar approach based on gridded USGS National Seismic Hazard Map (NSHM) inputs. Hazard maps include catalog and historical seismicity, but also synthesize geologic and geodetic slip rates, geophysical inputs and paleoseismic event information. They are also extensively reviewed. The US NSHM maps are extrapolated from hazard curves calculated on a grid. We extract from the hazard curves the return time at a given shaking intensity, e.g. PGA=0.1 g, as a proxy for how often a station at that location will contribute to EEW detection from nearby earthquakes. The ground motion intensity corresponds to an approximate magnitude and location distribution that can be used in a forward calculation with attenuation to find impact on population centers. Stations near active faults generally minimize detection time and maximize warning time. Stations with shorter return times contribute more often and are generally near larger earthquakes. The best station distribution will give the greatest number of people the most useful warning.

Description: Earthquake early warning (EEW) for California was initially conceived as a faster version of normal seismic network operations. For the Southern California Seismic Network (SCSN) this early view has been found to be well short of reality. We review lessons SCSN learned through implementation of the ShakeAlert EEW system on the west coast of the United States. EEW-based adaptations affect stations, telemetry, and data acquisition, processing and archiving. Station coverage in the SCSN was good at the start of EEW implementation but 5+ years of permitting and station installations have been required to reduce EEW detection times to target levels. SCSN started with an advantage not all ShakeAlert networks enjoyed as installed dataloggers could stream the 1-second data packets needed for low data latency. Telemetry priorities can differ between EEW and normal network operations. EEW can only use low-latency data; network operators want complete data to simplify archiving and can accept some latency. EEW only requires three channels of strong-motion data; networks prefer 6 channels, with velocity data. We found that during strong earthquake shaking some links could not support both and developed severe latency. To meet both requirements, telemetry links are now being re-engineered. Multicasting is needed to allow waveform data to be shared across multiple servers for redundant processing. EEW algorithms require separate waveform processing and trigger detection. Specialized algorithms are also required to estimate magnitudes from limited P-wave amplitude data. While similar in appearance, EEW places unique and unanticipated demands on our regional seismic network.