An Automated Procedure for Producing USGS ShakeMap Parametric Input from Non-Traditional Seismic Networks with Application to the Community Seismic Network

Abstract:

Regional seismic networks, such as the Southern California Seismic Network (SCSN), process earthquake waveform data to generate ShakeMap input parameters: PGA, PGV and response spectral accelerations at 0.3, 1, and 3 s. These networks typically consist of a mix of high-quality accelerometer and broadband sensors with well-tuned response and filtering algorithms. When using data from these sources, ShakeMap’s interpolation algorithms assume that the generated parametric values have no uncertainty. However, it is desirable to increase the number of ground motion recordings spatially by employing less expensive, lower-quality sensors. Data from non-traditional networks, like dense, community-hosted strong-motion networks such as the Community Seismic Network (CSN), can improve the resolution of ShakeMaps; however, their data must be processed using methods that are more network-specific. ShakeMap already has the ability to ingest macroseismic values, both traditional and “Did You Feel It?” (DYFI), employing assigned uncertainty values (a “nugget” effect) or computed uncertainties dependent on the number of DYFI responses, respectively, that allows each macroseismic “station” to be appropriately down-weighted as part of the interpolation process. With CSN, which employs MEMS sensors, or other Class-C sensor networks, the data must also be treated as uncertain, and we are establishing a protocol for employing such data in ShakeMap. We use CSN’s well-recorded 2019 Ridgecrest earthquake sequence to illustrate their utility, supplementing SCSN recordings in USGS’s ShakeMaps for these events, ultimately showing the increased resolution of amplification effects within the Los Angeles Basin. The challenge for routinely using CSN-derived data is establishing station-specific signal thresholds and corresponding station uncertainty values as a function of frequency for lower amplitudes. A procedure is presented to automate and streamline the data acquisition from CSN, employing USGS’s open-source software package gmprocess to generate parameters and station noise metrics so that ShakeMap can rapidly incorporate more-densely sampled CSN-type station observations.