Investigating early earthquake rupture characteristics with borehole strainmeters

Abstract:

Recent studies have suggested the existence of weak determinism in the earthquake rupture process. This is the concept that the final magnitude of a large earthquake can be estimated using data early on after earthquake origination, and before the rupture process is complete across the entire fault. It is hypothesized that ruptures begin as disorganized, accelerating cracks, and then at some point transition to a more organized slip pulse propagating out from the hypocenter, with the properties of this slip pulse correlating to the final magnitude of the earthquake. As these properties can be ascertained from features of early seismic wave arrivals on displacement recordings, applications of this knowledge have the potential to improve the timeliness and accuracy of earthquake early warning. However, traditional measurement techniques can be problematic when analyzing for determinism due to magnitude saturation in seismic data and the high noise floor of geodetic data, which obscures P-wave arrivals. Here, we explore a new method of analyzing the early rupture process using borehole strain sensors instead; as these instruments can record both small and large perturbations in strain, we can both observe the P-wave arrivals and avoid saturation at higher magnitudes. We began with a homogeneous slip model, where slip proceeds outward from the hypocenter in a uniform ring with a set rupture velocity of 2.8 km/s, as in Goldberg et al. (2018). We modeled earthquakes with magnitudes ranging from M6.5-8.5 by setting average rise times for each event according to the scaling relation described in Melgar and Hayes (2017). We then used the SW4 wave propagation modeling code to forward-model strain observations and found that generally within 5-10 seconds of data arriving at each station, earthquakes were distinguishable by magnitude. Next, we examined real earthquake data from events >M5 in the 2019 Ridgecrest Sequence. These data come from horizontal-component borehole strainmeters and collocated broadband seismometers managed by the UNAVCO Network of the Americas (NOTA). In analyzing the earliest-arriving data (less than ~20s after the P-wave arrival, depending on the station and event), we observed that there are different regimes of growth over time, potentially indicative of different rupture modes.