Local Stress Drop Estimates of Strong Earthquakes in the South Iceland Seismic Zone

Abstract eng:
In earthquake prone regions where strong-motion data is scarce, the reliability of strong ground motion simulations from large earthquakes depends heavily on realistic source, path and site models. Due to the inherent uncertainty and randomness of earthquake processes and crustal heterogeneities, reducing the standard error associated with empirical attenuation relationships has proven difficult. Additionally, such relationships are strictly speaking not valid outside the range that the data defines. Therefore, it is advantageous to use physically based models, especially of the earthquake source, for ground motion simulation. The specific barrier model (SBM) provides a complete, yet parsimonious and self-consistent description of the faulting processes that are responsible for the generation of high-frequency waves (> 1 Hz). It is especially versatile and has, when modeling the earthquake as a point-source, been applied in the context of the stochastic modeling approach and random vibration theory. Moreover, modeling the earthquake on a finite-size fault, the SBM has also been applied in more physically realistic strong-motion hybrid-simulations, both in the near-fault as well as far-field region of a finite earthquake source. While previous studies have successfully simulated key parameters of strong-motion and complete time histories, the uncertainty of the source parameters has not been quantified. A key source parameter of the SBM is the local stress drop which drives the slip on the fault. In this study, using the strong-motion database of Iceland, the most seismically active region in northern Europe, as a case-study, moderate-to-strong Icelandic earthquakes (77 records of 6 earthquakes in the South Iceland Seismic Zone with MW ranging from 5.1 to 6.5) were modeled by the SBM and their local stress drops estimated. While the SBM provides a simple but physically meaningful representation of the faulting process the Bayesian method gives mathematically well-defined answers to the question what we can learn about the stress drop value distribution from the given data. Therefore, the model parameter uncertainty was quantified within the Bayesian statistical framework, employing a Markov chain Monte Carlo (MCMC) method with the Metropolis algorithm. To set up the Bayesian probability system, the data were corrected for path and site effects to obtain derived source spectra for each earthquake. In order to avoid trade-offs with the local stress drop, the high-frequency diminution parameter 𝜅 was determined independently using a new automated scheme developed in this study. The results show that the uncertainties of local stress drop are constant for the earthquakes analyzed, and are lognormally distributed. Median values of interevent local stress drops are consistent with the exception of one event, the far-field data of which are contaminated by ground motions from triggered earthquakes. The results for the earthquakes in the South Iceland Seismic Zone are the first step towards applying a physically consistent earthquake source and ground motion model in strong-motion simulations from earthquake scenarios in North Iceland where the lack of data preclude such calibrations.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket