Hutchings, Lawrence

We present a physically based methodology to predict the range of ground-motion hazard for earthquakes along specific faults or within specific source volumes, and we demonstrate how to incorporate this methodology into probabilistic seismic hazard analyses (PSHA). By ‘physically based,’ we refer to ground-motion syntheses derived from physics and an understanding of the earthquake process. This approach replaces the aleatory uncertainty that current PSHA studies estimate by regression of empirical parameters with epistemic uncertainty that is expressed by the variability in the physical parameters of the earthquake rupture. Epistemic uncertainty can be reduced by further research.We modelled wave propagation with empirical Green’s functions. We applied our methodology to the 1999 September 7 Mw = 6.0 Athens earthquake for frequencies between 1 and 20 Hz.We developed constraints on rupture parameters based on prior knowledge of the earthquake rupture process and on sources within the region, and computed a sufficient number of scenario earthquakes to span the full variability of ground motion possible for a magnitude Mw = 6.0 earthquake with our approach. We found that: (1) our distribution of synthesized ground motions spans what actually occurred and that the distribution is realistically narrow; (2) one of our source models generates records that match observed time histories well; (3) certain combinations of rupture parameters produced ‘extreme,’ but not unrealistic ground motions at some stations; (4) the best-fitting rupture models occur in the vicinity of 38.05!N, 23.60!Wwith a centre of rupture near a 12-km depth and have nearly unilateral rupture toward the areas of high damage, which is consistent with independent investigations.We synthesized ground motion in the areas of high damage where strong motion records were not recorded from this earthquake. We also developed a demonstration PSHA for a single magnitude earthquake and for a single source region near Athens. We assumed an average return period of 1000 yr for this magnitude earthquake and synthesized 500 earthquakes distributed throughout the source zone, thereby having simulated a sample catalogue of ground motion for a period of 500 000 yr. We then used the synthesized ground motions rather than traditional attenuation relations for the PSHA.

We test the physically-based ground motion hazard prediction methodology of Hutchings et al. [Hutchings, L., Ioannidou, E., Kalogeras, I., Voulgaris, N., Savy, J., Foxall, W., Scognamiglio, L., and Stavrakakis, G., (2007). A physically-based strong ground motion prediction methodology; Application to PSHA and the 1999 M = 6.0 Athens Earthquake. Geophys. J. Int. 168, 569–680.] through an a posteriori prediction of the 26 September 1997, Mw 6.0 Colfiorito (Umbria–Marche, Italy) earthquake at four stations. By “physically-based” we refer to ground motion synthesized with quasi-dynamic rupture models derived from physics and an understanding of the earthquake process. We test five hypotheses proposed by Hutchings et al. [Hutchings, L., Ioannidou, E., Kalogeras, I., Voulgaris, N., Savy, J., Foxall, W., Scognamiglio, L., and Stavrakakis, G., (2007). A physically-based strong ground motion prediction methodology; Application to PSHA and the 1999 M = 6.0 Athens Earthquake. Geophys. J. Int. 168, 569–680.] that support application of the methodology to physically-based probabilistic seismic hazard or risk analysis. We use two methods to test the hypotheses. First, we test whether observed records fall within the 68% log-normal confidence interval for the distribution of absolute acceleration response (AAR), pseudo velocity response (PSV), and Fourier amplitude spectra (FFT) created by a suite of source models. We also used the godness of fit between synthesized seismograms to verify whether at least one of the source models in the suite generates seismograms that match the observed waveforms, and if good fits to seismograms are due to source models that are close to what is actually known about the Colfiorito earthquake. We tested the hypotheses with a range of source parameters proposed by Hutchings et al. [Hutchings, L., Ioannidou, E., Kalogeras, I., Voulgaris, N., Savy, J., Foxall, W., Scognamiglio, L., and Stavrakakis, G., (2007). A physically-based strong ground motion prediction methodology; Application to PSHA and the 1999 M = 6.0 Athens Earthquake. Geophys. J. Int. 168, 569–680.]. We synthesized records from 100 rupture scenarios that were generated by a Monte Carlo selection of parameters within the range. This range was based upon having some prior knowledge of where the earthquake would occur. Observed values of AAR, PSV and FFT fit within the 68% confidence interval for all four stations, and one of the models generated seismograms that had a good fit compared to the observations. Moreover, a strict test for validating a physically-based ground motion hazard prediction methodology is that as more information is known about the source, the uncertainty of the prediction should narrow, but still include the actual ground motion. Then, we tightened the source parameters to be centered about the known parameters for the Colfiorito earthquake, and allowed for less uncertainty in their values. We found this to be true for this test. While the 68% confidence interval narrowed from a factor of ± about 4 to ± about 2 for the distributions, observed values of AAR, PSV and FFT still fit within the distributions for all four stations. Ultimately, we have calculated peak ground velocity (PGV) and peak ground acceleration (PGA) for all the synthetic seismograms obtained from the computed scenarios, and we have found that they are comparable with the actual and with those from the attenuation relation. We conclude that the methodology of Hutchings et al. [Hutchings, L., Ioannidou, E., Kalogeras, I., Voulgaris, N., Savy, J., Foxall, W., Scognamiglio, L., and Stavrakakis, G., (2007). A physically-based strong ground motion prediction methodology; Application to PSHA and the 1999 M = 6.0 Athens Earthquake. Geophys. J. Int. 168, 569–680.] is promising in giving ground motion hazard prediction estimates.

Seismic attenuation is generally thought to be a constant, or a simple monotonic function of frequency, and generally not a function of time. Examples of exceptions include attenuation enhancement due to shallow earthquake-induced damage, and fluctuations due to fluid diffusion. In reality, seismic attenuation fluctuates continuously in time at all frequencies, and the presence of cracks, their density and connectivity, as well as the presence and saturation of fluids, play a central role in defining such behavior. Due to multiple mechanisms, the crack density within a fault’s damage zone varies throughout the seismic cycle. Moreover, non-tectonic stress loads, seasonal or tidal, can change the crack density of crustal rocks, and leave detectable signatures on seismic attenuation. A strong signature can also be left on the crustal attenuation by a stress transfer from a nearby fault. Here we show that attenuation time histories from the San Andreas Fault (SAF) at Parkfield are affected by seasonal loading cycles, as well as by 1.5–3 year periodic variations of creep rates, consistent with published results that documented a broad spectral peak, between 1.5 and 4 years, of the spectra calculated over the activity of repeating earthquakes, and over InSAR time series. After the Parkfield mainshock, the modulation of seismic attenuation is clearly correlated to tidal forces. Opposite attenuation trends are seen on the two sides of the fault up to the M6.5 2003 San Simeon earthquake, when attenuation changed discontinuously, in the same directions of the relative trends. Attenuation increased steadily for over one year on the SW side of the SAF, until the San Simeon earthquake, whereas it decreased steadily on the NE side of the SAF, roughly for the 6 months prior to the event. Random fluctuations are observed up to the 2004 M6 Parkfield mainshock, when rebounds in opposite directions are observed, in which attenuation decreased on the SW side, and increased on the NE side. Another example of changes of attenuation with time is given for the Geysers geothermal field, where a large data set of earthquake recordings from a dense temporal deployment are analyzed and results are given in terms of 1/Q(f,t).