Broadband Strong Ground Motion Simulations of Large Subduction Earthquakes by A. A. Skarlatoudis, P. G. Somerville, H. K. Thio, and J. R. Bayless Abstract The great subduction earthquakes that occurred recently in Peru, Chile, and Japan have provided unprecedented information about the ground motions gen- erated by such earthquakes. The 23 June 2001 M 8.4 Peru earthquake was recorded at eight strong-motion stations; the 27 February 2010 M 8.8 Maule, Chile, earthquake was recorded at over 10 strong-motion stations; and the 11 March 2011 M 9.0 Tohoku, Japan, earthquake was recorded at more than a thousand stations and produced the most extensive dataset of recordings for any earthquake. For the first time, data are available to guide the generation of ground-motion simulations from great subduction earthquakes. Broadband ground-motion simulations can enhance the usefulness of the recordings of these earthquakes by providing a means of interpolating and extrapolating the re- corded data. Once they have been validated, broadband ground-motion simulations can be used for forward predictions of the ground motions of great subduction events in regions such as Cascadia, in which there are no strong-motion recordings of large subduction earthquakes. In this study, we test our ability to use a hybrid method to simulate broadband strong-motion recordings of megathrust earthquakes by demonstrating that our sim- ulations reproduce the amplitudes of the recorded ground motions without systematic bias. We use simulations to study the distribution of various intensity measures of ground motion caused by these earthquakes and to validate our ground-motion simu- lation method by comparing the simulated ground motions with recorded ground mo- tions as well as with empirical ground-motion prediction models. Introduction Ground motions from great subduction earthquakes make a large contribution to ground-motion hazards in areas located in the vicinity of large subduction zones. In the U.S. probabi- listic seismic-hazard maps, the subduction source in the Pacific Northwest region is modeled by earthquakes having magnitudes as large as M 9.2. Because these magnitudes are larger than any of the earthquakes on which the most re- cent empirical ground-motion models (e.g., Atkinson and Boore, 2003; Zhao et al., 2006; Abrahamson et al., 2015) are based, it is important to use all the available information from these recent earthquakes to provide insight into the nature of ground motions from such large events. The recent occurrence of great subduction earthquakes in Peru, Chile, and Japan provides the first glimpse at what the ground motions from such large earthquakes may be like. The 23 June 2001 M 8.4, Arequipa, Peru earthquake (Somer- ville et al., 2008) was recorded at eight strong-motion stations; the 27 February 2010 M 8.8 Maule, Chile, earthquake (Somer- ville et al., 2013a) was recorded at over 10 strong-motion sta- tions; whereas the 11 March 2011 M 9.0 Tohoku earthquake (Somerville et al., 2013b) produced the most extensive dataset of recordings for any earthquake, with more than 1000 record- ings from each of K-NET and KiK-Net networks operated by the National Research Institute for Earth Science and Disaster Prevention (NIED) of Japan. For the first time, data are avail- able to guide the generation of ground-motion scenarios from great subduction earthquakes. Broadband ground-motion sim- ulations can enhance the usefulness of the recordings of these earthquakes by providing a means of interpolating and extra- polating the recorded data. The broadband ground-motion simulation method used in this study is based on the work of Somerville et al. (1991), Somerville (1993), and Graves and Pitarka (2004, 2010). It is a hybrid technique that computes the long- and short-period ranges separately and then combines the two to produce a sin- gle time history using appropriate matched filters. At periods longer than 3 s, the methodology is deterministic and contains a theoretically rigorous representation of fault rupture and wave propagation effects. At periods shorter than 3 s, it uses an empirical representation of source radiation and scattering derived from the recordings of a smaller earthquake, which is combined with a simplified theoretical representation of wave propagation. The method is relatively simple to apply because the only earthquake-specific parameters needed as input are 3050 Bulletin of the Seismological Society of America, Vol. 105, No. 6, pp. 3050–3067, December 2015, doi: 10.1785/0120140322