A crustal normal‐faulting earthquake (; Mw 6.7) occurred in eastern Tohoku, Japan, on 11 April 2011. K‐NET and KiK‐net stations recorded 82 records from within 100 km of fault rupture. These data and data from associated foreshocks and aftershocks will make a critical contribution to future improvements of ground‐motion prediction for normal‐faulting earthquakes.
Peak ground accelerations (PGA) and peak ground velocities (PGV) are compared with four ground‐motion prediction equations (GMPEs) that include the style of faulting as a predictor parameter. For distances under 100 km, and using a network average value of VS30, the average ratio of PGA to the selected GMPEs (the event term ) is high by factors of 2.3–3.7. Event terms for PGV are high by factors of 1.4–1.8. Adjusting PGA and PGV with customized site terms (Kawase and Matsuo, 2004a,b), the standard deviations of PGA and PGV residuals are reduced from 0.59 to 0.43, and from 0.53 to 0.35, respectively. The event terms decreased to relatively small factors of 1.1–1.8 for PGA and increased slightly to 1.5–2.0 for PGV. Thus, site terms are very important, but positive event terms remain. The remaining positive event terms are not explained by high stress drop, which was typical of crustal events of all mechanisms globally or in Japan. Two subparallel faults ruptured, but source inversions, which we reviewed, revealed that they ruptured sequentially, so simultaneous contributions from the two faults did not cause high motions. Although these observations may tend to suggest that ground motions in large normal‐faulting events are larger than predicted by the tested models, we are not aware of any observations from this event that contradict the precarious rock evidence of Brune (2000) that ground shaking is low on the footwall near the rupture.