Rock, NEHRP site classes A+B+C, Vs,30> 600m∕s, T < 0.2s. 1494 records (1563 in complete
Hard soil, NEHRP site class C, 300 < Vs,30≤ 600m∕s, 0.2 ≤ T < 0.4s. 995 records (786 in complete
Medium soil, NEHRP site class D, 200 < Vs,30≤ 300m∕s, 0.4 ≤ T < 0.6s. 360 records (284 in
Soft soil, NEHRP site classes E+F, Vs,30≤ 200m∕s, T ≥ 0.6s. 656 records (547 in complete dataset).
Classify stations for early data and for some K-Net stations from H/V response spectral ratios. Use site
terms derived in previous studies that account for nonlinear response (see article for details).
Select earthquakes that occurred at the interface between crust and the subducting slab or between mantle
wedge ad subducting slab at a depth of ≤ 50km. Also require: focal depth within 5km of subduction
interface defined by Slab1.0, reverse or thrust focal mechanism and dip angle for one of the nodal planes
is within 15∘ of dip of subduction interface.
Focal depths between about 5 and 49km with only 11 shallow events (798 records) with < 25km. Deep
events have focal depth ≥ 25km.
Dataset 1 includes 1222 records from 13 earthquakes with Mw≥ 7.1.
Plots intra-event residuals for Tohoku 2011 (Mw9.0) earthquake w.r.t. distance. Find that observed PGAs
are underestimated within 130km. Fit trend line (logey = gloger + ex + c) to residuals, which corrects
bias in residuals. Relate this observation to Moho reflections.
Note that predictions will be different for an event at the 25km boundary depending on whether model for
shallow or deep earthquakes is used. Try to derive model using continuous variables to avoid this problem
but could not obtain smooth predictions. Recommend that the average ground motions predicted from
both the shallow and deep models are used for earthquakes at the boundary.