Abstract: | We study path effects on prediction equations of pseudo‐velocity response spectra (natural period of 0.1–5.0 s) in northern Japan, where heterogeneous attenuation structure exists. The path effects have been examined by comparing the regression analysis results for two different prediction equations. The first equation consists of a single term of anelastic attenuation conventionally. The second equation consists of two terms of anelastic attenuation in consideration of the heterogeneous attenuation structure. In the second equation, we divide a source‐to‐site distance into two distances at the attenuation boundary beneath the volcanic front. The boundary is considered to separate the relatively high Q fore‐arc side mantle wedge (FAMW) from the low Q back‐arc side mantle wedge (BAMW). Strong motion records (hypocentral distances less than 300 km) from interplate and intraslab events with Mw 5.1–7.3 are used. Regression analysis results show that the standard errors are significantly reduced by the second prediction equation at short periods (0.1–0.5 s), whereas the difference in standard errors from both prediction equations is negligible at intermediate and long periods. The Qs values (quality factor for S‐wave) converted from two anelastic attenuation coefficients for the second prediction equation are remarkably similar to the path‐averaged Qs values for the FAMW and BAMW by other studies using spectral inversion method. From these findings, we conclude that the path effects on the prediction equation of pseudo‐velocity response spectra are satisfactorily accomplished by the second prediction equation. Copyright © 2009 John Wiley & Sons, Ltd. |