Distance scaling of earthquake-induced ground motion is studied in the Erzincan region, located in the eastern part of the North Anatolian Fault zone. The data set used in this study consists of 170 aftershocks of the MS = 6.8 Erzincan earthquake of 13 March 1992, with moment magnitudes between 1.5 and 4.0. In order to empirically obtain the scaling relationships for the high-frequency S-wave motion, regressions are carried out on 352 horizontal-component short-period seismograms, all recorded within a hypocentral distance of 40 km, to empirically obtain the scaling relationships for the high-frequency S-wave motion.
Peak ground velocities are measured in selected narrow-frequency bands, in the frequency range of 1.0–16.0 Hz, and are subsequently regressed to define a piecewise linear attenuation function, a set of excitation terms, and a set of site terms. Results are modeled in the framework of random vibration theory, using a bilinear geometrical spreading function, g(r), characterized by a crossover distance at 25 km: g(r) = r–1.1 is used for r ≤ 25 km, whereas g(r) = r–0.5 is used for larger distances. An extremely low-quality factor, Q(f) = 40(f/fref)0.45, is used to describe the anelastic crustal attenuation in the region, consistently with the independent results of Akinci and Eyidogan (1996, 2000).
Excitation terms are well matched by using a Brune spectral model with stress drop Δσ = 10 MPa (taken from the recent literature, Grosser et al., 1998). An effective high-frequency, distance-independent rolloff spectral parameter, κeff = 0.02 sec, is obtained in this study. Peak ground acceleration predictions based on these parameters show a much more rapid decrease with distance than the relations usually used in Turkey, indicating that our results should only be applied to the Erzincan region itself.