A source study of the October, 2007 earthquake sequence of Morelia, Mexico and ground-motion estimation from larger earthquakes in the region

Abstract

We analyze an earthquake sequence of seven small events (2.5<Mw<3.0) which occurred in Morelia, Mexico during a 33-hour period in October, 2007 and was recorded by two local stations. Morelia lies in Central Trans-Mexican Volcanic Belt (CTMVB). The waveforms and spectra of the events are surprising similar, suggesting that their locations and focal mechanisms were nearly identical. Waveform inversion, with constraints imposed from mapped faults in the area (strike ~E-W, dipping to the north) yields a focal mechanism defined by , , and , consistent with reported focal mechanisms in the region. For these small events, the signal is lost in the noise at f<0.2Hz. For this reason, we estimate seismic moment, , from S-wave spectrum in the frequency band of 0.2≤f≤1Hz. Unfortunately, in this band significant amplification of seismic waves, caused by upper low-velocity volcanic rocks, is expected at all sites in the CTMVB. In the estimation of and interpretation of the observed spectra, we approximate the amplification by H/Z spectral ratio. Assuming an -source model, the observed spectra can be explained by the (Δσ, t*, ) triplets of (5 MPa, 0.02 s, 20 Hz) and (20 MPa, 0.03 s, 20 Hz), where Δσ is the Brune stress drop, and t*and are attenuation parameters. We use these parameters and EGF and RVT techniques to simulate ground motions for a postulated earthquake. The estimated horizontal PGAs and PGVs at the two sites in Morelia with Δσ=5MPa range between 23 and 46 cm/ and 1.5 and 3.52 cm/s, respectively. The predicted values are almost twice as large for Δσ=20MPa: 44-89 cm/ and 2.5-6.1 cm/s. Our estimated PGAs, especially that for Δσ=5MPa, are significantly smaller than those predicted from regression of world-wide data, suggesting either higher attenuation in this volcanic region and/or inadequate estimation of the site effect, and the attenuation parameters t*and f.