Magnitude scaling of early-warning parameters for the M_w 7.8 Tocopilla, Chile, earthquake and its aftershocks

摘要

We studied the correlation between the final event magnitude and four parameters obtained from the early portion of P and S phases for a set of high quality subduction events. These relationships are used in the framework of earthquake earlywarning systems for real-time magnitude estimation. The investigated parameters are the low-pass-filtered peak displacement (PD), the integral of the velocity squared (IV2), and the predominant and characteristic periods (τ_p and τ_c). We created a dataset from the continuous records of the first two weeks following the 14 November 2007 M_w 7.8 Tocopilla (Chile) earthquake. The dataset includes 69 events with magnitudes greater than 4, among them the main event (M_w 7.8), the main aftershocks of M_w 6.7 occurred on November 15, and 4 events with magnitude greater than 6. The low-pass-filtered PD read on short P-phase and S-phase windows is well correlated with the finalmagnitude, confirming previous results. Indeed when examining 2-s time windows of P waves, we did not observe any saturation effect for magnitudes greater than 6.5; however, there is a slope change in the regression curve. A similar result is obtained from the integral of squared velocity computed over short windows around P and S waves. The characteristic and predominant periods are correlated with magnitudes up to M_w 6; but they clearly do not scale with the magnitude for the stronger events. Our observations offer insight into the feasibility of an early-warning system in Chile.