Detection of High-frequency Seismic Radiation from the Initiation and Termination of Asperity Rupture in the 2010 Chile, Maule Earthquake

The 2010 Chile, Maule earthquake (Mw 8.8, USGS) was located on the boundary of the Nazca and South American plate, where the slip distribution was well-constrained by a GPS network (Moreno et al., 2010). The rupture zone extends north-east and south-west from the epicenter along strike, which is the same for both the waveform and the geodetic inversion (Pulido et al., 2011; Moreno et al., 2012). However, the high-frequency beam power distribution obtained by the Back-projection (BP) method is inconsistent with the slip distribution inferred from these inversions; the former is concentrated at 25-50 km depth while the latter distributes in the upper 25 km (Lay et al., 2012). In general, the distribution of high-frequency energy radiation and the slip distribution are complementary to each other (Madariaga, 1976). The difference between the distribution of the beam power and the coseismic slip should be the clue to understand the source process of megaquakes. The BP method has been applied to many large earthquakes. But the resolution of the BP image may become worse, especially for shallow earthquakes, due to the effect of depth phases (pP and sP) . The Hybrid Back-projection method (Yagi et al., 2012), on the other hand, stacks the cross correlations of observed waveforms with the corresponding Green's functions. This process mitigates the bad effects of depth phases and improves the resolution of rupture imaging (Fukahata et al., 2013). Since the Green's functions strongly depend on the depth and focal mechanisms of the source point, in this study we construct a more realistic source model using a 3-D slab geometry model (Hayes et al., 2012), and then calculate the Green's function for each source point. We estimated the beam power distribution for two frequency bands: 0.1-0.5 Hz (F1) and 0.5-2.0 Hz (F2). We obtained almost the same rupture propagation for both frequency bands. There is no clear depth varying frequency-dependence. Two peaks of the beam power can be seen near and 200 km north-northeast of the epicenter. For the F2 dataset, the beam power distributes near the epicenter for the first 30 s. From 30 s to 70 s, the rupture starts propagating at about 60 km north-west of the epicenter toward north-east direction, and terminates at 200 km north-northeast of the epicenter with a strong peak of the beam power. Compared to the coseismic slip distribution, the distribution of the beam power surrounds the large southern and northern slip areas. Thus, we could say that the peaks of the high-frequency beam power represent the initiation and ending of the rupture of asperities. This feature is clearly seen in the northern part of the rupture. The discrepancy between the distribution of the high-frequency beam power and the coseisimic slip is understood from theoretical modeling, which show that the high-frequency waves are generated by the sudden rupture velocity change at the edge of the asperity (Madariaga, 1977). In the Maule earthquake case, the smooth and fast rupture propagation along the northern asperity seems to have contributed to the weak beam power at its center. Our study leads to the idea that the heterogeneity or/and the topographic property of an asperity may determine its rupture propagation characteristics.