11:00 AM - 1:00 PM
[SSS10-P05] Broadband source characteristics of the 2016 MW 7.6 Chiloé interplate earthquake off Chile
Keywords:interplate earthquake, Peru-Chile trench, source scaling relationship, large slip area, strong motion generation area
On December 25, 2016, an MW 7.6 interplate earthquake occurred at the interface between the South American and Nazca plates beneath Chiloé Island in southern Chile. The strong-motion seismograms of this event were recorded by the observation network of the University of Chile. We used these strong-motion data and constructed a broadband source model of this event. This study can contribute to the improvement of the prediction method of ground motions for interplate earthquakes.
We constructed two types of source models: long-period (<0.2 Hz) and short-period (>0.2 Hz) source models. Several source models based on waveform inversions have already been proposed (e.g., JMA, 2017; Ruiz et al., 2017; Lange et al., 2018; USGS, 2018). The source model of JMA (2017) was based on a teleseismic waveform inversion and its numerical data is available. Our long-period model consisted of large slip area (LSA) and background area and was constructed by tuning JMA (2017). First, we extracted rupture area by applying the criterion of Somerville et al. (1999) to JMA (2017). We located a fault plane with the rupture area at the plate boundary. Subsequently, we adjusted source parameters such as location, area, and slip magnitude of large slip area via trial and error so that the synthetic waveforms can reproduce the observed pulses and amplitude level. For calculating the synthetic waveforms, we tuned the 1-D velocity structure model of Lange (2008) using the waveforms observed during an MW 5.4 aftershock. Our short-period model consisted of only square strong motion generation areas (SMGAs). The parameters of SMGAs were estimated by forward simulations using the empirical Green's function method (Irikura, 1986). The estimation was performed without referring to the long-period model. The records observed during the above MW 5.4 event were used as empirical Green's functions.
We identified one LSA (2500 km2) in the long-period model and three SMGAs (1368 km2) in the short-period model. The LSA is located northeast of the hypocenter and at a depth of 25-40 km. The three SMGAs are located inside the LSA. The SMGAs and LSA are spatially overlapped primarily due to relatively small MW, while the separation of SMGA (namely, high-frequency seismic radiation) and LSA is reported for M >8 events (e.g., Lay et al., 2012; Tilmann et al., 2016). There was also a partial-separation event (the 1994 MW 7.7 offshore Sanriku earthquake; Miyahara and Sasatani, 2004) in spite of roughly the same MW as the Chiloé earthquake. Thus, the discussion on overlap/separation for M 7.5-8.0 events requires further case studies.
The rupture area and LSA are larger than those derived from Murotani et al. (2008), a validated scaling relationship for interplate earthquakes in Japan. The combined area of SMGAs is also larger than the scaling relationship of Satoh (2010). This implies the difference in source characteristics between Japan and Chile (Guo et al., this meeting). The LSA and SMGAs occupy 21% and 12% of the rupture area, respectively. Our source model clearly indicates that SMGA should have smaller dimension than LSA to well reproduce the short-period component of ground motions. This discrepancy between LSA and SMGA has also been discussed (e.g., Satoh, 2010; Tajima et al., 2013) and is considered to be a source characteristic common to M >7 events.
The short-period spectral level is 8.3 × 1019 Nm/s2, which is comparable to that of the events in northeastern Japan with similar MW. The stress drop of the SMGAs (19 MPa) is smaller than that of northeastern-Japan events (30-40 MPa). In Chile, stress drops of SMGAs may have large variability: ~20 MPa for the 2010 MW 8.8 Maule earthquake (e.g., Frankel, 2017) and 40 MPa for the 2015 MW 8.3 Illapel earthquake (Tohdo et al., 2019).
Acknowledgements: This study was based on the 2021 research project “the study on the characterized source model for interplate earthquakes” by the Nuclear Regulation Authority, Japan.
We constructed two types of source models: long-period (<0.2 Hz) and short-period (>0.2 Hz) source models. Several source models based on waveform inversions have already been proposed (e.g., JMA, 2017; Ruiz et al., 2017; Lange et al., 2018; USGS, 2018). The source model of JMA (2017) was based on a teleseismic waveform inversion and its numerical data is available. Our long-period model consisted of large slip area (LSA) and background area and was constructed by tuning JMA (2017). First, we extracted rupture area by applying the criterion of Somerville et al. (1999) to JMA (2017). We located a fault plane with the rupture area at the plate boundary. Subsequently, we adjusted source parameters such as location, area, and slip magnitude of large slip area via trial and error so that the synthetic waveforms can reproduce the observed pulses and amplitude level. For calculating the synthetic waveforms, we tuned the 1-D velocity structure model of Lange (2008) using the waveforms observed during an MW 5.4 aftershock. Our short-period model consisted of only square strong motion generation areas (SMGAs). The parameters of SMGAs were estimated by forward simulations using the empirical Green's function method (Irikura, 1986). The estimation was performed without referring to the long-period model. The records observed during the above MW 5.4 event were used as empirical Green's functions.
We identified one LSA (2500 km2) in the long-period model and three SMGAs (1368 km2) in the short-period model. The LSA is located northeast of the hypocenter and at a depth of 25-40 km. The three SMGAs are located inside the LSA. The SMGAs and LSA are spatially overlapped primarily due to relatively small MW, while the separation of SMGA (namely, high-frequency seismic radiation) and LSA is reported for M >8 events (e.g., Lay et al., 2012; Tilmann et al., 2016). There was also a partial-separation event (the 1994 MW 7.7 offshore Sanriku earthquake; Miyahara and Sasatani, 2004) in spite of roughly the same MW as the Chiloé earthquake. Thus, the discussion on overlap/separation for M 7.5-8.0 events requires further case studies.
The rupture area and LSA are larger than those derived from Murotani et al. (2008), a validated scaling relationship for interplate earthquakes in Japan. The combined area of SMGAs is also larger than the scaling relationship of Satoh (2010). This implies the difference in source characteristics between Japan and Chile (Guo et al., this meeting). The LSA and SMGAs occupy 21% and 12% of the rupture area, respectively. Our source model clearly indicates that SMGA should have smaller dimension than LSA to well reproduce the short-period component of ground motions. This discrepancy between LSA and SMGA has also been discussed (e.g., Satoh, 2010; Tajima et al., 2013) and is considered to be a source characteristic common to M >7 events.
The short-period spectral level is 8.3 × 1019 Nm/s2, which is comparable to that of the events in northeastern Japan with similar MW. The stress drop of the SMGAs (19 MPa) is smaller than that of northeastern-Japan events (30-40 MPa). In Chile, stress drops of SMGAs may have large variability: ~20 MPa for the 2010 MW 8.8 Maule earthquake (e.g., Frankel, 2017) and 40 MPa for the 2015 MW 8.3 Illapel earthquake (Tohdo et al., 2019).
Acknowledgements: This study was based on the 2021 research project “the study on the characterized source model for interplate earthquakes” by the Nuclear Regulation Authority, Japan.