Fast Identification of Near-Trench Earthquakes Along the Mexican Subduction Zone Based on Characteristics of Ground Motion in Mexico City
Abstract:The disastrous 1985 Michoacan earthquake gave rise to a seismic alert system for Mexico City which became operational in 1991. Initially limited to earthquakes along the Guerrero coast, the system now has a much wider coverage. Also, the 2004 Sumatra earthquake exposed the need for a tsunami early warning along the Mexican subduction zone. A fast identification of near-trench earthquakes along this zone may be useful in issuing a reliable early tsunami alert. The confusion caused by low PGA for the magnitude of an earthquake, leading to “missed” seismic alert, would be averted if its near-trench origin can be quickly established. It may also help reveal the spatial extent and degree of seismic coupling on the near-trench portion of the plate interface. This would lead to a better understanding of tsunami potential and seismic hazard along the Mexican subduction zone.
We explore three methods for quick detection of near-trench earthquakes, testing them on recordings of 65 earthquakes at station CU in Mexico City (4.8 ≤Mw≤8.0; 270≤R≤615 km). The first method is based on the ratio of total to high-frequency energy, ER (Shapiro et al., 1998). The second method is based on parameter Sa*(6) which is the pseudo-acceleration response spectrum with 5% damping, Sa, at 6 s normalized by the PGA. The third parameter is the PGA residual, RESN, at CU, with respect to a newly-derived ground motion prediction equation at CU for coastal shallow-dipping thrust earthquakes following a bayesian approach. Since the near-trench earthquakes are relatively deficient in high-frequency radiation, we expect ER and Sa*(6) to be relatively large and RESN to be negative for such events. Tests on CU recordings show that if ER ≥ 100 and/or Sa*(6) ≥ 0.70, then the earthquake is near trench; for these events RESN ≤ 0. Such an event has greater tsunami potential. Few misidentifications and missed events are most probably a consequence of poor location, although unusual depth and source characteristics may also be responsible in some cases. We propose routine computation of these parameters (along with location and magnitude) by the National Seismological Service of Mexico and dissemination of the information to other interested agencies which are in charge of tsunami alert, seismic alert, and near real time ground motion intensity maps for Mexico City.