FEASIBILITY STUDY ON EARTHQUAKE EARLY WARNING SYSTEM FOR THE CITY OF LIMA, PERU, USING A STRONG-MOTION NETWORK

Calderon C.; Hayashida T. · World Conference on Earthquake Engineering proceedings (2021)

Resumen

Feasibility of earthquake early warning system (EWS) for the city of Lima, Peru, was investigated using strong motion records data from the newly installed real-time strong motion observation network of the Japan Peru Center of Seismic Investigation and Disaster Mitigation (CISMID). The operation of the network started in 1988. We selected earthquakes occurred in 2017, the operation of the network obtained 347 records from 24 events. The magnitudes of the events are bigger than or equal to ML 3.5 and the focal depths are smaller or equal to 100 km. There were 19 observation stations in Lima with 2-12 km station intervals. At each station strong-motion accelerograph REFTEK 130-SMA is installed at sample frequency of 200 Hz. The sizes and locations of the selected events were estimated using the initial P-wave portions (3 seconds) of the observed waveforms, based on the conventional approaches. For the estimation of the magnitude, we determined the “tauc” parameter that relates to earthquake magnitude. We found a reasonable correlation between the estimated magnitudes and the catalogue magnitudes for shallower and relatively larger earthquakes in the vicinity (distance < 130 km). The hypocenter locations were determined with the detected P-wave arrival times. Since the observation stations are distributed only in the limited area of Lima, we found that the hypocenters are not well determined and we need at least eight stations to stably estimate the locations. To investigate the accuracy of the existing ground-motion prediction equations (GMPEs) in the real-time scheme, we selected two GMPEs for subduction zone earthquakes to compare the predicted peak ground accelerations (PGA) with the observed ones. In most cases the GMPEs tend to overestimate PGAs, possibly due to large site amplification in Lima. It could be also another reason that these GMPEs were developed for other environments using only larger earthquakes. The blind zone is defined as the region around the epicenter where no warning is issued because the strong shaking has already arrived by the time the alert is generated (Kuyuk and Allen, 2013). We also confirmed that the size of blind zones of the earthquake early warning system is not so large for most earthquakes and it is possible to give an alert to the city of Lima before S-wave arrivals, if the hypocenter locations (i.e. distance from the hypocenter) were correctly determined. © The 17th World Conference on Earthquake Engineering.