Earthquake Early Warning System Using A Dense Seismic Network
Abstract
We have developed an earthquake early warning system (EEWS) by using a spatially dense and high dynamic range seismic network that consists of 700 stations covering the Japanese Islands. The system is able to determine the hypocenter location, magnitude and a shaking intensity parameter within a few seconds from the P-wave arrivals at the closest stations and then transmits this information before the arrival of S-waves in areas of potentially serious earthquake damage. Since the available waveform data increases with time, our EEWS is designed to update the earthquake parameters every second. The methods and results of our EEWS are as follows: (1) An early warning system should be able to reliably determine earthquake parameters as quickly as possible; it is therefore unreasonable to wait until waveform data from numerous stations have been collected for analysis. Assuming that all stations will observe P-wave arrivals from a large earthquake, we have developed a novel method of determining the hypocentral location that uses the arrival times for only a few stations and also the lack of P-wave arrivals at other stations at a given time (Tnow). The use of Tnow makes it possible not only to reliably determine hypocenter parameters within a few seconds but also to detect spurious arrival times and remove them automatically. By using Tnow in our dense seismic network, our EEWS determines hypocenter parameters for 10 to 20 events per day and 99% of them are determined correctly. (2) We introduced a new parameter, in addition to the moment magnitude, to estimate the shaking intensity more reliably. Whereas the moment magnitude is estimated from the amplitude of the displacement spectrum, this new parameter is dependent on values of observed P-wave shaking intensity. We call it _gshaking intensity magnitudEh. We compared estimation errors of shaking intensity that were calculated from both magnitudes and found that the new parameter decreases estimation errors of shaking intensity by 50%. We also found that for events larger than M7, the shaking intensity magnitude can estimate the shaking intensity more quickly than the ordinary moment magnitude. These results therefore indicate that the shaking intensity magnitude can predict a more reliable shaking intensity at an earlier time. (3) Our EEWS was tested on waveform data recorded by southern California seismic networks. We copied 100 waveform data for larger earthquakes that occurred in the last two years and simulated a real-time analysis using our EEWS. This test showed: (a) our system became functional by only changing the format of the waveform data and the station identification codes, (b) the epicenter location errors in the first solution for 98% and 95% of the events are within 30 km and 20 km, respectively.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2005
- Bibcode:
- 2005AGUFM.S51D1028K
- Keywords:
-
- 0520 Data analysis: algorithms and implementation;
- 7219 Seismic monitoring and test-ban treaty verification;
- 7223 Earthquake interaction;
- forecasting;
- and prediction (1217;
- 1242);
- 7290 Computational seismology;
- 7294 Seismic instruments and networks (0935;
- 3025)