First European Conference on Earthquake Engineering and Seismology (A joint event of the 13 th ECEE & 30 th General Assembly of the ESC) Geneva, Switzerland, 3-8 September 2006 Paper Number: 741 EARTHQUAKE EARLY WARNING SYSTEM AT A LOCAL GOVERNMENT AND A PRIVATE COMPANY IN JAPAN Hiroaki NEGISHI 1 and Shunroku YAMAMOTO 2 SUMMARY We have developed a real-time earthquake information system, which provides earthquake information just a few seconds after P-wave arrival at the nearest station from a hypocenter [Horiuchi et al., 2005]. This system determines hypocenter and magnitude immediately by using telemetry seismographs network operated by National Research Institute for Earth Science and Disaster Prevention (NIED). Thus we get early warning for big ground motion before it begins in an area a little far from a hypocenter. Now we are conducting the actual proof experiment in collaboration with a local government and a private company, in order to verify the validity and practicality about the broadcasting and application of this information. The earthquake alarm information is automatically sent from NIED to the Fujisawa City Office, Kanagawa, Japan, having large earthquake occurrences. Data processing PCs at the office estimate seismic intensity and S-wave arrival time and display them. The PC has a function to control an eight-channel electric relay automatically according to the information, so we can control patrol lights, some instruments, etc. We started the system in July 2002, and now 13 public halls; two municipal schools and a city hospital have this system. Tokio Marine and Nichido Risk Consulting Co., Ltd., a risk-consulting subsidiary of one of the major casualty insurance companies in Japan, introduced this system in June 2003. The earthquake alarm information is mainly used for the employee s safety. Data processing PCs transmit information on seismic intensity and S-wave arrival time to the monitor unit (a display, speakers and a four-color patrol light) and employee s cellular phones. We are also using it for the business of the company. The earthquake information triggers Payment insurance calculation system to estimates the payment insurance and to call members for the damage assessment automatically. 1. INTRODUCTION In recent years, seismologists and engineers have designed a number of real-time seismic systems for quickly providing earthquake information following an earthquake. These systems have been developed in countries where large population centers are vulnerable to large earthquake disasters. The idea behind Early Warning is that it takes a finite amount of time for seismic waves to travel. For locations somewhat far from the hypocenter, the damaging radiation is mostly in the S-wave, which travels at 3 to 4 km/sec. If we know very quickly that an earthquake has occurred, it is possible to send a warning before the S-wave arrival at a site. Various kinds of earthquake early warning systems have been developed in some countries, such as the Urgent Earthquake Detection and Alarm System (UrEDAS) in Japan [Nakamura, 1988, 1996], the Earthquake Alarm System in Taiwan [Teng et al., 1997; Wu et al., 1998, 2001]. 1 Earthquake Disaster Mitigation Research Center, National Research Institute for Earth Science and Disaster Prevention, 4th Floor, Human Renovation Museum, 1-5-2, Kaigan-dori, Wakinohama, Chuo, Kobe, 651-0073, JAPAN Email : negishi@bosai.go.jp 2 National Research Institute for Earth Science and Disaster Prevention, 3-1, Tenno-dai, Tsukuba, 305-0006, JAPAN Email: syama@bosai.go.jp 1
Figure 1: High-sensitivity Seismic Network Operated by NIED in Japan. One of the most effective and successful systems is the Mexico City alert system developed by Espinosa Aranda et al., [1995]. Mexico City is built on an old lakebed, which greatly amplifies seismic waves, and a subduction zone run along the western coast of Mexico frequently produces large earthquakes. They placed 12 seismic stations near the coast, and established a quick information transmission system to broadcast earthquake alert to the city when the seismograph catches a large quake signals. On September 14, 1995 there was a M7.3 earthquake on the edge of the Guerrero Gap that activated the early warning system. An alert was successfully broadcasted in Mexico City 72 seconds before the arrival of the strong shaking. In the cases that major earthquake location is clear and far from cities, this system is comparatively easily effective, like in Mexico. In Japan, however, any places have potentials of the large earthquake occurrence and there are many cities with various distances from the earthquake source areas. So we need to have a system that can pick accurate arrival time of seismic phases, and determine hypocenter and magnitude with a few station data, to broadcast earthquake early warning practically. We have developed such total system, which automatically determines hypocenter and magnitude very quickly, and provides an earthquake early warning just a few seconds (almost less than 5 to 6 seconds) after P-wave arrival at the nearest station to a hypocenter. The detailed method of the determination algorithms is expressed in Horiuchi et al. [2005]. Here we will present mainly on the actual proof experiment in collaboration with a local government and a private company, in order to verify the validity and practicality about the broadcasting and application of this information. 2
2. EARTHQUAKE EARLY WARNING SYSTEM Before the main subject, we will present on the automatic processing system for the hypocenter determination briefly. Horiuchi et al. [2005] have developed an earthquake early warning system that determines earthquake parameters (origin time, hypocenter and magnitude) within a few seconds after P-wave s arrival at the closest station, and then transmits the information before the S-wave arrival in areas of possible serious earthquake damage. For the purpose of quick determination of hypocenter and magnitude, we have developed a new method of hypocenter location using a dense seismic network. This method uses not only P wave arrival time but also not-yet-arrival time. Quantitative definition of not-yet-arrival time data makes it possible to determine accurate epicenter location, depth and origin time with only two arrival stations. On the other hand, a wide and dense seismic network with real-time telemetry is also necessary to use this method effectively. After the 1995 Kobe earthquake National Research Institute for Earth Science and Disaster Prevention (NIED) have established a highly sensitive and dense seismic network in Japan, named Hi-net. The network uses three-component velocity seismometers in boreholes deeper than 100 m (the deepest one has depth of 3000 m), and sends waveform data to the NIED data center via frame-relay telemetry. The telemetry delay is almost shorter than 2 seconds for 95% of the data packets. As the result, we have a dense and uniform seismic network with real-time telemetry for almost all of Japan (Figure 1). The establishment of this seismic network makes the earthquake early warning system possible in Japan. Figure 2 shows an example of Earthquake Early Warning information broadcasted by our system. This event occurred in the offshore of Miyagi Prefecture, Tohoku, Japan in 26 May 2003. The rupture of this earthquake began at the depth of about 70 km, and 10.8 seconds after that the nearest station to the epicenter caught the P- wave. The first earthquake early warning information was sent momentarily at the time when P-wave reached the secondarily nearest station. As the result, we succeeded in giving a warning 10 to 20 seconds before the main shock hits to some major cities, such as Shiogama (about 13 seconds) and Sendai (about 16 seconds). Figure 2: Plots of acceleration seismograms for the 2003 Off Miyagi Prefecture Earthquake. 3
3. APPLICATIONS TO LOCAL GOVERNMENT AND COMMERCIAL COMPANY We started real-time operation of this system in July 2002 as a proof experiment. The earthquake alarm information is automatically sent to the Disaster Prevention Center of Fujisawa City, Kanagawa Prefecture, Japan, and the Tokio Marine Nichido Risk-consulting Co. Ltd., Tokyo Japan [Negishi et al., 2004]. We use two units of personal computers operated by Linux. Each unit has at least two computers because of redundancy. One unit determines earthquake parameters by using the method mentioned in the previous chapter, Figure 3: The outline of the Earthquake Early Warning System in Fujisawa City, Kanagawa, Japan. Figure 4: Example of graphical display of earthquake information displayed on client computer. 4
and the other PCs send the results by XML packets to users. The latter PCs send the earthquake information (hypocenter location, magnitude and origin time) automatically to the personal computers that placed at the user s data center via exclusive network lines. The host computer at each center receives the information and estimates seismic intensity, arrival time of S-waves (regarded as main shock) using the information and some local parameters, such as site location and site amplification response due to soil condition. The site information of earthquake early warning is used in various ways, for not only dissemination to citizens but also control of some instruments and mechanical systems. 3.1 Case Study for Local Government Fujisawa City is located in the south part of Kanagawa Prefecture that lies in the center of the Japanese Archipelago, and faces the western Pacific Ocean. Fujisawa is a cultural city with a population of approximately 390,000 and is endowed with housing, industry, agriculture, commerce, tourism and education. Fujisawa city is one of the cities that are expected to have damages due to the Tokai earthquake and/or Kanto earthquake near future. The collaborative experiment of real-time earthquake information system between NIED and Fujisawa City has started on July 2002. The outline of the system is shown in Figure 3. The earthquake-monitoring unit placed at NIED is checking waveform data telemetered from Hi-net stations continuously. Just a few seconds after the detection of large earthquake, the server determines earthquake parameters immediately and sends the information to the data distribution unit. The unit makes XML packets and sends them to the data server placed at the Disaster Prevention Center of Fujisawa City via exclusive network line. This unit also has a function to monitor the health condition of the monitoring unit, network line and the distribution server itself. The information about the system trouble is sent to the administrators in charge immediately when a certain part of the system is down. Earthquake information is shown by the graphic display plainly laid out, as shown in Figure 4. Hypocenter location and S phase wavefront (redrawn every second) are displayed on geographical map, and estimated seismic intensity and leading time are displayed, too. Synthetic voice speaks seismic intensity and counts down leading time. The earthquake warning is passed on to the person in charge of disaster mitigation in the city office by these display and voice. As the result, they can promptly take a necessary action. The earthquake information display device has an eight-channel relay that triggered by earthquake information, and can control various equipments. In this proof experiment, we are trying to control the radio communication system of the Disaster Prevention Center of Fujisawa city by the switching relay, and to report earthquake information to staffs by the voice at the same time. Now the system is placed at the Disaster Prevention Center, the district disaster prevention bases (there are 13 stations), Fujisawa City Hospital, and some municipal schools. Earthquake information is sent to these places via the regional intranet. 3.2 Case Study for Commercial Company Tokio Marine and Nichido Risk Consulting Co., Ltd. is a risk-consulting subsidiary of one of the major casualty insurance companies in Japan. The proof experiment with the company was begun in June 2003. This system is designed for the purpose to secure the safety of employees. The outline of the system is shown in Figure 5. The flow of data processing is almost the same as the experiment in the Fujisawa city. The terminal also has the patrol light with two or more colors, and we can identify the seismic intensity according to the color of the lamp without seeing the screen of the computer. Additionally, information is immediately delivered to employee's cellular phones by short mail (Figure 6a). This system has an another important function. In risk consulting companies, it is one of the most important businesses to estimate the claims paid of earthquake insurance and to construct the system of the damage assessment immediately after the occurrence of large earthquake. We tried to develop a system that automatically carries out these works by earthquake early warning information (Figure 6b). The distribution of the seismic intensity and damage is automatically estimated by using the real-time hypocenter information and focal mechanism solution that delivered soon after the earthquake early warning and calculation of claims paid and damage assessment are conducted with them. Thus, earthquake early warning is effective to not only the reduction of human damage but also reduction of economical damage. 5
Figure 5: The outline of the Earthquake Early Warning System in Tokio Marine and Nichido Risk Consulting Co. Ltd. Figure 6: a. (Left) Example of Earthquake Early Warning Information by Short-Mail-Service of Cellular Phone. b. (Right) Analysis Window of Focal Mechanism Data for Payment Insurance Calculation System. 4. DISCUSSION AND CONCLUSION For practical use of Earthquake Early Warning, several problems have to be solved, such as 1) Large and dense seismic array, 2) Network infrastructure of real-time telemetry and high-spec computer(s), 3) Quick determination algorithms for earthquake parameters, 4) Effective applications for disaster mitigation using earthquake early warning, and 5) Society that can utilize real-time earthquake information effectively, e.g., education. 6
In addition, we should research on the correspondence when mis-information is sent, and the social economy information and the given influence. The above-mentioned matter appears in the pioneering early warning system in Mexico [Espinosa Aranda et al., 1995]. The system hopes to provide a 60 sec warning to Mexico City. The warnings are disseminated by radio and public speakers. There are speakers in public schools and 25 campuses. Early warning signals also go to various government, utility, and transportation agencies. An estimated 2,000,000 people will hear a warning. Early in the project, there were many false alarms that affected public confidence, but recently many of the technical problems causing the false warning have been solved. In case of the M7.3 earthquake in 1995, although there was only minor damage in Mexico City from the earthquake the system showed it could work well. All the participating radio stations broadcasted the pre-determined alert messages. Due to a good education effort in some of the large high rises that had speakers, there were orderly evacuated out of the building. This experience shows that with good education efforts, early warning systems in certain situations have the potential for preventing heavy casualties from earthquakes. These are the valuable data for us to operate earthquake early warning system effectively. Since 2003, the Ministry of Education, Culture, Sports, Science and Technology (MEXT) and NIED started a joint research project entitled Research Project for the Practical Use of Real-time Earthquake Information Networks. The research object of this project has wide fields, from upgrading earthquake early warning system, developing effective user applications, to researching its economical influences and psychological influences. Practical earthquake early warning system should not be designed only by seismologists. Incorporative knowledge of seismologists, engineers, computer system experts, and emergency management professionals, is necessary to design and develop a reliable and stable earthquake early warning system. 5. ACKNOWLEDGEMENTS We are grateful to Messrs Shinji Abe, Toshio Mogi, Takashi Satta and Noboru Sugiyama of Disaster Prevention Center of Fujisawa City for support of the proof experiment in their city. We also would like to express our appreciation to Messrs Masaru Matsumoto, Tomohisa Sashida and Yoshiaki Ogane of Tokio Marine and Nichido Risk Consulting Co. Ltd. for their support of the proof experiment in their company. Valuable advices from Drs. Shigeki Horiuchi and Shoji Sakata of National Research Institute for Earth Science and Disaster Prevention are also appreciated. 6. REFERENCES Epinosa Aranda, Jimenez, J.H., Ibarrola, G., Alcantar, F., Aguilar, A., Inosrroza, M., and Maldonado, S. (1995), Mexico Coty seismic alert system, Seismological Research Letters, 66, 42-53. Horiuchi, S., Negishi, H., Abe, K., Kamimura, A., and Fujinawa, Y. (2005), An Automatic Processing System for Broadcasting Earthquake Alarms, Bulletin of Seismological Society of America, 95, 708-718. Nakamura, Y. (1988), On the Urgent Earthquake Detection and Alarm System (UrEDAS), Proceedings Ninth World Conference, Earthquake Engineering VII, Japan Association for Earthquake Disaster Prevention, Tokyo, Kyoto, 673-678. Nakamura, Y. (1996), Real-time Information Systems for Hazards Mitigation, Proceedings Eleventh World Conference, Earthquake Engineering, CD-ROM, Paper No. 2134, Pergamon, Oxford. Negishi, H., Yamamoto, S., Sakata, S., Abe, S., Mogi, T., Satta, T., Sugiyama, N., Matsumoto, M., Sashida, T., and Ogane, S. (2004), Use and Application of the Real-time Earthquake Information Fujisawa City and The Tokio Marine Risk Consulting CO., LTD. -, 2004 Japan Earth and Planetary Science Joint Meeting, S046-P027. Teng, T.L., Wu, Y.M., Shin, T.C., Tsai, Y.B., and Lee W.H.K. (1997), One Minute After: Strong Motion Map, Effective Epicenter, and Effective Magnitude, Bulleting of Seismological Society of America, 87, 1209-1219. Wu, Y.M., Shin, T.L., and Tsai, Y.B. (1998), Quick and Reliable Determination of Magnitude for Seismic Early Warning, Bulleting of Seismological Society of America, 88, 1254-1259. Wu, Y.M., Shin, T.L., and Chang, H. (2001), Near Real-time Mapping of Peak Ground Acceleration and Peak Ground Velocity Following A Strong Earthquake, Bulleting of Seismological Society of America, 91, 1218-1228. 7