S13C-1433
Status of the SAFER Project (Seismic eArly warning For EuRope)
Since July 2006, the SAFER (Seismic eArly warning For EuRope) Project, a European Commission funded Specific Targeted Research or Innovation Project, has coordinated efforts to develop methodologies for the mitigation of the effects of earthquakes in and around Europe. SAFER aims to develop new methodologies, and expand upon existing ones for a range of actions that will respond to the consequences of damaging earthquakes. The time span over which such actions will take place extends from the earliest detection of the first seismic waves to tens of minutes after the main earthquake has finished, and to days and weeks afterwards so as to consider the subsequent aftershocks. The sort of actions envisaged to be developed by SAFER include the detection and analysis of the first P-wave arrivals, decision- making processes (i.e. is the earthquake potentially dangerous or not), the issuing (or not) of alarms, instigating safety measures for buildings, industrial processes and lifelines, and predictions of possible secondary effects such as landslides and subsequent aftershocks. It is the intention of this poster to present an overview of SAFER, some of the preliminary results and the expected products. This will include detection and analysis algorithms, new ways of optimising seismic networks, utilization of the USGS ShakeMap program, and predictive tools of aftershock hazard.
S13C-1434
The Irpinia Seismic Network: An Advanced Monitoring Infrastructure For Earthquake Early Warning in The Campania Region (Southern Italy)
A new seismic network (ISNet, Irpinia Seismic Network) is now operating in the Southern Italy. It is conceived as the core infrastructure for an Earthquake Early Warning System (EEWS) under development in Southern Italy. It is primarily aimed at providing an alert for moderate to large earthquakes (M>4) to selected target sites in Campania Region and it also provides data for rapid computation of regional ground-shaking maps. ISNet is deployed over an area of about 100×70 km2 covering the Apenninic active seismic zone where most of large earthquakes occurred during the last centuries, including the Ms=6.9, 1980 Irpinia earthquake. ISNet is composed of 29 seismic stations equipped with three components accelerometers and velocimeters aggregated in six smaller sub-nets. The sub-net stations are connected with a real-time communications to a central data- collector site (LCC, Local Control Center). The different LCCs are linked among them and to a Network Control Center (NCC) located in the city of Naples 100 km away from the network center, with different type of transmission systems chosen according their transmission mode robustness and reliability. The network is designed to provide estimates of the location and size of a potential destructive earthquake within few seconds from the earthquake detection, through an evolutionary and fully probabilistic approach. For the real time location we developed a methodology which extends and generalizes the one Horiuchi et al. (2005) by a) starting the location procedure after only one station has triggered, b) using the Equal Differential Time (EDT) approach to incorporate the triggered arrivals and the not-yet-triggered stations, c) estimating the hypocenter probabilistically as a pdf instead of as a point, and d) applying a full, non-linearized, global-search for each update of the location estimate. Following an evolutionary approach, the method evaluates, at each time step, the EDT equations considering not only each pair of triggered stations, but also those pairs where only one station has triggered. The size of earthquake is also evaluated by a real time, evolutionary algorithm based on a magnitude predictive model and a Bayesian formulation. It is aimed at evaluating the conditional probability density function of magnitude as a function of ground motion quantities measured on the early part of the acquired signals. The predictive models are empirical relationships which correlate the final event magnitude with the P-displacement amplitudes measured on first 2-4 seconds of record after the first-P arrival. The methods previously described for rapidly estimating the event's location and magnitude, are used to perform a real-time seismic hazard analysis allowing to compute the probabilistic distribution, or hazard curve, of ground motion intensity measures (IM) i.e. the peak ground acceleration (PGA) or the spectral acceleration (Sa), at selected sites of the Campania Region. We show the performances of the earthquake early warning system through applications to simulated large events and recorded low magnitude earthquakes.
S13C-1435
A Virtual Simulator As A Tool For Testing The Performance Of The ISNet Early Warning Infrastructure.
The ISNet (Irpinia Seismic Network) consists of 29 six-components seismic stations deployed in a 100×70 km2 area in Southern Italy, containing the fault system that generated the 1980, M=6.9, Irpinia earthquake. The seismic stations, equipped with both accelerometers and seismometers, are able to follow both the strong and the weak motion. The computation is distributed along the network, with local control centers collecting data-streams from the single stations and performing a redundant computation of the early warning parameters (picking, location, magnitude and hazard integral). The decision is finally sent to the network control center, which is located far-away from the network itself. The performance of the infrastructure therefore depends on the latency of the information in the acquisition and in the transmission, on the computation time and on the failures of the system. We build up a virtual simulator to analyze off-line both the lead-time of the whole infrastructure and the influence of the failures of the system on the final decision. We use the simulator as a platform to test and optimize the algorithms for the computation of the early warning parameters, to investigate the optimum topology for the existing network, to asses weaknesses of the present infrastructure and define interventions plans to make the alarm decision robust. The simulator is a flow chart, in which the single blocks (acquisition, transmission, computation and decision) are either mathematical operators based on empirically estimated physical quantities or code lines running on the same platform. We evaluate the performances of the network on a synthetic database of records simulating the 1980, Irpinia earthquake and a set of real data from an earthquake of magnitude 3.0.
S13C-1436
Vrancea Earthquake Early Warning System Development in Romania
A prototype early warning system was developed in Romania in order to provide 25-35 seconds warning time for Bucharest facilities for earthquake with M >6.5. The prototype system consists of four components: network ground motion sensors in the epicentral area, an area centre in epicentre, a digital radio communication link and a central data centre. These components are fully functional at Romania seismic data centre (RO_NDC) in Bucharest since August 2003. The system is in testing phase for one important user from Bucharest the Van de Graff Accelerator were we provide about 25 seconds warning time before shaking occurs. Because the prototype system demonstrated that potentially useful warnings of strong shaking from Vrancea earthquakes are feasible, the National Institute for Earth Physics decide to implement this system for other four important facilities from Romania. At every user was installed a dedicate device which is able to stop trigger the specific safety rules for each facility. The system was tested only with offline earthquakes due to the fact that since August 2003 there was no strong earthquake such that the real time detection algorithm to issue an alarm. The detection algorithm is based on a combination of spectral and amplitude analysis for the first 4 seconds of the P wave.
S13C-1437
PreSEIS: Earthquake Early Warning for Finite Faults
Earthquake early warning (EEW) systems have to meet two requirements: they have to be both fast and highly reliable. We have developed a method for EEW, called PreSEIS, which is as quick as methods that are based on single station observations and, at the same time, shows a higher robustness than most other approaches. At regular time steps after the triggering of the first EEW sensor, PreSEIS estimates the most likely source parameters of an earthquake using the available information on ground shaking at different sensors in a seismic network. The approach is based on two-layer feed-forward neural networks that allow estimating the location of the earthquake hypocenter, its moment magnitude, and the expansion of the evolving seismic rupture. We apply PreSEIS to the Istanbul Earthquake Rapid Response and Early Warning System (IERREWS). The moment magnitudes of 280 simulated finite faults scenarios (4.5 <= M <= 7.5) are estimated with errors of less than +-0.8 units after 0.5 s, +-0.5 units after 7.5 s, and +-0.3 units after 15.0 s. The mean location errors can be reduced in the same time intervals from 10 km over 6 km to less than 5 km, respectively. Our analysis demonstrates that the uncertainties of the estimated parameters (and thus of the warnings) decrease with time. This reveals a trade-off between the reliability of the warning on the one hand, and the remaining warning time on the other hand. The regular up-date of predictions with time allows PreSEIS to handle complex ruptures, in which the largest fault slips do not occur close to the point of rupture initiation. The estimated expansions of the seismic ruptures lead to a clear enhancement of alert maps, which visualize the level and distribution of likely ground shaking in the affected region seconds before seismic waves will arrive.
S13C-1438
Contribution of seismic arrays to earthquake early warning in Greece: Preliminary results for the Gulf of Corinth test site
An experimental small-aperture array of 4 elements (TRISAR) was installed in July 2003 in central Peloponnese, in order to investigate the contribution of arrays to earthquake location in Greece. Data analysis was based on the software developed and provided by NORSAR in order to simulate real-time processing followed by analyst review. The resulting earthquake locations were compared with those of conventional seismographic networks in order to quantify location errors, evaluate array performance and investigate enhancement possibilities. Analysis revealed that array performance in terms of event location is restricted by its very small aperture and limited number of sensors. Furthermore, detailed investigation of errors in automatic location results suggests structural and local geology effects. The possibility to automatically correct for systematic deviations was verified. However, future research with an extended array configuration was deemed necessary in order to provide clearer results by reducing phase misidentifications and wrong groupings made by the automatic algorithm, resulting from the poor slowness resolution of the array in its initial configuration. The opportunity to continue research on seismic array implementation for near-real time earthquake location as part of an early warning system presented itself within the framework of the SAFER project funded by EU. Following the detailed assessment of the initial results, the number of elements of the TRISAR array was increased to 7 and the geometry was modified in order to maximize resolution and performance focusing in the area of the Gulf of Corinth. This area, where optimum array performance was previously observed, has already been selected as one of the project test sites. Data processing will once again be carried out in real-time simulation mode while earthquake location results will be compared with those from a dense local network installed around the test-area. In addition, a second 4 element small-aperture array is installed in the Atalanti area (ATASAR), at a similar distance from the test site as TRISAR, in order to improve the accuracy of array-based earthquake location. Data telemetry will provide the basis for the implementation and testing of real-time processing during the operation of this second array. The preliminary results obtained during the first months of operation of the 2 arrays will be presented and discussed.
S13C-1439
The Early Warning System(EWS) as First Stage to Generate and Develop Shake Map for Bucharest to Deep Vrancea Earthquakes
EWS made by NIEP is the first European system for real-time early detection and warning of the seismic waves in case of strong deep earthquakes. EWS uses the time interval (28-32 seconds) between the moment when earthquake is detected by the borehole and surface local accelerometers network installed in the epicenter area (Vrancea) and the arrival time of the seismic waves in the protected area, to deliver timely integrated information in order to enable actions to be taken before a main destructive shaking takes place. Early warning system is viewed as part of an real-time information system that provide rapid information, about an earthquake impeding hazard, to the public and disaster relief organizations before (early warning) and after a strong earthquake (shake map).This product is fitting in with other new product on way of National Institute for Earth Physics, that is, the shake map which is a representation of ground shaking produced by an event and it will be generated automatically following large Vrancea earthquakes. Bucharest City is located in the central part of the Moesian platform (age: Precambrian and Paleozoic) in the Romanian Plain, at about 140 km far from Vrancea area. Above a Cretaceous and a Miocene deposit (with the bottom at roundly 1,400 m of depth), a Pliocene shallow water deposit (~ 700m thick) was settled. The surface geology consists mainly of Quaternary alluvial deposits. Later loess covered these deposits and the two rivers crossing the city (Dambovita and Colentina) carved the present landscape. During the last century Bucharest suffered heavy damage and casualties due to 1940 (Mw = 7.7) and 1977 (Mw = 7.4) Vrancea earthquakes. For example, 32 high tall buildings collapsed and more then 1500 people died during the 1977 event. The innovation with comparable or related systems worldwide is that NIEP will use the EWS to generate a virtual shake map for Bucharest (140 km away of epicentre) immediately after the magnitude is estimated (in 3-4 seconds after the detection in epicentre) and later make corrections by using real time dataflow from each K2 accelerometers installed in Bucharest area, inclusively nonlinear effects. Thus, developing of a near real-time shake map for Bucharest urban area is of highest interest, providing valuable information to the civil defense, decision makers and general public on the area where the ground motion is most severe. EWS made by NIEP can be considered the first stage to generate and develop the shake map for Bucharest to deep Vrancea earthquakes. http://www.infp.ro
S13C-1440
The Puerto Rico Seismic Network Broadcast System: A user friendly GUI to broadcast earthquake messages, to generate shakemaps and to update catalogues
The Puerto Rico Seismic Network (PRSN) has historically provided locations and magnitudes for earthquakes in the Puerto Rico and Virgin Islands (PRVI) region. PRSN is the reporting authority for the region bounded by latitudes 17.0N to 20.0N, and longitudes 63.5W to 69.0W. The main objective of the PRSN is to record, process, analyze, provide information and research local, regional and teleseismic earthquakes, providing high quality data and information to be able to respond to the needs of the emergency management, academic and research communities, and the general public. The PRSN runs Earthworm software (Johnson et al, 1995) to acquire and write waveforms to disk for permanent archival. Automatic locations and alerts are generated for events in Puerto Rico, the Intra America Seas, and the Atlantic by the EarlyBird system (Whitmore and Sokolowski, 2002), which monitors PRSN stations as well as some 40 additional stations run by networks operating in North, Central and South America and other sites in the Caribbean. PRDANIS (Puerto Rico Data Analysis and Information System) software, developed by PRSN, supports manual locations and analyst review of automatic locations of events within the PRSN area of responsibility (AOR), using all the broadband, strong-motion and short-period waveforms Rapidly available information regarding the geographic distribution of ground shaking in relation to the population and infrastructure at risk can assist emergency response communities in efficient and optimized allocation of resources following a large earthquake. The ShakeMap system developed by the USGS provides near real-time maps of instrumental ground motions and shaking intensity and has proven effective in rapid assessment of the extent of shaking and potential damage after significant earthquakes (Wald, 2004). In Northern and Southern California, the Pacific Northwest, and the states of Utah and Nevada, ShakeMaps are used for emergency planning and response, loss estimation, and communication of earthquake information to the public. We develop a tool to help the PRSN personnel on duty with the generation of ShakeMaps for the felt events in Puerto Rico and the Virgin Islands. Automatic or reviewed locations came from different sources and the user can select the method to broadcast the message using several ways like direct email trough service lists, a server/client tool to push messages to a remote display client, generate shakemap web pages and update the catalogues.
S13C-1441
ElarmS and the next large earthquake in Italy
We evaluate the capability of ElarmS to work as an Early Warning System for the next destructive earthquake in Italy. Potential sources were retrieved from the Database of Individual Seismogenic Sources (DISS; http://www.ingv.it/DISS/) and from the catalogue of historical earthquakes (CPTI04; http://emidius.mi.ingv.it/CPTI/). Based on the geometry of the new Italian Broad-Band Seismic Network, we estimate the "Alert time" for each source from different nucleation points and the size of the shadow zone, the area already affected by S-waves. We present our results for the whole country with a particular attention to the metropolitan area of Naples (Southern Italy) that have been selected as a test site for the SAFER project (http://www.saferproject.net/).
S13C-1442
Calibration of ElarmS using earthquakes in southern California
We derive empirical magnitude scaling relationships for southern California using a dataset of 59 past earthquakes recorded in southern California by the Southern California Seismic Network (SCSN) between 1992 and 2003. The events range in magnitude from 3.0 to 7.3. We use the maximum predominant period (tau-p-max) and the peak displacement amplitude (Pd) measured from the first 4 seconds of P-wave arrivals to determine period-magnitude and amplitude-magnitude scaling relationships respectively. Our calibration study shows that the scaling relationships are similar to those derived for northern California. The average error in magnitude estimates is 0.2 magnitude units for events with magnitudes smaller than 4.5, 0.3 magnitude units for events with magnitudes ranging from 4.5 to 6.5, and 0.5 magnitude units for events with magnitudes greater than 6.5. All 59 events are processed offline by the ElarmS algorithms to generate the realtime AlertMap predictions of peak ground shaking. When the station density is high (as in the greater Los Angeles metropolitan region) the performance of ElarmS is good. In regions with lower station density the system becomes slower and the errors become larger. The AlertMap animations for these events will be available at the meeting. http://www.ElarmS.org
S13C-1443
Application of ElarmS to earthquakes in Japan
Earthquake Alarms Systems, or ElarmS, analyzes seismic waves in realtime for the purpose of providing rapid earthquake information and warning of damaging ground shaking. P-wave arrival times are used to detect and locate earthquakes, peak amplitudes and maximum predominant period of the initial P wave are used to estimate the event magnitude, and the estimated magnitude, combined with available observations of peak ground shaking, is used to predict damaging ground motions for the surrounding region. ElarmS has been calibrated using seismic data from northern and southern California. Here we both test and calibrate the ElarmS algorithms with seismic data from Japan's broadband networks. This provides a test of ElarmS in another geographical location and geological setting and also provides calibration of the algorithms for a substantial number of large magnitude (M>6) earthquakes.
S13C-1444
A Real-Time CISN Test Bed for the ElarmS Early Warning Algorithm
The ElarmS earthquake early warning (EEW) algorithm, initially designed and tested on data collected from earthquake in southern California, is being tested on real-time network data from the California Integrated Seismic Network. In order to take advantage of the real-time broadband and strong motion data available in California, we have decomposed the ElarmS algorithm into several fundamental components: a waveform processing and phase detection module, and the ElarmS event processing module (EVM). Using a framework generator developed by Caltech that automatically generates stub code for waveform processing, we implemented the E waveform processing algorithms for filtering, peak ground motions, tau-p, and phase detection in a separate real-time module that runs at the distributed CISN waveform acquisition and processing centers at UC Berkeley, USGS Menlo Park, and Caltech. We developed a low latency real-time distribution system to distribute and aggregate the processed waveform attributes to one or more ElarmS processing centers. A newly developed API provides the EVM component of the ElarmS algorithm with the real-time waveform attributes and phase detections. In order to monitor, analyse and refine the ElarmS algorithm, we implemented a real-time monitoring module that monitors the state of EVM, which archives all processed waveform data in the order they were received whenever EVM detects an event or aborts. The waveform datasets will allow us to replay the exact waveform sequence seen in real-time in order to identify coding or algorithmic problems, and to evaluate refinements in the ElarmS EVM algorithm. The monitoring system also controls the event post-processing software used to generate the ElarmS Alertmaps and event web pages, and to submit data to the CISN EEW algorithm evaluation database developed by SCEC.
S13C-1445
Earthquake Early Warning: Real-time Testing of an On-site Method Using Waveform Data from the Southern California Seismic Network
We have implemented an on-site early warning algorithm using the infrastructure of the Caltech/USGS Southern California Seismic Network (SCSN). We are evaluating the real-time performance of the software system and the algorithm for rapid assessment of earthquakes. In addition, we are interested in understanding what parts of the SCSN need to be improved to make early warning practical. Our EEW processing system is composed of many independent programs that process waveforms in real-time. The codes were generated by using a software framework. The Pd (maximum displacement amplitude of P wave during the first 3sec) and Tau-c (a period parameter during the first 3 sec) values determined during the EEW processing are being forwarded to the California Integrated Seismic Network (CISN) web page for independent evaluation of the results. The on-site algorithm measures the amplitude of the P-wave (Pd) and the frequency content of the P-wave during the first three seconds (Tau-c). The Pd and the Tau-c values make it possible to discriminate between a variety of events such as large distant events, nearby small events, and potentially damaging nearby events. The Pd can be used to infer the expected maximum ground shaking. The method relies on data from a single station although it will become more reliable if readings from several stations are associated. To eliminate false triggers from stations with high background noise level, we have created per station Pd threshold configuration for the Pd/Tau-c algorithm. To determine appropriate values for the Pd threshold we calculate Pd thresholds for stations based on the information from the EEW logs. We have operated our EEW test system for about a year and recorded numerous earthquakes in the magnitude range from M3 to M5. Two recent examples are a M4.5 earthquake near Chatsworth and a M4.7 earthquake near Elsinore. In both cases, the Pd and Tau-c parameters were determined successfully within 10 to 20 sec of the arrival of the P-wave at the station. The Tau-c values predicted the magnitude within 0.1 and the predicted average peak-ground-motion was 0.7 cm/s and 0.6 cm/s. The delays in the system are caused mostly by the packetizing delay because our software system is based on processing miniseed packets. Most recently we have begun reducing the data latency using new qmaserv2 software for the Q330 Quanterra datalogger. We implemented qmaserv2 based multicast receiver software to receive the native 1 sec packets from the dataloggers. The receiver reads multicast packets from the network and writes them into shared memory area. This new software will fully take advantage of the capabilities of the Q330 datalogger and significantly reduce data latency for EEW system. We have also implemented a new EEW sub-system that compliments the currently running EEW system by associating Pd and Tau-c values from multiple stations. So far, we have implemented a new trigger generation algorithm for real-time processing for the sub-system, and are able to routinely locate events and determine magnitudes using the Pd and Tau-c values.
S13C-1446
Development of the CISN Earthquake Early Warning Web Site: Establishing a Basis for Comparison of Algorithms
As a part of the California Integrated Seismic Network (CISN) earthquake early warning (EEW) algorithm development, we have developed the CISN EEW web site to collect the results of multiple EEW algorithms and to display the algorithm results in a comparative manner. The CISN web site provides a time-based query capability which allows users to specify a time-window of interest. The CISN EEW web site can then produce a time-based display which shows all available algorithm results for the specified time window. These time-based plots provide a means of comparing algorithm performance and results. The CISN EEW web site developers have designed an XML-based data-exchange format which represents a common data format in which all the CISN EEW algorithm developers can format their results, and a data transfer mechanism for moving the data from the real-time network centers to non-real-time web site hosts. Incoming EEW data sets are automatically discovered by the web site processing system and entered into a data-management database at which time the data are available for discovery and display through the web site's query tools. The CISN web site is one part of a larger program of algorithm evaluation currently underway by CISN EEW algorithm development groups. The CISN EEW program of algorithm evaluation includes definition of common EEW algorithm data exchange formats, the establishment of a collection point for EEW algorithm results, and the development of web- based tools for comparing these results. Further techniques for comparing and evaluating the performance of EEW algorithms are under development by the group and will be implemented on the web site when available. http://www.scec.org/eew
S13C-1447
The Development of Real-time Strong-motion Observation on the Earthquake Early Warning in Taiwan
For the sake of seismic hazards mitigation, a real-time strong-motion monitoring system was implemented by the Central Weather Bureau (CWB) since 1995. After successive refinements during the past decade, the system has been utilized as the basis for the development of the early warning (EWS) application in Taiwan. In order to shorten the earthquake response time, a virtual sub-network approach is utilized at first. Under the practical experiment since 2001, for monitoring inland or near offshore earthquakes with magnitude greater than 4.5, the response time can be shorten as 18.8+-3.8 sec averagely. Therefore, it can provide early warning before S-wave arrival for metropolitan areas located 60 km away from the epicenter. For the sake of further reducing the area of so-called blind-zone which cannot provide early warning, we also attempted to utilize vertical displacement records of P-wave as a basis to issue warnings. As results, we found that the amplitude 0.1 cm can be used as a criterion to judge if an earthquake above magnitude 6.0 is occurring in Taiwan Island. Furthermore, we also derived a set of frequency-based formulas simultaneously, which can be used to estimate earthquake magnitude rapidly. To combine the research results, and under the framework of real-time strong-motion observational network, we designed an earthquake early warning algorithm which is suitable for Taiwan Island. According to the experiment on 7 magnitude-above-6.0 earthquakes took place after 2002, the averaged response times is within 15 sec. Furthermore, the processing time for inland earthquakes can be shorten beneath 10 sec, and the radius of blind-zone is reduced to 30 km. Since 2003, the CWB has been replacing the existing 16-bit digital accelero- graphs used in its real-time strong-motion monitoring to 24-bit instruments at an annual replacement rate of 5 to 10% of the total. Besides, some high quality borehole seismic stations and a cable-based Ocean Bottom Seismographic (OBS) system are planed to implement on and off north-eastern Taiwan since 2007. We will further develop the EWS in response to these new instruments, as well as to any new research ideas.
S13C-1448
Disaster Mitigation by Quick Response Against Strong Earthquake Motion
The concept of EEW, earthquake early warning, was reported on San Francisco Daily Evening Bulletin on 3rd November 1868 by Dr. Cooper first for San Francisco area. According to him this idea was triggered by a failure of earthquake prediction. It is impressive that the thinking way is same as present. In 1982 more than 100 years later, the warning system for Tohoku Shinkansen realized the Cooperfs idea for the first time on the world. After that, SAS for Mexico City started operating in 1991, and UrEDAS for Tokaido Shinkansen, an evolutional P-wave detection/warning system, began to operate in 1992. The UrEDAS technology is based on new concepts and methods to realize a real time system for estimating the earthquake parameters as magnitude, location and depth. In Japan at 1992, a new information service using UrEDAS technology had been prepared, but it was not born due to objection of JMA, Japan Meteorological Agency. By the same JMA, an information service, so called "Kinkyu Jishin Sokuho", will be broadcasted in nation wide from the first of October 2007. This implies that our UrEDAS Information Service plan has been correct, and it is my pleasure. However, it shall be rare case in Japan that JMA's information will reach faster than arriving of M7 class or less earthquake at the possible damaged areas, because it takes a time relatively long for processing and transmitting. Only for M8 class earthquakes of which the occurrence probability is about once in ten years in Japan, it is possible to receive the information before arriving of strong shaking in a possible damaged area far from the epicenter. JMA should popularize "Kinkyu Jishin Sokuho" after understandable explanation of these facts. I'm afraid that it will mislead people to broadcast a film on an evacuation training in unbelievable condition which is assuming an earthquake of seismic intensity 7 (corresponding to MMI scale XII) arriving 20 seconds later. Anyway, we can rely only on the onsite alarm in focal area. Accurate and quick earthquake information just after shaking is more useful than the late early warning. And I hope public organizations to make best effort that they can announce the earthquake information not only for main shock but also aftershocks, because these information is quite important for the quick response.
S13C-1449
Onsite Portable Alarm System - Its Merit and Application
Recently an existence of the earthquake early warning system (EEWS) becomes popular. In general, the EEWS will be installed in a fixed observation site and it may consist of several separated components such as a sensing portion, A/D converter, an information processing potion and so on. The processed information for warning may be transmitted to network via fixed communication line, and therefore this kind of alarm system is called as Network Alarm System. On the other hand, after the severe earthquake damage, it is very important to save the disaster victims immediately. These rescue staffs are also under the risk of aftershocks and need a local alarm not depending on the network, so this kind of alarm can be called as Onsite Alarm. But the common early warning system is too complex to set onsite temporary, and even if possible to install, the alarm is too late to receive at the epicentral area. However, the new generation earthquake early warning system FREQL can issue the P wave alarm by minimum 0.2 seconds after P wave detection. And FREQL is characterized as the unique all-in-one seismometer with power unit. At the time of the 2004 Niigata-Ken-Chuetsu earthquake, a land slide attacked a car just passing. A hyper rescue team of Tokyo Fire Department pulled the survivor, one baby, from the land slide area. During their activity the rescue team was exposed to the risk of secondary hazards caused by the aftershocks. It was clear that it is necessary to use a portable warning system to issue the onsite P wave alarm. Because FREQL was originally developed as portable equipment, Tokyo Fire Department asked us to modify it to the portable equipment with the loud sound and the light signal. In this moment, this portable FREQL has equipped in nation wide. When the hyper rescue team of Tokyo Fire Department was sent to Pakistan as a task force for rescue work of the 2005 Pakistan earthquake, the portable FREQL was used as important onsite portable warning system and P wave alarms was actually issued by three times during the rescue work. Although this is one example for the actual application of portable onsite alarm, it is possible to apply the other field as the construction field. In this presentation, Portable Onsite Alarm is discussed from views of its necessity and application.
S13C-1450
3D Models of Major Cities for Estimating Losses due to Earthquakes
Estimating losses due to earthquakes in real-time is essential for launching appropriate rescue operations. For loss estimates, information on the number of buildings, their type and their quality of construction is required. Because this information is often not available, or it is poor or incomplete for large cities in earthquake-prone areas, we are developing a method to derive as much information as possible about the building stock from 3D models of cities. The steps in our procedure to obtain an approximate model for a large city in a developing country are the following. (1) Orthorectify a high-resolution satellite image of the city. (2) Construct a 3D model, including all visible buildings. (3) Count the number of buildings in each of 5 height classes (one story, two stories, 3-5 stories, 6-10 stories, taller). (4) Identify each building's use approximately (residential, office, industrial, monument). (5) Use local information from engineers to construct fragility curves for the existing building types. (6) Assign the appropriate percentage of each height class to a fragility class. This information is developed for each administrative district of the city for which population numbers are available. Then each district, with its building composition, is subjected in a computer simulation to the strong motion corresponding to the earthquake magnitude and hypocentral distance in question. If information on wave amplification due to soil conditions is available an amplification factor is applied for each district separately. If necessary, districts are subdivided according to soil conditions. We present here the example of Bucharest, where 6 administrative districts exist. The losses we calculate are the distribution of buildings into 5 damage classes, as well as the numbers of fatalities and injured for each district. This process assumes that the errors due to the many uncertainties inherent in the loss estimation average out, when the estimates are based on averages, using large numbers of buildings (order of 10,000 per district) distributed over large areas (several to 10 kilometers) with varying soil conditions. Based on this information, we calculate loss scenarios for Bucharest, assuming different earthquake depths and distances from the city. http://www.wapmerr.org
S13C-1451
Modelling transmission properties for seismic waves near important cities worldwide for loss estimates in real-time after earthquakes
For estimating losses due to earthquakes in real-time and scenario mode, the intensity of ground motion should be calculated as accurately as possible. We have started to assemble a database for transmission properties of seismic waves beneath and in the vicinity of important cities worldwide. Although one may argue that so little is known about local and regional differences in attenuation that a "one-size-fits-all" approach may be justified, it seems clear that, with time, more information will accumulate to justify differentiating regional attenuation properties. Thus, we begin already now to compile a list of regional attenuation relationships that will cover the world for a database of the loss estimating tool QUAKELOSS2. However, QUAKELOSS2 will allow the user to replace our suggestions for attenuation laws with his own, or one single function for the entire globe. In addition, we are compiling a catalog of microzonation information on cities in seismically active regions. Currently we estimate losses without information on local amplification of ground motion due to soil conditions. This approach works, if a thousand settlements with hundreds of thousands of buildings are affected because then the differences in soil conditions tend to average out. To improve this first approximation approach, we are beginning to model cities by districts. Each district has its own coordinates, its amplification factor due to soil conditions, its population number, and its composition of building stock, in the ideal case where all this information is available. In our mission to construct a worldwide database, some of these factors will not be known, but with time the database will improve. Up to now, we have identified about 50 cities for which some information on soil condition is described in the literature. As a first step we are implementing a system of ranking this information according to the detail and usefulness for our purpose, namely to obtain an amplification factor. The testing phase on how much better we may be able to model intensities of past earthquakes, using microzonation data, will begin shortly. We request that specialists with microzonation information contact us so we can include their data in this worldwide database. In turn, we will assist our collaborators with loss estimates in their country.
S13C-1452
Calibration of Subsurface Amplification Factors Using Surface/Borehole Strong-motion Records from the KiK-net
The Real-time Earthquake Information System (REIS, Horiuchi et al., 2005) detects earthquakes and determines event parameters using the Hi-net (High-sensitivity seismograph network Japan) data in Japan. The system also predicts the arrival time and seismic intensity at a given site before ground motions arrive. Here, the seismic intensity is estimated based on the intensity magnitude which is derived from data of the Hi-net. As the Hi-net stations are located in the boreholes, intensity estimation on the ground surface is evaluated using a constant for subsurface amplification. But the estimated intensities based on the conventionally used amplification constants are not always in agreement with those observed at specific sites on the ground surface. The KiK-net (KIBAN Kyoshin network Japan) consists of strong motion instruments. Each station has two sets of accelerometers, one set is installed on the ground surface and the other one is co-located with a Hi-net station in the borehole. We use data recorded at the KiK-net stations to calibrate subsurface site amplification factors between the borehole and the ground surface. We selected data recorded for over 200 events during the period of 1997 to 2006 in Hiroshima prefecture and calculated the ratios of peak velocity amplitudes on the ground surface ( Asurf) to those in the borehole ( Abor). The subsurface amplification varies from station to station showing dependency on the propagation distance as well as on the incident direction of seismic waves. Results suggest that the site amplification factors shall be described as a function of distance and incident direction, and are not constants. Thus, we derived empirical amplification formulas between Asurf and the peak velocity amplitudes on the engineering bedrock ( Abed) as a function of distance in place of the conventionally used amplification constants. Here, the engineering bedrock is defined as the depth where the S- wave velocity is 600 m/s. The estimated intensities show substantial improvement in the accuracy at most stations as compared with those calculated using conventional constants. When the amplification dependence on the incident direction was accounted for, the estimated intensities somewhat improved. This calibration will help an earthquake early warning system such as REIS provide more accurate intensity estimates.
S13C-1453
Limitation of the Predominant-Period Estimator for Earthquake Early Warning and the Initial Rupture of Earthquakes
Earthquake early warning is an important and challenging issue for the reduction of the seismic damage, especially for the mitigation of human suffering. One of the most important problems in earthquake early warning systems is how immediately we can estimate the final size of an earthquake after we observe the ground motion. It is relevant to the problem whether the initial rupture of an earthquake has some information associated with its final size. Nakamura (1988) developed the Urgent Earthquake Detection and Alarm System (UrEDAS). It calculates the predominant period of the P wave (τp) and estimates the magnitude of an earthquake immediately after the P wave arrival from the value of τpmax, or the maximum value of τp. The similar approach has been adapted by other earthquake alarm systems (e.g., Allen and Kanamori (2003)). To investigate the characteristic of the parameter τp and the effect of the length of the time window (TW) in the τpmax calculation, we analyze the high-frequency recordings of earthquakes at very close distances in the Mponeng mine in South Africa. We find that values of τpmax have upper and lower limits. For larger earthquakes whose source durations are longer than TW, the values of τpmax have an upper limit which depends on TW. On the other hand, the values for smaller earthquakes have a lower limit which is proportional to the sampling interval. For intermediate earthquakes, the values of τpmax are close to their typical source durations. These two limits and the slope for intermediate earthquakes yield an artificial final size dependence of τpmax in a wide size range. The parameter τpmax is useful for detecting large earthquakes and broadcasting earthquake early warnings. However, its dependence on the final size of earthquakes does not suggest that the earthquake rupture is deterministic. This is because τpmax does not always have a direct relation to the physical quantities of an earthquake.
S13C-1454
Investigations Into Early Magnitude Estimation From Predominant Period, Using Synthetic Rupture Models
Considerable interest has been shown in a method for estimating predominant period in the time domain (TpMax), first proposed by Nakamura (1988) and currently being developed for other early warning systems (e.g. Lockman and Allen, BSSA, 2005). Issues still exist as to the causes of the scatter evident in empirical work, and how effective the method is for characterising large events whose time to rupture is longer than the few seconds desired to estimate the magnitude. Our work on applying this method to an aftershock dataset motivated us to investigate the method through the use of synthetic rupture models. The rupture model we use prescribes a stress-drop with a prescribed rise-time over a small patch of the fault surface. This stress-drop is propagated to other patches of the fault according to a prescribed rupture rate. The same finite difference model geometry and fault patch size was then used to model events ranging from magnitude 3.7 to 7.2. Moment Magnitude was calculated directly by integrating the resultant slip on the fault, and TpMax was calculated from seismograms recorded on surface 50 km from the centre of the fault. The initial modelling used a homogenous stress drop, rise-time, and rupture rate. A dataset of 165 events, showed a significant increasing relationship between the TpMax calculation and magnitude. Isolating similar events initiating at the same point on the fault, gave a near straight-line trend. Scatter in the relationship is shown to result from variations in the position, initiation point, stress drop, rise time, and rupture velocity. Low frequency filtering was found to significantly affect the TpMax calculations and trends. Without filtering, the relationship saturated from just after magnitude 6, as the time to rupture becomes longer than the window used to calculate TpMax. However, low frequency filtering actually reduces the time to reach a maximum in the calculation, and this can cause the increasing trend to continue into somewhat higher magnitudes. This mapping may explain some of the previously reported results that TpMax can often be calculated in less time than the time to rupture (Olson and Allen, Nature, 2005). Extensions to this work are being made to look at whether these conclusions remain true for heterogenous rupture, and whether any advantages can be gained by using either displacement or acceleration seismograms in the calculation rather than velocity seismograms.
S13C-1455
A method for Real-time inversion of earthquake source ruptures
With the enhancement of regional seismic networks, real-time seismology becomes a practical way for effective damage mitigation. For example, using P-wave arrival times and amplitudes from only a few stations, a real-time earthquake analysis system developed by Horiuchi et al. (2004) can precisely locate the earthquake within a few seconds and issue an early-warning for the S-wave shaking. This automatic procedure, however, models an earthquake as a point source, which is insufficient to issue shaking intensity or evaluate strong motion distribution caused by a large earthquake, since large earthquakes are usually observed as several asperities. Several methods to rapidly and automatically determine earthquake fault plane or to reveal rupture process have been proposed (e.g., Dreger and Kaverina, 2000; Kuge, 2003; Hori, 2004). Unfortunately, all the methods are time consuming and unpractical for a real-time warning system. In this study, we proposed an automatic method toward real-time determination of finite-fault slip distribution. Instead of waiting for moment tensor solutions of a larger earthquake, we determine the focal mechanism using the amplitude (with polarity) of the P-wave first motions as soon as P-wave arrivals are detected. Of the two nodal planes of the focal mechanism, the fault plane is identified by comparison of moment indexes of the two nodal planes by using P-wave only. Given that magnitude of a larger earthquake is known (it has been practically issued by Horiuchi system), waveform inversion for a finite fault plane source model is then carried out Here we show three simulation results for the 1999 Taiwan Chi-Chi earthquake, the 2007 Noto hanto, Japan, earthquake and the 2007 Chuetsu-oki, Japan earthquake. The inversion results can be obtained in one minute given that the hypocenter location and the magnitude are known (note that all these parameters can be in fact reliably issued the Horiuchi system). Our simulation results show that a real-time inversion of the source process is able to reveal a slip distribution,reasonable and comparable with that of an off-line analysis. Due to time delay for inversion, the kinematic source models might not be helpful for shaking intensity predictions. However, it may be of great importance for tsunami warning since the tsunami wave arrives at a relatively slow speed.
S13C-1456
A Method for Real-Time Magnitude Estimation from Inversion of Displacement Spectra
An effective regional earthquake early warning system (EEWS) requires the capability to estimate earthquake location and magnitude while the event is still occurring. Attenuation relations are then used to predict the spatial distribution of expected peak ground motion amplitudes. To maximize available warning time, early warning algorithms should be capable of producing source and ground motion estimates as soon as possible after the initial event detection (using a few seconds of data on a few stations), and updating these estimates as additional data become available at stations further from the source region. The Spectrum Matching (SM) method is a Bayesian approach to estimate magnitude (and its uncertainties) by matching the observed displacement spectra of available waveforms with theoretical source spectral models. Magnitude and uncertainty estimates are updated at each second after the first P-arrival. The method consists of three main steps: a) comparison between observed data and theoretical model, b) estimate of magnitude and its error, c) introduction of a prior distribution: This inversion technique is based on the comparison between the recorded displacement spectra and a set of theoretical omega-square curves, evaluated for several magnitude values from the range 3.0-9.0 Mω. The misfit between the observed spectra and theoretical curves is measured, defining a cost function based on the L2 norm. The comparison between spectra is performed in the frequency range 1/n-15 Hz, where n indicates the length of the recorded signal in seconds. As the duration of the recorded signal increases, the spectrum becomes richer in low frequencies content and thus allowing more accurate magnitude estimates. For every station recording, the estimated earthquake magnitude at each time step is provided as a probability density function (PDF), which incorporates in its definition the uncertainties related both to the model employed and to the available data. The total PDF is expressed as the likelihood product of all the PDF of magnitude relative to the recording stations at the given time t. To better constrain our magnitude estimation, we used a Bayesian approach. We introduced as a prior the Gutenberg-Richter earthquake size distribution to constrain our initial estimate, and then use the estimated PDF for subsequent updates. Once we start to record the S-wave, we use as a prior the entire P-phase PDF. This guarantees that an outlier observation does not strongly influence the final estimation. The SM method has been tested on a database of records from the 1999 Mω 7.6 Chi Chi, Taiwan earthquake and its aftershocks, the 2003 Mω 6.8 Tottori, Japan earthquake, and a suite of events from Switzerland and California, with magnitudes ranging between 4-7.3 Ml
S13C-1457
Epicenter Estimation Using Empirical Traveltime
A robust and efficient scheme to determine the epicenter in a local seismic network is proposed which uses an empirical traveltime from previous earthquakes, but does not rely on a seismic velocity structure of the region. An empirical traveltime is generated from the relationship between the epicentral distance and traveltime of the first arrival P phase. For a fast convergence of the method, two steps of the location scheme are used. A preliminary location is obtained by a geometrical method and the final epicentral location by a grid search algorithm. Empirical traveltimes are obtained from two different regions, southern California and South Korea, and the method is applied to both regions. Especially for the case of 138 earthquakes in southern California, occurring in 2007, the located epicenters in this study are so close to those from the bulletin that the average misfit is only 2.2 km. In both regions, the location misfits are especially small for earthquakes occurring within seismic networks. Large misfits are related with lack of traveltime information and velocity heterogeneity around the epicenter. As with other location algorithm, this method is also more accurate when the numbers of P arrivals are larger. In most cases of estimations in this study, however, only five or more arrivals can locate the epicenter accurately, which shows the potential for earthquake early warning and the usefulness in such regions where the velocity model is poorly known.
S13C-1458
GEOFON and the Indian Ocean Tsunami Warning System
After the Mw=9.3 Sumatra earthquake of December 26, 2004, which generated a tsunami that effected the entire Indian Ocean region and caused approximately 230,000 fatalities, the German government funded the so- called German Indian Ocean Tsunami Early Warning System (GITEWS) Project. The GEOFON group of GFZ Potsdam was selected to develop and implement its seismological component. In this presentation we describe the concept of the Earthquake Monitoring System and report on its present status.The major challenge for a Earthquake Monitoring System (EMS) within a tsunami warning system is to deliver information about location, size, source parameters and possibly rupture process as early as possible before the potential tsunami hits the neighboring coastal areas. Tsunamigenic earthquakes are expected to occur in subduction zones close to coast lines. This is particularly true for the Sunda trench off-shore Indonesia, but also in the Macran subduction zone off- shore Iran. Key for an Indian Ocean monitoring system with short warning times is therefore a dense real-time seismic network in Indonesia, supplemented by a substantial number of stations in other countries and territories within and around the Indian Ocean. Up to 40 new broadband and strong motion stations will be installed until 2010 with real-time data collection using a private VSAT communication system.The GITEWS EMS Control Center in Jakarta will be based on an enhanced version of the widely used SeisComP software and the GEOFON earthquake information system prototype presently operated at the GFZ-Potsdam (http://geofon.gfz- potsdam.de/db/eqinfo.php). However, the Control Center software under development (SeisComP3) will be more reliable, faster and automatic but with operator supervison. It will use sophisticated visualisation tools, offer the posibility for manual correction and re-calculation, flexible configuration and support for distributed processing. Its large redundancy for algorithms, moduls and hardware assures easy integration into larger multi-sensor, multi- hazard control centers and decision support systems. A first prototype of the EMS Control Center software is already operational. http://geofon.gfz-potsdam.de
S13C-1459
Earthquake Precursors An Example of Data Acquisition and Processing at Science Horizons
Science Horizons has used COTS products, namely its AQUINAS® System and SeismicRadar®, to identify earthquake precursors from seismic data 1,000 miles from the earthquake's epicenter. We have assumed that a precursor has the same characteristics, specifically the same azimuth and horizontal phase velocity, as the earthquake's P-wave. In addition, we have also shown the superiority of seismic arrays to detect precursors. The precursors identified by SeismicRadar® preceded the earthquakes with times ranging from thirteen hours to one minute. Therefore, one can conclude that the same assumption and the same technology should be able to detect numerous precursors that are a few miles from a pending earthquake's epicenter. If this conclusion is shown to be valid in seismic areas, earthquake precursors will no longer be elusive. Then we will be able to create algorithms, based on the precursor's location, duration, and frequency, that reliably predict earthquakes. Science Horizons has a patent pending on this earthquake prediction method. For more information regarding Science Horizons' earthquake prediction method, please go to our website at http://www.horizon.com . http://www.horizon.com