Worldwide EEW Development - Earthworm平台應用於臺灣地震速報預警研究

1 Introduction

1.3 Worldwide EEW Development

The EEW system is becoming a key practical tool for mitigating loss due to seismic events. Depending on the distance to the earthquake, it provides a few seconds to a few tens of seconds warning for people and automated facilities. Currently, many countries have an online operating or experimental EEW system, such as Japan (Nakamura 1988;

Odaka et al., 2003; Horiuchi et al., 2005; Wu and Kanamori 2008b), Taiwan (Wu et al., 1998; Wu et al., 1999; Wu and Teng 2002; Hsiao et al., 2009; Hsiao et al., 2011), Mexico (Espinosa-Aranda et al., 1995; Espinosa-Aranda et al., 2009), the United States (Allen and Kanamori 2003; Wu et al., 2007; Allen et al., 2009; Bose et al., 2009a), Italy (Zollo et al., 2006; Zollo et al., 2009), Turkey (Alicik et al., 2011), Beijing (Peng et al., 2011), and Romania (Bose et al., 2009b).

The station coverage gap (GAP), defined as the angle between epicenter and two adjacent stations, can be used as a metric for evaluating the quality of an EEW report (Wu et al., 1997; Wu et al., 2013a). A dense seismic network can provide a sufficient number of triggered stations to reach the good coverage of seismic stations (e.g., a small value of GAP) within a relative short time after an earthquake occurs. Therefore, it can be a potential solution to provide faster and more reliable earthquake early warnings. However, it is expensive to deploy a large number of traditional seismic stations. Fortunately, recent advances in electrical and mechanical technologies have made it possible to build low-cost seismometers for constructing dense seismic networks. Holland (2003) first monitored earthquakes using seismic data streams from low-cost seismometers and

short-period seismic sensors. The concept of home seismometers has been implemented in Japan (Horiuchi et al., 2009). The Quake Catcher Network (QCN) project is able to rapidly expand and increase the density of ground-motion observations with relative low cost (Cochran et al., 2009). The QCN initiated Rapid Aftershock Mobilization Programs (RAMP) following the 2010 M7.2 Darfield, New Zealand, earthquake (Lawrence et al., 2014), respectively. The results demonstrated that the QCN can be used to detect and locate moderate to large earthquakes, and estimate their magnitudes using ground-motion parameters. The Self-organizing Seismic Early Warning Information Network (SOSEWIN) has been tested in Istanbul based on wireless communications (Fleming et al., 2009).

1.4 Taiwan EEW Development

Over the years, many studies have been conducted regarding the development of an EEW system in Taiwan. In 1995, an earthquake rapid reporting system began operating on the basis of 16-bit strong-motion seismometers and was a type of early-stage EEW system for Taiwan (Wu et al., 1997). Although the system could not issue warnings prior to large ground shaking, it provided rapid reporting within 102 s for the Chi-Chi earthquake and was the leading technology at that time (Wu et al., 2000). As EEW system necessity demanded, the Central Weather Bureau (CWB) was the first to test an EEW prototype system within the Hualien area in Taiwan. To reduce reporting times and provide early warnings for distant metropolitan regions, a new idea, based on the prototype system, was proposed for applying the subnetwork method to earthquake monitoring (Wu et al., 1999). Using the subnetwork concept and ML10, a quick magnitude

determination method (Wu et al., 1998) that adopted 10 s records following the first P-wave arrival, the current EEW system (the virtual subnetwork [VSN] system) was built and achieved an average 22 s reporting time (the time between an earthquake’s origin time and the time the EEW system issues a report) (Wu and Teng, 2002). However, due to the limits of the ML10 method, the reporting time could not be reduced to within 10 s.

To further reduce reporting times, the P-wave method, based on the peak amplitude of displacement records (Pd) for the vertical component using a 3 s time window for magnitude determinations (Wu and Zhao, 2006), was tested and operated (Hsiao et al., 2009, 2011). The CWB has recently upgraded seismic facilities within the original seismic network and deployed 30 borehole stations, as well as one cable-based ocean-bottom seismic station. At the same time, to enhance the density and coverage of station distributions, real-time seismic data streams from various seismic networks were integrated using the Earthworm platform, a program originally developed by the U.S.

Geological Survey (Johnson et al., 1995). Based on the above, an Earthworm-based EEW prototype system was constructed and has been tested since 2007 (Hsiao et al., 2011;

Chen et al., 2012).

In addition, some experimental on-site EEW systems have been tested and operated as well. Wu et al., (2006) determined the relationships between the earthquake magnitude and characteristic parameters from the first three seconds of the P-wave. They demonstrated that single-station approach can be used to estimate earthquake magnitudes well. Wu et al., (2011) demonstrated that the on-site EEW system can provide valuable information to the Taiwan High Speed Railway in the 2010 JiaSian earthquake. The National Center for Research on Earthquake Engineering (NCREE) has developed neural

network method for predicting structural response in on-site EEW system (Lin et al., 2011). The on-site EEW system has been put into practice in elementary schools in Taiwan (Lin, 2011). Some Micro Electro Mechanical Systems (MEMS) sensors have been developed for EEW system. The National Taiwan University (NTU) and the San Lien Corporation, a high-tech oriented company (http://www.sanlien.com.tw), have developed an accelerometer, named Palert, based on MEMS technology. The Palert Seismic Network (PSN) has been tested and operated for both on-site and regional EEW systems by NTU since 2010 and is capable of providing high quality and stable data streams for earthquake monitoring (Wu et al., 2013b; Hsieh et al., 2014; Wu, 2014).

1.5 Earthworm for EEW system

Earthworm is a popular software for real-time earthquake monitoring. It has been used all over the world. There are five advantages of the Earthworm system. First, Earthworm is free and open source. It makes the system operator easy to modify it and save cost. Second, Earthworm can receive real-time data streams from different kinds of seismic instruments. Even those sensors are made from different companies, Earthworm is able to integrate all data in the same platform. Third, Earthworm was composed by modules. Users can take different set of modules to construct their own Earthworm system. Moreover, because modules are running separately, users can create new modules without disturbing current modules. Forth, in the same computer, Earthworm uses shared memories for communicating message with other modules. Among different computers, Earthworm use TCP/IP protocol to exchange messages. In this way, Earthworm can efficiently exchange message among modules and process data in parallel.

Compared to the former EEW system in Taiwan, Earthworm system provides an excellent opportunity to improve the construction of the EEW system. Instead of using telephone line for real-time data transmitting in old EEW system, for modern system, data are packed as 1-sec length packet and transmitted based on TCP/IP protocol. Earthworm can integrate all data and be a server to provide real-time waveforms to clients as long as the internet is available. In addition, Earthworm can process data in memory. It is more efficient than processing data using text or binary files.

1.6 Dissertation Plan

In this dissertation, the fundamental EEW concepts and the review of EEW researches are introduced in chapter 1. Methods of location and magnitude estimations, and EEW modules are described in chapter 2. EEW system in CWB is described in chapter 3. Integrating low-cost seismic network and official seismic network is described in chapter 4. A case study of the Mw 7.6 Chi-Chi earthquake is described in chapter 5.

Discussion and Conclusions are described in chapter 6.

Chapter 2 Methods and EEW Modules

2.1 Earthquake Location Estimation

Consider a one-dimensional continuous velocity model, shown as Figure 2-1. In this case, the ray equation becomes:

where the velocity, v(z), is a function of depth (z), ds is the differential of ray path. In Figure 2-1, the direction cosines are:

Then, a ‘Snell’s Law’ can be obtained:

where p is called the ray parameter. The velocity is given by:

Where go and gz are constants, z is depth. In Figure 2-1, the center of this arc is given by:

The travel time of this linear velocity is:

Finally, the spatial derivatives of travel time, T, at the source are:

In the procedure of the Geiger’s method (1910, 1912), a half-space model was used to calculate the predicted travel times. Figure 2-2 shows the relationship between the travel time and the distance.

Figure 2-1. Geometry for velocity given by v = v(z) (Lee et al., 1992).

Figure 2-2. Travel time vs. distance for layer over half-space model. (Lee et al., 1992)

2.2 Earthquake Magnitude Estimation

To precisely measure the size of an earthquake, we must take a certain length of time window extending after the P-wave arrival until the enough observed waveforms are available. This time window has variant values depending on different EEW algorithms and is one of the components adding a delay to the overall alert time (Behr et al., 2015).

For EEW purposes, it is necessary to detect earthquake magnitude in the beginning stage of the earthquake occurrence. Wu and Teng (1998) used an empirical method to correlate local magnitude and the predicted magnitude over 10 seconds after the first P-wave arrival is detected. Recently, P-wave methods has been widely studied and implemented in EEW systems. There are two kinds of the P-wave methods. One is associated with the frequency content of the initial waveforms. Allen and Kanamori (2003) has proposed a method based on the predominant period (τp) measured over a varying time window after the P-wave arrival. When 1-, 2-, 3-, and 4-s time window of data are available, the τp

values are measured and the magnitude would be updated. In addition, the average period parameter (τc) of the initial 3-s P waves can be used for estimating magnitudes (Wu and Kanamori, 2005). The other kind of P-wave method is associated with the amplitude content of the initial waveforms. Wu and Zhao (2006) take the peak amplitude in vertical displacement (Pd) over a 3-s time interval after P-wave arrival. They showed that the upper limit of the magnitude prediction is 6.5 because the time window is too short to contain whole rupture information from larger events. Using the combinations of P and S wave signals, Zollo et al., (2006) demonstrated that the peak displacements measured in 2-s P-wave time window and 2-s S-wave time window can be correlated with magnitude in the ranged from 4.0 to 7.4. Lancieri and Zollo (2008) used peak displacement over 2- and 4-s P-wave time window and 1- to 2-s S-wave time window with Bayesian approach to estimate magnitude at each time step.

2.2.1 τ

Method

Following the procedure from Wu and Kanamori (2005a), take the ground-motion displacement, u(t), and velocity, u’(t), from the vertical component record and compute the following ratio r by

where the integration is over the time interval (0, τ0) after the onset of the P wave.

Usually, τ0 is set at 3 sec. Using Parseval’s theorem,

where uˆ(f) is the frequency spectrum of u(t), and <f ²> is the average of f ² weighted by

|uˆ(f)|² . Thus,

can be used as a parameter representing the period of the initial portion of the P wave.

The largerτc is, the larger the event is. Following Wu et al., (2007), a regression equation can be used for magnitude estimation:

2.2.2 P

Method

The peak amplitude of the initial P-wave displacement, Pd, reflecting the attenuation relationship of the ground motion with distance, can be used as an amplitude parameter to predict sizes of earthquakes. Therefore, if we can determine the attenuation relationship of Pd, then we can use Pd to estimate the magnitude when the hypocentral distance is available. Only vertical-component records are used to determine Pd. The seismograms are integrated once or twice to obtain the displacement and then a 0.075 Hz high-pass recursive Butterworth filter is applied to remove the low-frequency drift after the numerical integration. We assumed a linear relationship among the logarithmic Pd, the magnitude M and the logarithmic hypocentral distance R:

where A, B and C are constants to be determined; R is hopocentral distance; M is magnitude; the units of Pd and R are cm and km, respectively.

2.3 Earthworm System

Earthworm is a software originally developed by the United States Geological Survey (USGS) since 1994. The preliminary purpose was to construct a system which is able to quickly notify earthquake information to the public. Earthworm has been developed and improved continuously by users because Earthworm is a free and open-source software. Currently, Earthworm has become a robust and well tested software. Many earthquake monitoring center use this software to detect earthquakes and archive waveform records. The software has also been successfully extended to volcano observation and is also used in many tsunami centers.

Because of two main components in the earthworm, the system can be enlarged and become more dedicate. Figure 2-3 shows the two main components in the Earthworm (module and shared memory). With shared memories, modules can exchange information directly in the memory. Every Earthworm system can have different compositions of modules and shared memories. Based on this design, the Earthworm system is very flexible and maintainable.

Figure 2-3. Two main components in the Earthworm diagram. The rectangle represents a module; the circle represents a shared memory. Modules can exchange data with shared memories.

Earthworm is a command-line based system. It is not easy to install and be understood.

The procedure of Earthworm installation described in Appendix A.1 is useful for quickly setup Earthworm system. In addition, a summary of the Earthworm features are described in Appendix A.2.

2.4 EEW Modules

An Earthworm diagram that describes data flow within the eBEAR system is provided in Figure 2-4. For system calibration, we ran the system in offline mode using the TANKPLAYER module. To receive real-time data for online operations, we applied the IMPORT module. The three circles provided in Figure 2-4 represent shared memories within Earthworm. The first shared memory, WAVE_RING, contains waveform data that can be processed using the PICK_EEW module to determine P-wave arrivals, as well as the peak amplitudes for P-wave displacement (Pd), velocity (Pv), and acceleration (Pa) within a 3 s time window. The second shared memory, PICK_RING, not only contains

information from the PICK_EEW module, but also provides information to the TCPD module for generating earthquake messages, including source parameters. When an earthquake occurs, the TCPD module may update information for the event and create earthquake messages. Updated earthquake messages are stored within the third shared memory, HYPO_RING. At the end of the process, the DCSN module filters earthquake messages using specific criteria (as discussed later) and generates EEW reports for broadcasting as an XML-formatted file.

Figure 2-4. A flowchart for data processing within the eBEAR system.

2.4.1 PICK_EEW Module

The original Earthworm module, PICK_EW, requires time to check the seismic coda term within the auto-picking procedure. The work is time consuming and not suitable for EEW systems. Therefore, we created a new module named PICK_EEW by revising the module to run without checking the seismic coda term. To avoid false pickings caused by

background noise, we also added two parameters, Pa and Pv. Because seismic waveforms from field stations have different noise levels depending on vibrations from the natural

Figure 2-5 displays the procedure for P-wave autopicking. The PICK_EEW module declares possible picks based on the short-term average (STA) and long-term average (LTA) algorithm. To become a candidate pick of a seismic trace, the ratio of STA/LTA should be greater than two times a certain threshold. Following a pick based on the threshold, and to distinguish ground noise and the seismic signal, we considered three additional conditions: the number of zero crossings, the signal-to-noise ratio, and the Pa and Pv. Using this procedure, the module was able to qualify the candidate pick as a valid seismic pick. In practice, because each seismic station has different background noise, we tested different sets of picking parameters by performing an offline test.

Figure 2-5. A flowchart of the algorithms designed for the PICK_EEW module.

2.4.2 TCPD Module

After the TCPD modules jointly trigger using a space–time window based on

expected travel times, the event hypocenter is estimated using two steps. For determining the event epicenter, the module first adopts Geiger’s method, an inversion process using a half-space velocity model in which velocity linearly increases with depth. For estimating event depth, the module then uses a grid search method with depths ranging from 10 to 100 km in steps of 10 km. Theoretical travel times to each station are calculated and compared to those observed at each depth. Finally, the depth with minimum residuals and the epicenter determined by Geiger’s method are considered as the event hypocenter. The procedure is performed within the TCPD module via an updating process. At the beginning of the process, after at least six picks of seismic waveforms, the TCPD module begins to locate an event. When the root mean square of travel-time residuals resulting from the inversion process is larger than 0.8, the pick with the largest travel-time residuals is removed and the inversion process is again performed. When additional picks of seismic waveforms participate, the procedure of hypocenter determination is repeated and the estimated hypocenter is updated.

Earthquake magnitudes are predicted using the initial portion of P-wave peak displacement Pd within the 3 s time window. Following a double-integrated, strong-motion, and integrated broadband, the PICK_EEW module applies a 0.075 Hz high-pass filter to displacement records. The Pd value is then used to estimate magnitude (MPd) based on empirical formula. The empirical formula for borehole stations has not yet been established. Earthquake magnitude is estimated by obtaining an average for each MPd value from each seismic station. However, the false picking of P-wave arrivals, the directivity effect, and site effects may lead to unreasonable MPd values. For obtaining robust estimations of magnitude and to reduce errors, three steps are applied. First, only

MPd values within one standard deviation of the dataset are used. Next, each record is weighted according to P-wave travel-time residuals. The weighting factor is expressed as

in which Wi is the weighting factor of each MPd value and Ri (in seconds) is the P-wave travel-time residual for each corresponding MPd value. Finally, a weighted average for waves propagate away from the epicenter. As a result, the TCPD module determines the earthquake message and continuously updates that message. We propose that the numbers of updating earthquake messages will increase quickly and will be significant for large and local earthquakes. In contrast, for small earthquakes or for noise, the number of updating earthquake messages will increase slowly and will be small. Therefore, if the EEW system determines a large number of updating earthquake messages for an ongoing earthquake, we consider the EEW information as a reliable warning. To prevent false alarms, the DCSN module always skips the first and second earthquake message generated from the TCPD module. The third earthquake message is the first EEW report

to users. The EEW report is written in an XMLformatted file for broadcasting. The EEW report is updated either when differences in the magnitude or the epicenter are larger than 0.5 or 20 km, respectively, as compared to the last EEW report. A user display pops up automatically when an XML-formatted message is received. The display estimates the seismic intensity, the wave fronts of P- and S-waves, and the remaining warning time (defined as the time between the reporting time and the arrival of the S wave to the target area). If the EEW report is updated, the user display directly changes the location of the epicenter and again re-estimates EEW-related parameters.

The DCSN module takes the EEW report from the HYPO_RING for other applications such as generating the XML-formatted messages for clients running the EEW display and warning program provided by the CWB (Chen et al., 2015). The DCSN module will also pop up EEW messages on the corresponding CWB staff’s computers,

在文檔中 Earthworm平台應用於臺灣地震速報預警研究 (頁 16-0)