KEYWORDS Earthquakes; MEMS seismic stations; trigger data; warning time delays. Page 144

Transcription

1 Event Detection Time Delays from Community Earthquake Early Warning System Experimental Seismic Stations implemented in South Western Tanzania Between August 2012 and December 2013 Asinta Manyele 1, Alfred Mwambela 2 ¹Electronics and Telecommunications Engineering Department, Dar es Salaam Institute of Technology, Tanzania ²Electronics and Telecommunication Department & university of Dar es Salaam, Tanzania ABSTRACT Urban areas in South Western Tanzania (SWTZ) are vulnerable to seismic risk due to presence of East African Rift Valley (EAR) system. To protect these communities from earthquake hazards, the Community Earthquake Early Warning System (CEEWS) has been tested for providing earthquakes impending warnings. The network of MEMS accelerometer sensors were deployed to catch seismic waves propagating away from earthquakes and transmit triggers data to Quake- Catcher Network (QCN) server. In this study the efficiency of CEEWS based on observed data delays for event detections using station triggers transmitted via landline and mobile telephone based internet to QCN server is presented. Analyzing received triggers at the QCN server from SWTZ stations, results for observed mean event detection delays ranged between 6.7 and 14 seconds, but the actual individual station delays ranged from less than 1 second to 14 seconds. With these mean observed detection latency, it is possible to provide earthquake warnings of few seconds for people to get to a safe location like taking cover under sturdy desk to minimize earthquake risks across SWTZ. KEYWORDS Earthquakes; MEMS seismic stations; trigger data; warning time delays I. INTRODUCTION The Community Earthquake Early Warning (CEEW) is a system that aims to issue timely warnings before the earthquake produces catastrophic damage in the targeted areas. The CEEWS uses hybrid type of Earthquake Early Warning (EEW) systems where sensors are deployed in the protected area, as well as in remote sites near the expected earthquake epicenters. When a large earthquake occurs and seismic waves propagate away from the epicenter, it reaches a number of seismic sensors at a time according to their epicentral distances. Sensor station computer reached by the event processes the detected real-time seismic data to capture P-wave onset parameters and send them to server for predictions of the later coming S-wave parameters at target sites. The predicted S-wave parameter that include arrival time and the expected amplitude of ground shaking at target sites, are used into deciding sites to warn as well as type of warning to issue. The CEEWS methodology can provide warning before the arrival of S-waves, which usually brings the strong ground shaking that causes most damages. Other types of EEW systems are operational or in testing stages in many seismically active regions where hazards occur frequently [3]. The amount of warning time possible with CEEWS depends on the speed of the system to detect the event and process data and the distance of the target site from the event [1]. Generally, the speed of the CEEW system relies on a dense sensor network deployed near possible earthquake sources to provide quick detection. Because it is expensive to deploy a large number of high quality and expensive seismic sensors, recent advances in sensor technologies provide solution of low-cost Micro Electro Mechanical System (MEMS) sensors for construction of dense community seismic networks [5]. The Quake Catcher Network (QCN) project is example of initiative for building denser seismic network using low cost MEMS sensors around the world [4]. The Rapid Aftershock Mobilization Programs (RAMP) following the 2010 M 8.8 Maule, Chile, earthquakes and the 2010 M 7.2 Darfield, New Zealand, earthquakes are examples of the use of QCN based low cost denser seismic network in earthquake disaster managements [2], [5]. The CEEWS trigger data transmission depends on MEMS sensors installed on internet connected computers to trigger up on the detection of threshold level of ground motion and the computer to send trigger data to server after pre-processing. Therefore, apart from the time the earthquake takes to reach the required number of sensors and the time for the CEEWS algorithm to process the data, the time taken to transmit data to the server also adds to system event detection delays. When the speed of data communication networks in a region is fast and reliable, small time delay will be imposed on the event detection time but with slow and unreliable transmission links, significant time delays will be realized. Page 144

2 The CEEWS event detection delay is defined in this work as the time difference from seismic onset detection at the sensor station and the moment that the event is confirmed and registered at the QCN server. The event detection delay time at the station includes time the seismic signal travel to the station from the event epicenter, the time taken by the station computer to record two seconds window waveform length of seismic data as well as the time for the algorithm to estimate magnitude of event at the station. The event detection time delay at the server, includes the time taken for the sensor computer to transmit trigger data to the server, time used by the QCN server algorithm to confirm the event and assign event parameters (magnitude and location), as well as time for the server to post event parameter into its web site for further processing by CEEWS. For QCN network deployment in California and other regions, real-time algorithms are able to determine the parameters of events with median detection times of 9.1 seconds after the origin time. The primary delay in determination has been found to be the propagation of the P waves from the source to threshold of 5 7 stations as well as latencies to transfer the ground-motion parameters from the station to the central server over the Internet [5], [7]. Using trigger data recorded by experimental MEMS sensor stations in Tanzania between August 2012 and December 2013, the study presents the analysis of the seismic station data delay from seismic sensors to QCN server. II. SEISMICITY OF THE REGION Using the U.S. Geological Survey catalogue, Fig. 1 shows the earthquakes with magnitude greater than 3 recorded between 1973 to May 2012 in Tanzania [6]. Fig.1 Events (red circles) recorded between 1973 and May 2012 in Tanzania. From Fig. 1, most earthquake epicenters are in line with the position of Eastern and Western branches of East African Rift system that crosses in Tanzania. The study concentrates on the earthquake epicenters on the western branch of EAR system, between longitude 28 and 35, between latitude -2 and -12. III. METHODOLOGY For detecting earthquakes using MEMS sensors and estimating the warning time delays, experimental MEMS sensor stations were installed as shown in Fig. 2. That is, sensor station sites were located closer to major population centers in the region which includes Kyela, Tukuyu, Mbeya city, Namanyere, and Sumbawanga town. For purpose of exploring other unknown earthquake epicenters in Tanzania, experimental MEMS sensor stations were also installed in Dar es Salaam city. Page 145

3 Fig. 2 Location of Experimental MEMS Seismic Sensor Stations in Tanzania. Red circle is the event detected during the test, car-like figures are the USB connected MEMS sensor, and laptops are the built-in laptop MEMS sensor stations. Seismic stations sent recorded trigger data packets to QCN server via internet during the test period where triggers were registered with identification numbers. The parameters in the sent trigger data packet included the trigger time of the event, the maximum amplitude of the seismic wave in the first two second window of the event and the estimated magnitude of the event at trigger. IV. DATA COLLECTION Trigger data received by QCN server from experimental stations within the time period of August 2012 to December 2013 were downloaded and sorted according to recording stations as well as the used internet service provider type [7]. Because seismic stations were operational only during working hours and when the station computers were in use, trigger data experienced several delays before transmission due unavailability of the communications when the computer is turned off after working hours before transmission. In order to reach meaningful estimation of system delays, 60 seconds delay was set as the maximum delay that the CEEWS can accept and trigger data with delays greater than 60 seconds were not considered for analysis. The trigger data were grouped according to type of internet used (Landline and cellular phone) as well as station locations before performing statistical analysis. V. DELAY RESULTS Station trigger data delay results are presented for three groups of stations and for different communication links, in order to compare the expected CEEWS event detection delays for warning issuance. A. Event Detection Delays Result for Stations in Dar es Salaam For Dar es Salaam based seismic stations, observed event detection delays ranged from 0.02 seconds to 60 seconds for wireless based internet services and 0.12 seconds to 60 seconds for landline based internet services. The mean, maximum and minimum event detection delays were seconds, 60 seconds, and 0.02 seconds for landline internet services, and 9.22 seconds, seconds, and 0.12 seconds for landline based internet services. Fig. 3 presents the result for time delays variations against the trigger dates for wireless and landline internet services. (a) Trigger delays for Landline internet links (b) Trigger delays for Wireless internet links Fig.3 Delays Results for Dar es Salaam Stations Page 146

4 From Fig. 3(a), most station event detection delay for landline based communication links concentrated between 0.12 seconds and 12 seconds. According to Fig. 3(b), most station time delays for wireless based internet services, were observed between 0.02 seconds and 15 seconds. B. Event Detection Delays Result for Stations in Rukwa region Event detection delays results from the data downloaded from QCN database for seismic station in Sumbawanga and Namanyere towns and transmission via internet services are presented in this section. The mean detection delays were 8.6 seconds and seconds, for landline and wireless internet service, respectively. The maximum and minimum delays were 27.3 seconds and 0.18 seconds for landline, and seconds and 1.18 seconds for wireless internet services. The variations of event detection delays according to trigger dates are presented in Fig. 4. (a) Detection delays for Landline internet links (b) Detection delays for Wireless internet links Fig.4. Delays Results for Rukwa Stations From Fig. 4, most station event detection delay for landline based communication links were concentrated between 0.18 seconds and 10 seconds. For wireless based internet services, most observed station detection delays were between 1.18 seconds and 20 seconds. C. Event Detection Delays Result for Stations in Mbeya region Results for event detection delays from Stations in Mbeya city, Tukuyu and Kyela towns are all grouped under Mbeya stations. For landline based communication links, the observed station detection delays had a mean of 6.72 seconds, minimum of 1.14 seconds and maximum of seconds. For mobile phone based internet link, the observed mean detection time delay was 8.37 seconds, minimum was 0.07 seconds and maximum was seconds. The variations of event detection delays according to trigger dates are presented in Fig. 5. (a) Delays for Landline Internet Link (b) Delays for Wireless Internet Link Fig.5 Delay Results for Mbeya Stations Page 147

5 According to Fig. 5, most station event detection delays for landline based communication links concentrated between 0.18 seconds and 10 seconds. For wireless based internet services, most observed station detection delays were between 1.18 seconds and 20 seconds. Also the delays observed across the region are compared among station for similar internet services, as shown in Fig. 6 and Fig. 7. (a) Dar es Salaam Stations (b) Rukwa Stations Fig. 6 Delays for Landline internet links (c) Mbeya Stations (a) Dar es Salaam Stations (b) Rukwa Stations Fig. 7 Delays for Wireless internet links (c) Mbeya Stations According to Fig. 6, event detection delays using landline internet is better for Mbeya stations as compared to Dar es Salaam and Rukwa. From Fig. 7, event detection delays in Dar es Salaam are better (about 10 seconds) than the delays in other stations for wireless internet links. VI. CONCLUSIONS With the mean event detection time ranging from 6.7 seconds to 14.1 seconds, implementations of CEEWS in Tanzania to minimize earthquake risks is feasible. Considering the positions of major population centers in relation to known earthquake epicenters of at least 46 km, a case of magnitude 7.4 earthquake of 1919, several towns and cities can be warned if the CEEWS is operational. ACKNOWLEDGMENT We thank various sponsors of Quake-Catcher Network project through Stanford University for providing MEMS sensors and monitoring facilities available for us and to make this project possible. REFERENCES [1] A. Manyele, and A. Mwambela, Feasibility Study of Community Earthquake Warning System Proposed for Mbeya City and Surrounding Regions, OJER> Vol.3 No.3, August [2] A. I Chung, J.F Lawrence, and C. Christensen, Evaluating the Integrability of the Quake-Catcher Network. Proceedings of the 10th International ISCRAM Conference Baden-Baden, Germany, May 2013 [3] R.M Allen,, P. Gasparini, O. Kamigaichi, and M. Böse,. The Status of EarthquakeEarly Warning around the World: an Introductory Overview, Seism. Res. Lett., 80(5),doi: /gssrl , [4] E. S Cochran, F. J.Lawrence, C. Christensen, and R. S Jakka, The Quake-Catcher Network: Citizen Science Expanding Seismic Horizons. Seismological Seismol. Res. Lett. 2009, 80, Page 148

6 [5] J. F. Lawrence, S. E. Cochran, A. Chung, A. Kaiser, C. M. Christensen, R. Allen, J. W. Baker, B. Fry, T. Heaton, D. Kilb, M. D. Kohler, and M. Taufer, Rapid Earthquake Characterization Using MEMS Accelerometers and Volunteer Hosts Following the M 7.2 Darfield, New Zealand, Earthquake, Bulletin of the Seismological Society of America, Vol. 104, No. 1, pp , February 2014, doi: / , [6] USGS Web Site 2015,. [7] QCN Web site quakecatcher.net/sensor/trigger.php?. Page 149