Abstract—Regional earthquake early warning (EEW) systems are not suitable for Taiwan, as most destructive seismic hazards arise due to in-land earthquakes. These likely cause the lead-time provided by regional EEW systems before a destructive earthquake wave arrives to become null. On the other hand, an on-site EEW system can provide more lead-time at a region closer to an epicenter, since only seismic information of the target site is required. Instead of leveraging the information of several stations, the on-site system extracts some P-wave features from the first few seconds of vertical ground acceleration of a single station and performs a prediction of the oncoming earthquake intensity at the same station according to these features. Since seismometers could be triggered by non-earthquake events such as a passing of a truck or other human activities, to reduce the likelihood of false alarms, a seismometer was installed at three different locations on the same site and the performance of the EEW system for these three sensor locations were discussed. The results show that the location on the ground of the first floor of a school building maybe a good choice, since the false alarms could be reduced and the cost for installation and maintenance is the lowest. Keywords—Earthquake early warning, Single station approach, Seismometer location I. INTRODUCTION VER the last two decades, effective EEW techniques have emerged due to advancements in digital seismology, communications, automatic processing, and algorithms for the rapid estimation of earthquake parameters [1]. Based on the requirements of information for algorithms to estimate earthquake parameters, EEW techniques can be divided into two groups: regional warning and on-site warning. Generally, since regional warning techniques leverage information of several stations next to the epicenter, the accuracy of earthquake parameter estimation of regional warning techniques is usually higher than that of on-site warning techniques. However, for regions closer to the epicenter where seismic intensity is usually much higher than regions outside, the lead-time before a destructive wave arrives provided by a regional warning can be null. On the other hand, an on-site warning can provide more lead-time at the region closer to an T. Y. Hsu is an associate research fellow with the National Center for Research on Earthquake Engineering, Taipei, Taiwan. He is also an adjunct assistant professor with National Taipei University of Technology, Taipei, Taiwan. (phone: 886-2-6630-0863; fax: 886-2-6630-0858; e-mail: dysheu@ncree.narl.org.tw). S. Y. Wu, S. K. Huang, H. W. Chiang, K. C. Lu and P. Y. Lin are with the National Center for Research on Earthquake Engineering, Taipei, Taiwan. (e-mail: 1306026@ncree.narl.org.tw, sghuang@ncree.narl.org.tw, hwchiang@ ncree.narl.org.tw, kclu@ ncree.narl.org.tw, pylin@ ncree.narl.org.tw). K. L. Wen is with the National Center for Research on Earthquake Engineering, Taipei, Taiwan. He is also with National Central University, Taoyuan, Taiwan. (e-mail: wenkl@ncree.narl.org.tw). epicenter since only the seismic information on the target site is required. Therefore, an increase in accuracy and lead-time of an on-site warning is a key point in improving the effectiveness of EEW techniques. An on-site warning system issues an alarm a few seconds after a trigger based on the initial P-wave motion at a single station. According to the records of EEW stations at the National Center for Research on Earthquake Engineering (NCREE), the present on-site warning system may be triggered by certain vibration signals that are not caused by an earthquake movement, which may consequently lead to many false alarms at the station. Normally, seismometers for EEW systems are mounted on a surface of a free field where no civil structures are around within a certain range. However, these seismometers could be triggered by non-earthquake events. The EEW stations of NCREE in Taiwan are mostly implemented at schools where regular human activities take place. In order to reduce the possibility of false alarms due to non-earthquake events, a seismometer was installed at three different locations, namely on the ground of the first floor of a school building, on a concrete foundation with a depth of 2 meters, and in a downhole with a depth of 40 meters. This paper discusses the performance of the EEW system for these three sensor locations. II.PREDICTION METHODS FOR ON-SITE EEW SYSTEMS A. Support Vector Regression Method Recently, a new method for the estimation of seismic intensity using support vector regression (SVR) was developed [2]. Estimating the predicted peak ground acceleration (PGA) based on the SVR method is achieved by two steps. The first step extracts six P-wave features from the first three seconds of the vertical ground acceleration after the arrival of the P-wave. The second step predicts the PGA using a regression model established by supervised learning with the P-wave features as the inputs. The six P-wave features are the predominant period (TauC), peak acceleration (Pa), peak velocity (Pv), peak displacement (Pd), cumulative absolute velocity (CAV), and the integral of the squared velocity (IV2). The peak values, Pa, Pv, and Pd are defined as the maximum values of absolute acceleration, absolute velocity, and absolute displacement, respectively. The other related formulae are summarized as: ! ' & ( & * ! $     * !     (1)   * "!%" (2) Performance of On-site Earthquake Early Warning Systems for Different Sensor Locations Ting-Yu Hsu, Shyu-Yu Wu, Shieh-Kung Huang, Hung-Wei Chiang, Kung-Chun Lu, Pei-Yang Lin, Kuo-Liang Wen O World Academy of Science, Engineering and Technology International Journal of Geological and Environmental Engineering Vol:10, No:8, 2016 830 International Scholarly and Scientific Research & Innovation 10(8) 2016 ISNI:0000000091950263 Open Science Index, Geological and Environmental Engineering Vol:10, No:8, 2016 publications.waset.org/10005106/pdf