On-site Earthquake Early Warning Models Based On S-Wave And P-Wave Spectral Ratios

Earthquake early warning(EEW)is an emerging real-time earthquake risk mitigation technology.In an EEW system,the prediction of the ground-motion parameters at the target location is a key technical part.Conventional EEW algorithms use P-wave records to locate the earthquake,estimate the magnitude,and then use the ground motion prediction equations(GMPEs)to estimate the ground motion parameters of the S-wave.The error of hypocenter locating,magnitude estimating,and the inherent uncertainty of GMPEs may lead to a relatively large uncertainty to the estimated S-wave ground-motion parameters.The O n-site EEW models,such as the Pd-PGV models,use the correlations between the P-wave ground motion parameters and the S-wave ground motion parameters.But the ignorance of effects from some of the source parameters,propagation path,and site conditions would lead to an impaired accuracy.It is necessary and possible to develop an improved On-site EEW algorithm.In this thesis,the prediction algorithm of the S-wave ground motion parameters at the target site has been investigated,and a new O n-site EEW model,namely the model of the response spectral ratios between S-wave and P-wave(referred to as S/P model)is proposed.After the first 3s of P-wave being recorded and processed by an EEW system,the predicted response spectrum of the subsequent S-wave can be obtained by multiplying the response spectrum of the first 3 second of P-wave with the predicted spectral ratios from the S/P model.By comparing the predicted S-wave spectral with a preset alarm threshold,it is possible to determine whether an alarm needs to be triggered.The S/P model enriches t he EEW algorithms and improves the prediction accuracy of the S-wave ground-motion parameters at the target site.The contents of this thesis are summarized as follows:1.In this thesis,the scaling relationship between the response spectrum of S-wave and the response spectrum of the first 3 seconds of P-wave was studied,based on earthquake records from Japan,the western United States,and southwestern C hina.The results show that the main influencing factors of S/P spectral ratios are magnitude,hypocentral distance and site condition.For relatively frequent earthquakes with a moderate magnitude,the S/P spectral ratios do not depend on the magnitude.Within moderate distance which an EEW is practically necessary,the S/P spectral ratios do not depend on the hypocentral distance.Site conditions have a significant effect on the S/P spectral ratios and the site parameters can be easily obtained for the target site.In the subduction zone such as Japan,the S/P spectral ratios are basically unaffected by the earthquake source category and this is a significant advantage of the proposed model,because earthquake category cannot be easily obtained from the first 3s of P-wave.The focal mechanism has little effect on the S/P spectral ratios and is not statistically significant at most spectral periods.Thus,the focal mechanism does not need to be included in this model.2.Three separate S/P models are presented in this thesis,one for the subduction zone in Japan,one for the Western United States,and one for the Southwestern part of China.3.The spectral of S-wave was predicted using the S/P model and the recorded value of the spectral of the first 3s of P-wave.And then,the predicted and recorded values of S-wave spectral are compared.The results show that the accuracy of the model is considerable.4.According to the emergency response plan for the EEW system for the C hina High-Speed Railway,the peak ground acceleration(PGA)= 40,80,100 or 120 gal were selected as the alarm threshold,and the EEW performance of the S/P model was tested.The proportion of correct alarms,false alarms,and missed alarms are counted.The results show that the S/P model performed reasonably well,the correct alarm rates are much larger than the false alarm rates and the missed alarm rates at most EEW thresholds,suggesting that the S/P model can be effectively applied to the high-speed railway EEW.5.The prediction variables of the S/P model were expanded.In addition to the response spectrum and PGA,models for peak ground velocity(PGV),the spectral intensity SI,and the instrumental intensity IJMA were also developed based on the same principles of the S/P model.