| As one of the largest natural disasters on Earth,earthquakes have caused significant damage and unpredictable consequences,leading to tremendous losses in both the natural environment and human life and property security.Earthquake early warning,as a new method to mitigate earthquake disasters,has rapidly developed in recent years and is gradually being applied.Single-station earthquake early warning is an important approach commonly used in regions with scarce seismic stations or where establishing an effective real-time seismic network is not feasible.Single-station localization is a crucial step in single-station earthquake early warning and often involves estimating the epicentral distance and azimuth using seismic information obtained from a single station.In this study,focusing on earthquake early warning localization,we conducted research on single-station localization methods.Initially,the traditional B-Δ method was applied to calculate and analyze the epicentral distance,revealing the need for improved accuracy and stability.Therefore,by studying Bayesian principles,we developed a Bayesian model for calculating the epicentral distance and improved the B-Δ method.Experimental results demonstrate that the application of the Bayesian method significantly improves both the accuracy and stability of the calculated epicentral distance.Building upon this,we combined the smoothing method from the Japanese early warning system to calculate the azimuth and constructed a complete Bayesian single-station localization model,enabling the accurate determination of epicenter coordinates using a single station with good stability in the calculated results.The main contributions of this study are as follows:(1)Estimation of the epicentral distance Δ using the traditional B-Δ method.Simulating data from 949 seismic records in Japan,we preprocessed the seismic signals by applying a 0.075 Hz Butterworth low-pass filter and a mean removal method.We utilized a two-step picking approach to determine the P-wave arrival times and analyzed the magnitude of the slope coefficient B.We compared the B-Δfitting coefficients with previous studies.Experimental results revealed the limitations of the traditional B-Δ method,including unstable results and low accuracy.(2)Development of a Bayesian model for calculating the epicentral distance.We analyzed and verified the influence of different prior information on the model’s accuracy.Using the 949 seismic records from Japan,we conducted simulation experiments and compared the results with those obtained using the traditional B-Δmethod.The experiments demonstrated significant improvements in both the accuracy and stability of estimating the epicentral distance when applying the Bayesian method to enhance the traditional B-Δ estimation.(3)Utilization of the smoothing method from the Japanese early warning system to determine the azimuth,combined with the Bayesian epicentral distance calculation model,resulting in a complete Bayesian-based single-station localization system.This system achieved real-time single-station localization functionality.Through experimental simulations with 43 randomly selected stations during a magnitude Mj5.4 earthquake event in Honshu,Japan,the results demonstrated that the proposed model can achieve single-station localization,providing valuable references for single-station early warning localization,and yielding stable calculated results. |
