A relationship between back-azimuth estimation accuracy for earthquake early warning and seismometers site environment

When earthquakes which threaten the safety of railway facilities and running trains occur, earthquake early warning (EEW) systems aim to stop the trains as soon as possible. In order to quickly issue an alarm during earthquake if necessary, the seismometers installed for these systems automatically detect the P-wave arrival and estimate seismic parameters such as epicentral distance, back-azimuth and magnitude using information from the initial few seconds of the P-wave. To improve the estimation accuracy, we examined the influence of the seismometers site environment. In this research we focused on the back-azimuth estimation accuracy particularly.

To quantify the seismometers site environment, we adopted indices as noise RMS, AVS30 and seismic basement depth. The noise RMS, which presents the quietness at the seismic observation sites, is defined from 2 seconds of the waveform recorded before P-wave arrival. The AVS30, which presents the hardness of subsurface ground at the site, is defined the averaged S-wave velocity from the depth of 30m to the ground surface. And the seismic basement depth, which presents the thickness of the sedimentary layer around the site, is defined the depth from the layer of S-wave velocity 400m/s as an engineering bedrock to that of S-wave velocity 3100m/s as a seismic bedrock.

We examined the relationship between the back-azimuth estimation accuracy and these indices, and found that the estimation error tends to decrease as the noise RMS decreases, the AVS30 increases and the seismic basement depth decreases. As the reasons for these, the accuracy of the double integration from acceleration waveform to displacement waveform to estimate the back-azimuth is higher in the quiet site, and the horizontal motion of the initial P-wave becomes clearer in the site where the subsurface ground is harder and the sedimentary layer is shallower.

According to the results, the back-azimuth estimation accuracy will be improved in a quiet environment where the ambient noise is small, the subsurface ground is hard and the depth of the sedimentary layer is shallow. When we select the seismometers installation site for EEW, we should give priority to site environments with these characteristics.