Examination of scaling between proposed early signals in P waveforms and earthquake magnitudes

Abstract

We examine seismic waveforms generated by clusters of similar earthquakes and additional events located on the Karadere-Duzce branch of the North Anatolian fault to investigate possible scaling between proposed early signals in the P waveforms and magnitudes. The employed signals can be divided into two classes. The first is associated with Ellsworth-Beroza and Iio type signals that are assumed to reflect signatures of seismic nucleation phases. The second are associated with Nakamura type period and Wu-Zhao type maximum displacement in the early waveforms that are related to measures of the corner frequency and local magnitude, respectively. The impact on each signal of artefacts generated by the Finite Impulse Response (FIR) filter is also examined. The use of repeating event clusters reduces considerably the variability of path and site effects compared to previous regional and global studies. This allows us to focus more accurately on variations of the potential scaling parameters with the source magnitude. However, the magnitude range of the employed events is limited to about 4 units. The FIR filter artefacts can generate spurious signals prior to the P -wave arrival, potentially affecting the Ellsworth-Beroza type signals, but are not a significant influence on any of the other measurements. The results indicate that candidates for the Ellsworth-Beroza signals, associated with identification of a weak arrival before the main P phase, exist in less than 20 per cent of the cases. The Iio type signals can be determined by definition on all waveforms. Both of these potential signatures of nucleation phases show little or no scaling with the final event size in the examined data. In contrast, the Nakamura and Wu-Zhao type signals in the first few seconds, which can also be determined by definition on all waveforms, appear to scale with the final event magnitude, albeit with large scatter. The relatively large time windows used in the analysis, compared to the rupture time of the employed events, implies that the results provide little if any information on the physical process of the rupture initiation. The scaling of the Nakamura and Wu-Zhao type early signals with the final event size may be explained by statistical tendency of stronger initial rupture phases to propagate larger distances. © 2007 The Authors Journal compilation © 2007 RAS.