the relevance of High - frequency analysis artifacts to remote triggering

The high-frequency content observed in teleseisms recorded by seismometers can be produced either by the nonlinear behavior of seismometers and digitizers (Delorey et al. 2008; Hellweg et al. 2008) or by the real Earth response. The latter include scattering from small-scale heterogeneities during seismic wave propagation (e.g., Chen and Long 2000) and high-frequency radiations from the earthquake source (e.g., Peng et al. 2006) or near-surface regions near the recording site (Fischer et al. 2008). Recent studies have shown that large-amplitude surface waves generated by earthquakes at regional and teleseismic distances could also trigger high-frequency seismic sources, either in the form of regular earthquakes at seismogenic depth near the recording site (Hill and Prejean 2007 and references therein) or “non-volcanic” tremor in the lower crust (Rubinstein et al. 2010; Peng and Gomberg 2010 and references therein). In teleseisms the presence of high-frequency content (e.g., >5 Hz) in the seismogram is inconsistent with the expected attenuation of waves from a distant source (i.e., >1,000 km). The lack of frequencies above 5 Hz in a teleseism makes it easy to separate the seismic signals of locally triggered events from those of the teleseism by applying a high-pass or bandpass filter to broadband continuous recordings (e.g., Hill and Prejean 2007; Velasco et al. 2008). Another effective way to demonstrate triggered seismicity is the spectrogram display (i.e., frequency-time plot) of the seismic data (e.g., West et al. 2005; Hill and Prejean 2007; Peng and Chao 2008; Peng et al. 2008). In such a plot, the locally triggered seismic signals typically show as narrow vertical bands rich in high-frequency energy within the low-frequency body and/or surface waves of teleseismic events. When examining high-frequency signals for evidence of remote triggering, it is important to distinguish between genuine high-frequency signals from triggered events and those from seismic instruments or analysis procedures. For example, Hellweg et al. (2008) found that due to digitization errors, large long-period surface waves of the 2002 Mw 7.8 Denali fault earthquake recorded at the STS-1 broadband sensors in northern California produced high-frequency noises that mimic the pattern of remotely triggered tremor and earthquakes. In this article we show that signal processing artifacts could also introduce high-frequency energy in the spectrogram plot that mimics remote triggering of earthquakes and/or tremor. In the following section, we first describe the general observation, followed by a detailed explanation. Next, we offer several procedures to correct for such artifacts, and discuss our results in the context of previous observations.