Searching for Tectonic Tremor on the North Anatolian Fault

Tectonic tremor, a relatively newly-discovered seismologic phenomenon, has been helpful in increasing our understanding of fault evolution and slip. Tectonic tremor was first identified in subduction zones, such as Cascadia and Japan. More recently, tremor was discovered on a transform boundary, the San Andreas Fault (SAF) in California. The North Anatolian Fault (NAF) in Turkey is also a transform boundary and has a similar slip rate to the SAF but is younger. Because tremor has not previously been identified on the NAF, our search was designed to identify signals with characteristics similar to those of tremor found in previous studies. In our search for tremor, we analyzed data from the North Anatolian Fault Passive Seismic Experiment that was deployed along the NAF in central Turkey from 2005 to 2008. This experiment included 39 broadband stations, some of which were located within a few kilometers of the NAF or its major splays. In other regions, tremor has been triggered by the surface waves of large earthquakes so we first checked if this was also the case for the NAF. For the 22 earthquakes of M ≥ 5.5 in or around Turkey during the time period of the experiment, we filtered the seismograms between 2-8 Hz, the dominant tremor passband for other regions. We inspected the filtered records but did not see any tremor triggered by surface waves. We then focused on identifying ambient tremor along the NAF using a frequency ratio scanning method. Since different types of seismic signals have peak amplitudes in different frequency bands, time periods with increased tremor activity may be identified by an increase in amplitude in the appropriate frequency band. In this step of the analysis, data from all stations were filtered from 2-5 Hz, 10-15, and 0.02-0.1 Hz to differentiate tremor, local earthquakes, and surface waves, respectively, and the amplitude was computed for 5-minute-long windows for each passband. We more closely investigated time periods with increases in the amplitude of the tremor passband relative to the local earthquake and surface wave passbands. Early December 2007 was one of the times that fit these criteria at multiple stations, but closer inspection of seismograms from this time period showed a diversity of seismic signals. In addition to signals that resembled tremor, we observed many local earthquakes and unidentified cultural or instrumentation signals. The tremor episodes had peak amplitudes from 5-15 Hz, which is higher than in other regions, whereas the local earthquakes had significant energy in the 2-5 Hz band. This suggests that the passbands used in the frequency ratio scanning method, which were based on the characteristics of tremor and earthquakes observed in other areas, need to be revised before the method can be applied in Turkey. Future research should focus on defining the characteristics of tremor on the NAF and also more closely examine the geologic parameters that may be influencing our observations.