Slow Earthquakes Dating Back to the Era of Recording Paper -A Trial Detection of Deep Low Frequency Tremor in 1980s for a Long-Term Catalog in the Tokai Region, Japan-

Digital continuous data from dense seismic and geodetic networks opened the era of slow earthquakes (i.e., low frequency tremor (LFT), very low frequency earthquakes and slow slip events) which lasts since the pioneering works by Dragert et al. (2001) and Obara (2002). However, the number of catalogs of slow earthquakes in the 20th century is limited, as these events were not well known before 1990s and continuous seismic data is rare due to the cost and performance of data storage device. Thus, analog seismograms are solely available data to reveal the LFT activity in such period. In this study, we attempt to detect deep LFT in the Tokai region, Japan, using analog seismograms on ink recording paper in 1980s.

We visually examined waveforms at Shimoyama (SMY) seismograph in the Kanto-Tokai Observation Network by NIED, to detect episodes of LFT. At first, we generated image files of simulated analog seismograms with the same scale in time and amplitude, to test the detectability of LFT in 2006-2012, as digital continuous data is available at the same station. We used LFT episodes detected by the method in Maeda and Obara (2009) and Obara et al. (2010) as a reference catalog. Even if the single station is used, the waveforms are easily identified as the tremor signal with the duration of several minutes which is repeatedly found for several days. When the number of low frequency signal is small within several hours, we discard the event, as it is not clear whether the signal is caused by a regular earthquake or a small LFT episode. After visual inspection of the waveforms from Apr. 1980 to Dec. 1987, we found possible LFT episodes, for example, in Oct.-Nov., 1980, Jun. 1981, May 1983, Aug. 1984, Apr. 1986, Dec. 1986, and May 1987.

Numerical studies predict the long-term change of LFT behavior, for example, in terms of the recurrence interval (Matsuzawa et al., 2010), and amplitude (Ariyoshi et al., 2009). Such information would be beneficial to reveal the stress accumulation process of megathrust earthquakes. However, as the above LFT is not located, the obtained result is not directly applicable to the discussion. In the next step, we need to digitize analog seismograms for further analysis. Our study shows the importance of saving analog seismograms. Past data could be essential in the centennial time scale, even if it seems to be not useful to date.