5:15 PM - 6:30 PM
[SCG39-P14] A trial digitization of analog seismograms of the Kanto-Tokai Observation Network for the analysis of tectonic tremor
Keywords:Slow Earthquakes, Analog Seismogram, Deep Tectonic Tremor
Slow earthquakes have been studied by many researchers since the discovery of tectonic tremor (hereinafter, tremor) in the Nankai region (Obara, 2002, Science). However, the past activity of tremor in 20th century is still not clear. Continuous digital seismic data in this period is rare due to the cost of data storage. Thus, analog seismograms are essential dataset to reveal the tremor activity in such period. In addition, digitization of analog data is important for the waveform analysis of tremor. In terms of the analysis of tremor activity, manual digitization requires much time, as tremor episodes lasts several days, typically. Therefore, automatic digitization seems to be suitable. National Research Institute for Earth Science and Disaster Resilience has stored the paper recordings of seismograms of the Kanto-Tokai Observation Network, which started since late 1970s. In our previous study, we have reported the tremor activity in the Tokai region in 1980s, based on the visual inspection of analog seismograms (Matsuzawa and Takeda, 2020, JpGU). We aim to digitize these analog seismograms for the analysis of tremor, and tried to develop an automatic digitization tool of analog seismograms.
The analog seismograms are recorded by ink recorders. Two hours data are divided to four blocks in a sheet. Traces of seismograms in the left and right blocks are 0-35s, 30-65s in each minute, respectively. Upper and lower blocks correspond to the former and the latter hour, respectively. In each trace, time marks for each second are added as streaks. Short offset is added in each minute. We scanned these analog recordings and saved to files in the Portable Network Graphics (PNG) format (400dpi, grayscale).
We tried to develop a tool to digitize these data in the following procedure. At first, we converted the PNG format file to the PGM (Portable Gray Map) format, and handled bitmap data in our program. In our program, we locate four trace blocks, then set the analysis area for each trace, based on the sum of pixel values in the horizontal direction in each trace block. The horizontal, and vertical width of analysis area is set to the block length, and interval of baselines of traces, respectively. We calculated weighted sum in the vertical direction based on the distance from the baseline, and set to the value as the waveform amplitude.
To validate our developed program, we applied this tool to image data which simulate analog seismograms of the Kanto-Tokai Observation Network, using Hi-net digital data. The recovered digital data from images are generally similar to the original data in the case of small amplitude signal, although high frequency components are lacked. We compared power spectra between original and recovered data. The amplitude is similar below 10Hz, while peaks in 1Hz and overtones are found in the recovered data due to time marks for every second. This result supports that the analog seismograms and our tool can extract tremor signal, as the tremor signal is dominant at several Hz.
Our digitization method assumes that amplitude of waveforms is smaller than the vertical width of analysis area (i.e. the interval of traces). This means that larger signal than this width cannot be recovered, due to the saturation in the trace and contamination to the adjacent traces. Active tremor signals sometimes exceed this amplitude, while most tremor has small amplitude. Another problem is related to the time marks. Envelope correlation technique or matched filter techniques are frequently used in locating tremor. As these are based on correlation functions, we need to develop a method to avoid the apparent effect of time marks.
The analog seismograms are recorded by ink recorders. Two hours data are divided to four blocks in a sheet. Traces of seismograms in the left and right blocks are 0-35s, 30-65s in each minute, respectively. Upper and lower blocks correspond to the former and the latter hour, respectively. In each trace, time marks for each second are added as streaks. Short offset is added in each minute. We scanned these analog recordings and saved to files in the Portable Network Graphics (PNG) format (400dpi, grayscale).
We tried to develop a tool to digitize these data in the following procedure. At first, we converted the PNG format file to the PGM (Portable Gray Map) format, and handled bitmap data in our program. In our program, we locate four trace blocks, then set the analysis area for each trace, based on the sum of pixel values in the horizontal direction in each trace block. The horizontal, and vertical width of analysis area is set to the block length, and interval of baselines of traces, respectively. We calculated weighted sum in the vertical direction based on the distance from the baseline, and set to the value as the waveform amplitude.
To validate our developed program, we applied this tool to image data which simulate analog seismograms of the Kanto-Tokai Observation Network, using Hi-net digital data. The recovered digital data from images are generally similar to the original data in the case of small amplitude signal, although high frequency components are lacked. We compared power spectra between original and recovered data. The amplitude is similar below 10Hz, while peaks in 1Hz and overtones are found in the recovered data due to time marks for every second. This result supports that the analog seismograms and our tool can extract tremor signal, as the tremor signal is dominant at several Hz.
Our digitization method assumes that amplitude of waveforms is smaller than the vertical width of analysis area (i.e. the interval of traces). This means that larger signal than this width cannot be recovered, due to the saturation in the trace and contamination to the adjacent traces. Active tremor signals sometimes exceed this amplitude, while most tremor has small amplitude. Another problem is related to the time marks. Envelope correlation technique or matched filter techniques are frequently used in locating tremor. As these are based on correlation functions, we need to develop a method to avoid the apparent effect of time marks.