Revisiting historical seismograms of the 1923 M7.9 Kanto earthquake and aftershocks

Historical analog records of earthquakes occurred in the past 100-140 years are archived at various universities and institutes throughout Japan. In 2023, the 1923 M7.9 Kanto earthquake will mark the 100th anniversary. This 1923 Kanto earthquake triggered abundant aftershocks. However detailed waveform analyses to estimate fault parameters for aftershocks are not yet performed. Such examples include five M~7 earthquakes occurred within the 2 days following the mainshock. Understanding the slip models for these aftershocks will provide the new insight into the stress interaction between the mainshock and aftershocks, and also could provide additional contribution to refine the current earthquake hazard model. We revisit historical seismograms to determine fault models for large aftershocks (M > 6.5) including the 1923 Kanto mainshock based on modern waveform analysis methods with high-performance computing and current knowledge on the velocity structure. Our effort has been focusing on collecting seismic records for the 1923 Kanto earthquake sequence, and we find that large aftershocks were recorded by a variety of seismographs installed globally. At the Univ. of Tokyo, Japan, Imamura's seismographs (V=1 and V=2), Omori's long period seismographs (T=60s, V=15-20) and other seismographs were operational to register events signals from this earthquake sequence. The Earthquake Research Institute, the Univ. of Tokyo has archived the copies of the seismograms recorded at overseas stations including Victoria, Uccle, Uppsala, Riverview, Strasbourg, and De Bilt. However, it should be noted that the large shaking of the mainshock masked the seismic phases from two M~7 aftershocks. The Berkeley Digital Seismic Network also operated horizontal Bosch-Omori and vertical Wiechert seismographs for the 1923 Kanto earthquake sequence and continuous analog records are archived. These continuous data provide the detection threshold of seismic phases from aftershocks. We find that the magnification of the Bosch-Omori type (V=50) seems to be not enough to capture event signals for M < 7 aftershocks. We are proceeding with digitization of those records to systematically identify seismic phases for this earthquake sequence. We also plan to establish a database of digitized records that will be freely available to public.