Historical Seismograms - Evidence from the AD 2000 Izu Islands ...

60 N. V. Kondorskaya and Yu. F. KopnichevTo use the proposed method, it is necessary to carry out spectral-temporal analysisof P-wave records on short period instruments. This involves digital samplingof the records at intervals not exceeding 0.2 sec and follow-up fast Fourier transform(FFT) of the time series; the amplitude Fourier spectrum is multiplied by thecharacteristic of the respective TCHISS filters (bandpassed between 1.0 to 1.6 Hzat the 0.7 level) (Zapolsky, 1971). Then the reverse Fourier transform is made.As a result, we get a record equivalent to the record of the TCHISS channel. Theenvelope of the record is plotted and compared to average standard curves to revealthe rise time and also to determine the total duration of oscillations ( r ) throughcomparison of the codas of a weak earthquake from the same focal region with theenvelope of the strong earthquake under study. Studies of specific characteristics ofshort-period radiation enabled us to identify a strong relationship between the amplitudeof the velocity increment in the first P-wave arrivals at teleseismic distancesand source mechanism (Figure 2).For shallow earthquakes (h < 100 km) amplitudes build up slower for normalfaults, quicker for strike-slip faults and overthrusts, and very quick for reverse faults.This relationship is not valid for deep earthquakes.Thus, we may trace the distribution of oceanic earthquakes by two parameters:by the inclination of the P-wave envelope at the beginning of amplitude growth andby the time r, of reaching the maximum amplitude in the train of P-waves. Thefirst parameter provides information on the mechanism and depth of the earthquakesource. The second parameter provides information on the duration of one or severalgreat pulses making up short-period radiation (for most "single pulse" sourcescontributing to strong earthquakes, the time r, practically coincides with ro - theduration of the propagation process in the source). Figure 3 shows the envelopes ofsome oceanic earthquakes (they all gave rise to intensive tsunami waves). It is seenthat the envelope in stretches of amplitude growth are close to average relationshipsobtained through analysis of combined continental and oceanic earthquakes withvarious displacements in the source. The distribution of rrn for tsunamigenic andnon-tsunamigenic earthquakes is given in Figure 4. A sharp distinction betweentsunamigenic and non-tsunamigenic events is seen and threshold values of rm differfor earthquakes with different source mechanisms.It is necessary to emphasize that the data from one station are quite enough toprovide information on the tsunamigenic capacity of earthquakes and the displacementsin the source, since the above parameters are relatively stable according todata of various stations. In this context we undertook a preliminary interpretationof the records of a strong Japan earthquake of December 20, 1946 made atthe Tucson station with short-period Benioff instruments (Figure 5). Plotting ofthe envelope and estimation of rm equal to 139 sec indicate that this earthquakewas tsunamigenic and its source mechanism was close to an overthrust, which is inaccord with other determinations (Abe, 1980).Very recently we received a record of this earthquake, made with a short-perioddevice at the Pasadena station, which agrees well with the results of the Tucsonst at ion.Short-period records of historical earthquakes may be used in the evaluation ofsource parameters employing the method suggested by Koyama and Zhang (1985),where the smoothed envelope of a wave packet is plotted and the short-periodseismic moment is calculated.