Determination of earthquake fault parameters from long-periodPwaves

The theory for propagating earthquake sources predicts a ‘Doppler-type’ effect in the time duration of the radiated body wave pulses. With the assumption of a constant fracture velocity, the time duration of these pulses is directly related to the length of the fault but has different spatial patterns of variation for the unilateral and bilateral faulting cases. To relate this time duration of energy release to the observed long-period P waves, various pulse forms are postulated whose time history is controlled by the fault parameters of length and fracture velocity. These pulses are convolved with the impulse response of a standard crust-seismograph system to yield theoretical P-wave seismograms. A relation is thereby established between the time durations of the radiated pulses and the time durations observed at the surface of the earth. This relationship is employed to correct the times of observed long-period P waves so as to recover the original time-duration radiation pattern. By a minimum deviation technique, the fault length and fracture velocity can be found that best fits this observed pattern. This procedure is shown to yield reasonable results for four different earthquakes from three different regions.