STOCHASTIC MODELS FOR EARTHQUAKE OCCURRENCE

Seismic data for the period 1446—1969 in the North China seismic active region have been used in this study. We selected 36 earthquakes which occurred at the points of inflection and at the main peaks of the active periods. Hermitian interpolation was made between the energies of two consecutive earthquakes. In this way, a curve of equal sampling intervals showing the variation of the activity of the North China seismic region during the past 500 years was obtained.This curve may be assumed to consist of two parts, namely, the deterministic part and the stochastic part. A periodic model and an auto-regressional model of stationary stochastic process were respectively proposed for carrying out extrapolation and prediction.By posteriori-test analysis, we mean the interceptions of the recent data 15 times, each followed by computing, extrapolating and predicting separately. Again, by summing up the number of time intervals predicted to be earthquake dangerous, a curve of prediction could be obtained.Based on these results of posteriori-test we predicted in April 1974 the appearance during the time interval from 1975 to 1977 a seismic active period in North China, within which the Haicheng earthquake of 1975 and the Tangshan earthquake of 1976 actually occurred. Moreover, the said curve is still significant for predicting future earthquakes in North China. It seems that the seismic active period of North China starting from the Haicheng earthquake would last for a certain length of time, till about 1980, but it would be again active at around the year 2001.