PREDICTION OF A SECOND LARGE EARTHQUAKE BASED ON AFTERSHOCK SEQUENCE

Abstract

The recognition problem is addressed in detection of the aftershock sequences of large earthquakes where the occurrence of large events is likely. Seismological data were used to select several functionals and to develop a rule that correctly classified 16 of 17 aftershock sequences in California. The algorithm was tested using independent data without parameter fitting, namely, aftershock sequences in eight seismic regions of the world. The testing results are satisfactory, since the errors do not exceed 15\%. The result is statistically significant at the 96\% level. Two successful forward predictions have been made.

Back to
Computational Seismology, Vol. 2.