VAN EARTHQUAKE PREDICTIONS - AN ATTEMPT AT STATISTICAL EVALUATION - REPLY

A number of successful VAN predictions were omitted in Kagan's [1996] calculations. Furthermore, Kagan assumed an arbitrary cutoff in the ma gnitude of the earthquakes, We show that, for the SI-NOA catalog, the VAN high rate of success is not due to a retroactive adjustment of pre diction rules, or to the non-randomness of seismicity. If we remove de pendent events from the catalog, the prediction effect still remains s tatistically significant when considering the prediction rules suggest ed by VAN since 1986, and not omitting a number of successful predicti ons referring to main shocks. The simple prediction algorithm, suggest ed by Kagan [1996], cannot be considered as giving similar results wit h VAN; this is so because, as we show with simple examples, the number of successes is not a safe criterion to compare two methods. Kagan's [1996] algorithm predicts almost exclusively aftershocks, while VAN pr edicts a comparable (or even larger) number of earthquakes most of whi ch are main shocks. For the PDE(NOAA) catalog, a re-analysis shows tha t the VAN predictions' rate of success cannot be attributed to chance.