Premonitory changes in seismicity prior to the Great Sumatra-Andaman earthquake of December 26, 2004

We studied changes in seismicity in a source region of the Sumatra earthquake prior to the earthquake on 26 December 2004. Rate increases were detected for earthquakes of M ≥ 5.0 using the ETAS method, with the assumption that independent events follow the Poisson process with at most only one rate change. Based on the declustered catalogue, we obtained consistent changes with simple Poisson models. Temporal changes in mean magnitude were obtained using the CUSUM method. b -values for three periods were calculated, taking into consideration the results of the CUSUM analysis, which showed a 10-year period of an anomaly high b -value followed by a 13-year period of a low b -value. The statistical significance of this change was confirmed by a bootstrap method. Both changes imply activation of seismicity in the source region.