Karen R. Felzer, Rachel E. Abercrombie, and Göran Ekström

Annual Meeting, Southern California Earthquake Center, 2002.

Whenever an earthquake happens, there is the potential for large and damaging aftershocks.  Often the assumption is made that these aftershocks are directly caused by stress changes induced by the mainshock, and mainshock stress change calculations are used to forecast aftershock locations.  We find, however, that a majority of aftershocks may actually be triggered by previous aftershocks.  As a result, many aftershocks might best be forecast by an updatable algorithm that uses the times, locations, and magnitudes of previous aftershocks.  Success with approaches of this type has already been demonstrated by previous studies, including Kagan and Jackson (2000) and Wiemer et al. (2002).  We perform a direct comparison of whether static Coulomb stress change produced by the mainshock or previous aftershock parameters can best be used to forecast aftershock locations for four California sequences.  We find that while both methods work comparably at narrowing down the locations where the most spatially concentrated 20% to 40% of the aftershocks will occur, the rest of the locations can be best forecast by using previous aftershock parameters.