Apparent earthquake rupture predictability

Abstract

To what extent can the future evolution of an ongoing earthquake rupture be predicted? This question of fundamental scientific and practical importance has recently been addressed by studies of teleseismic source time functions (STFs) but reaching contrasting conclusions. One study concludes that the initial portion of STFs is the same regardless of magnitude. Another study concludes that the rate at which earthquakes grow increases systematically and strongly with final event magnitudes. Here, we show that the latter reported trend is caused by a selection bias towards events with unusually long durations and by estimates of STF growth made when the STF is already decaying. If these invalid estimates are left out, the trend is no longer present, except during the first few seconds of the smallest events in the data set, M_w 5–6.5, for which the reliability of the STF amplitudes is questionable. Simple synthetic tests show that the observations are consistent with statistically indistinguishable growth of smaller and larger earthquakes. A much weaker trend is apparent among events of comparable duration, but we argue that its significance is not resolvable by the current data. Finally, we propose a nomenclature to facilitate further discussions of earthquake rupture predictability and determinism.