Intraplate seismic hazard: Evidence for distributed strain and implications for seismic hazard - Intraplate Earthquakes - page 319

Table 12.1 Plate rate versus propensity for
large event clustering for four tectonic plates

Plate rate
(cm/yr)

Short-term
clustering? Tectonic plate

Australian 5.8 Yes

Indian 2.6 No?

North American 4.1 Yes

Eurasian 2.3 No

low viscosity and, therefore, to a shorter recurrence interval for individual source zones.
This hypothesis is highly speculative; however, both Australia and North America are
characterized by relatively fast plate rates and clustering, whereas peninsular India and
Eurasia are characterized by significantly lower plate rates and no clustering.
In any case, in all of the intraplate regions discussed above, evidence points to earthquake
activity that in some SCR regions tends to be clustered on timescales of centuries to a few
millennia, and that in all SCR regions is distributed throughout a broad region over longer
timescales.

12.4 Statistical considerations

In the absence of geological and geodetic constraints on fault slip rates or direct evidence
for large characteristic earthquakes, PSHA calculations have long relied on the observed
rate of small-to-moderate earthquakes in a region to estimate the expected rate of large
earthquakes. This practice is motivated by good evidence that, through any given region,
seismicity is characterized by a Gutenberg-Richter magnitude distribution (Gutenberg and
Richter, 1944 ) with a b -value of 1.0 (e.g., King, 1983 ; Felzer, 2006 ) . The practice is further
justified by appealing to the assumption of stationarity; the same rationale is sometimes also
used to estimate M max . However, it has been shown that a stationary process, for example
the Epistemic Triggering of Aftershock (ETAS) model (Kagan and Knopoff, 1981 ; Ogata,
1988 ) , will give rise to a constant b -value but an apparent a -value that can vary considerably
if estimated from a relatively short earthquake catalog (Page et al ., 2010 ) . As shown by
Stein and Newman ( 2004 ) , within a short catalog large earthquakes can appear to be either
characteristic or “uncharacteristic”; i.e., to not be represented in the catalog at the average
long-term rate.
The ETAS model, which provides an integrated framework for seismicity including
foreshock and aftershock probabilities, has been shown to match the degree of clustering
observed in typical earthquake catalogs (e.g., Hardebeck et al ., 2008 ) . An ETAS model
predicts not only traditional foreshock/aftershock clustering, but also regional clustering of
moderate earthquakes. In California, for example, such clustering can give rise to periods of