Intraplate seismic hazard: Evidence for distributed strain and implications for seismic hazard - Intraplate Earthquakes - page 304

as demonstrated, in addition to other evidence discussed in this paper, by abun-
dant and widely accepted evidence that the New Madrid seismic zone itself
“turned on” in the relatively recent geological past. One logic-tree approach to
PSHA might therefore be to include a low probability that recent clusters have
ended along with a non-zero probability that future activity within other rifted
zones will exceed the rate predicted by an extrapolation of current background
rates.
12.1 Introduction
Quantifying probabilistic seismic hazard remains a vexing problem in intraplate regions
throughout the world. Along active plate boundary regions, expected long-term earthquake
rates can be estimated from seismicity catalogs, geological evidence of past earthquakes,
and GPS estimates of ongoing deformation. Hazard assessment remains challenging, and
under active investigation, even in active regions (e.g., WGCEP, 2013 ) ; however, in a well-
studied region such as Japan or California, geological results and deformation data provide
at least a first-order constraint on seismic moment accrual and release rates. The magnitude
of the maximum possible earthquake in a region, M max , can also be estimated given an
overall strain rate, paleoseismic constraints on the rate of surface-rupturing earthquakes,
and the commonly accepted assumption that, in most regions if not for specific faults,
regional earthquake magnitudes follow a Gutenberg-Richter distribution (Gutenberg and
Richter, 1944 ; King, 1983 ; Felzer, 2006 ; Hough, 1996 ) .
In intraplate regions, estimation of long-term earthquake rates is far more challeng-
ing. A fundamental difficulty for hazard assessment is that, unlike active plate boundary
zones, no basic, integrated physical framework can explain the first-order processes con-
trolling seismogenesis in intraplate regions. Given the lower rate of earthquakes, even
relatively long historical catalogs are inadequate to capture long-term average rates; direct
geological evidence of past earthquakes is generally scant, and hopes of constraining low
strain rates from GPS data have not generally been borne out to date, with improved
data often revealing little to no detectable surface deformation (e.g., Calais et al ., 2005 ;
Frankel et al ., 2012 ) . With no long-term constraint on fault slip rate or deformation from
geological or geodetic data, estimates of earthquake rates and thus hazard are estimated
from seismicity data; i.e., the earthquakes observed during historic times and/or doc-
umented in prehistoric times, and their estimated magnitudes. Even where the largest
known earthquakes - the M max , or characteristic earthquakes - occurred during pre-
instrumental times, magnitude estimates can be uncertain by a full magnitude unit or
more (see Hough and Page, 2011 ) . Recurrence times for an assumed characteristic event
can sometimes be estimated from geological evidence; however, as I discuss in a later
section, even these long-term recurrence rate estimates are highly uncertain in intraplate
regions.
In this chapter I focus initially on our current understanding of seismic hazards in the
Central/Eastern United States (CEUS), where (1) large earthquakes have struck in historical