Intraplate seismic hazard: Evidence for distributed strain and implications for seismic hazard - Intraplate Earthquakes - page 314

is roughly consistent with the Holocene paleoseismic record if characteristic earthquakes
are low magnitude 7.
Any model derived from surface measurements of strain is non-unique, with gener-
ally poor resolution at depth (e.g., Page et al ., 2009 ) . The assumed rupture length for a
Reelfoot fault event is moreover uncertain. To estimate a magnitude for a Reelfoot fault
event, Frankel et al .( 2012 ) assume a rupture length of 60 km, with rupture extending over
a full 24 km down-dip width. This rupture length is based primarily on the distribution of
present-day microseismicity. Based on analog and box modeling, Pratt ( 2012 ) concludes
that thrust faulting does extend over the full length of the central NMSZ limb as defined by
microseismicity (see Figure 12.3 . ) However, several other lines of evidence suggest pre-
dominant moment release associated with NM3 involved a shorter rupture length, bounded
between the intersections of the Reelfoot and the northern and southern limbs of the NMSZ
( Figure 12.3 ) : (1) an extension of thrust motion south/southeast of the intersection of the
Cottonwood Grove and Reelfoot faults is kinematically inconsistent, as right-lateral strike-
slip movement on the Cottonwood Grove fault would lower the likelihood of failure of the
thrust fault extending south/southeast of the junction with the Reelfoot fault (e.g., Mueller
et al ., 2004 ) ; (2) the scarp/surface flexure associated with the Reelfoot fault is less well
expressed geomorphically to the south/southeast of the intersection (e.g., Champion et al .,
2001 ) ; (3) Mueller et al .( 2004 ) show that the “side limbs” on the NMSZ are consistent with
off-fault lobes of increased Coulomb stress generated by ruptures on two master faults, the
Cottonwood Grove and Reelfoot faults. Further, it is possible that the southeastern exten-
sion of the Reelfoot fault ruptured not during the February mainshock but rather during
the dawn aftershock (NM1-A), as proposed by Hough and Martin ( 2002 ) . Thus, while a
plausible case can be made for a longer rupture length, a length of 40 km is also defen-
sible. If one assumes a rupture length of 40 km and a depth of 20 km, the moment of a
“500-year event” inferred by Frankel et al .( 2012 ) would be reduced by nearly a factor
of 2 (i.e., the ratio of fault area: 800 km 2 /1440 km 2 ), implying a M w of 7.1 rather than
7.3. It has sometimes been suggested that the magnitudes of the 1811-1812 earthquakes
could be higher than predicted from standard scaling relationships by virtue of high stress
drop and/or unusually high shear modulus (e.g., Johnston, 1996 ) . However, as discussed by
Hanks and Johnston ( 1992 ) , there is a strong dependence of high-frequency strong ground
motions on stress drop. It is thus not possible to appeal to high stress drop as an explana-
tion for high magnitudes, if the magnitudes are determined from high-frequency ground
motions (i.e., intensities). In fact Hanks and Johnston ( 1992 ) themselves conclude that both
the 1811-1812 mainshocks and the 1886 Charleston earthquake might be “no larger than
M

6.5 to 7, provided their stress drops are higher than average by a factor of 2 or so.”
(Note that this estimate was based on the original intensity assignments for the 1811-1812
earthquakes.)
The strain-rate observations from the NMSZ are thus consistent with two interpretations:
(1) As argued by Hough and Page ( 2011 ) , and corroborated by Frankel et al .( 2012 ) , a
localized low but non-zero surface strain rate (e.g., 10 − 9 /yr), within the bounds imposed by
the most recent GPS studies, could be sufficient to account for a sequence with an equivalent

=