Tweet
http://geohab.coas.oregonstate.edu/p...2011135_p2.pdf
Full text at link. 
Goldfinger, Chris, et al. "Superquakes and Supercycles." Seismological Research Letters 84.1 (2013): 1.
Superquakes and Supercycles
by Chris Goldfinger, Yasutaka Ikeda, Robert S. Yeats, and Junjie Ren
INTRODUCTION
The recent 2011 Mw 9.0 Tohoku, Japan, and the 2004
Mw 9.15 Sumatra?Andaman superquakes have humbled many
in earthquake research. Neither region was thought capable of
earthquakes exceeding Mw ∼ 8:4. Appealing proposed relationships
to predict the size of earthquakes in subduction
zones, such as that between earthquake magnitude and parameters
such as lower plate age and convergence rate (Ruff and
Kanamori, 1980) and plate coupling based on anchored slabs
(Scholz and Campos, 1995), at least have many exceptions and
may not be valid. Both earthquakes occurred where the subducting
plate edge was quite old, ∼50?130 Ma. The role of
thick sediments smoothing the plate interface and maximizing
rupture area has been considered a contributing factor, and it
seems to influence many recent great earthquakes (Ruff, 1989).
The Tohoku event is also contrary to this hypothesis. Clearly,
much remains to be learned about these great events, so much
so that most previous estimates of maximum earthquake size in
subduction plate boundaries should be considered suspect, and
perhaps other fault systems as well (McCaffrey, 2007, 2008).
Our perspective on this issue is clearly hampered by short
historical and even shorter instrumental records. The examples
noted earlier indicate that basing estimations of maximum
earthquake size or models of earthquake recurrence on such
short-term records alone clearly cannot encompass the range
of fault behavior, even when historical records may be
>1000 years long as in Japan. Here, we present several examples
of areas where long geologic and paleoseismic records can
illuminate a much wider range of seismic behaviors compared
with those deduced from historical and instrumental data, and
speculate on models of long-term fault behavior based on very
long records.
Superquakes and Supercycles
by Chris Goldfinger, Yasutaka Ikeda, Robert S. Yeats, and Junjie Ren
INTRODUCTION
The recent 2011 Mw 9.0 Tohoku, Japan, and the 2004
Mw 9.15 Sumatra?Andaman superquakes have humbled many
in earthquake research. Neither region was thought capable of
earthquakes exceeding Mw ∼ 8:4. Appealing proposed relationships
to predict the size of earthquakes in subduction
zones, such as that between earthquake magnitude and parameters
such as lower plate age and convergence rate (Ruff and
Kanamori, 1980) and plate coupling based on anchored slabs
(Scholz and Campos, 1995), at least have many exceptions and
may not be valid. Both earthquakes occurred where the subducting
plate edge was quite old, ∼50?130 Ma. The role of
thick sediments smoothing the plate interface and maximizing
rupture area has been considered a contributing factor, and it
seems to influence many recent great earthquakes (Ruff, 1989).
The Tohoku event is also contrary to this hypothesis. Clearly,
much remains to be learned about these great events, so much
so that most previous estimates of maximum earthquake size in
subduction plate boundaries should be considered suspect, and
perhaps other fault systems as well (McCaffrey, 2007, 2008).
Our perspective on this issue is clearly hampered by short
historical and even shorter instrumental records. The examples
noted earlier indicate that basing estimations of maximum
earthquake size or models of earthquake recurrence on such
short-term records alone clearly cannot encompass the range
of fault behavior, even when historical records may be
>1000 years long as in Japan. Here, we present several examples
of areas where long geologic and paleoseismic records can
illuminate a much wider range of seismic behaviors compared
with those deduced from historical and instrumental data, and
speculate on models of long-term fault behavior based on very
long records.