Geology Reference
In-Depth Information
0.2-1.0 km of the coastline. The uppermost soil
contains rooted snags of dead, but still-standing,
spruce trees (ghost forests), even though the soil
is presently 1 m below the modern tide-marsh
surface. Above the soil, a layer of fine-grained
sand can be traced inland, where it thins and
becomes finer-grained along the banks of a
coastal river. Lying directly above the sand, mud
contains roots of pioneer species from saltwater
tideflats. Soils of the buried spruce swamps can
be traced seaward into buried saltwater marsh
soils, some of which contain rooted stems and
leaves of a tufted grass. Tufts project up out of
the soil and are encased in a sandy layer that is
analogous to that found above the spruce
swamp soil. Preservation of the tufted grasses
indicates that they were buried in less time
than that needed for subaerial decomposition
(
<
2 years). Similarly, study of tree rings from the
snags suggests that they died suddenly, rather
than being gradually submerged and killed
(Atwater and Yamaguchi, 1991; Yamaguchi
et al.
,
1997). Given the elevation of modern coastal
spruce forests above sea level, the presence of
the tideflat species above the soil that bears the
rooted trees dictates that these forests were
submerged by at least 0.5 m. Detailed inves-
tigation of ostracod assemblages, which are
zoned vertically within the tidal marsh,
corroborate this estimate of vertical displacement
(Hemphill-Haley, 1995). The overlying sand
layer, which becomes finer-grained and thinner
inland and which entombs grasses in growth
position, is interpreted to result from a tsunami
that swept across the coastal region. (Note that,
along rivers, the normal trend is for downstream
fining (Paola
et al.
, 1992), whereas tsunami
deposits become coarser toward the coast.) The
two underlying buried soils are similarly
associated with roots, stems, overlying sands,
and tideflat muds. All of these observations
are consistent with repeated subduction-zone
earthquakes that caused instantaneous coastal
submergence followed by gradual emergence
(Atwater
et al.
, 1995).
As a result of these stratigraphic studies,
which documented the presence of large
Cascadian earthquakes, the key remaining
questions changed from whether they occurred
in the past, to “When did they occur?” and “Did
the entire length of the subduction zone rupture
(a single night of horror) or were there several
individual earthquakes (a decade of terror)?”
High-precision radiocarbon dates (Atwater
et al.
, 1991) and studies of the tree rings of the
spruce snags (Yamaguchi
et al.
, 1997) suggested
that the most recent earthquake occurred about
300 years ago, probably in the winter according
to the tree rings and local oral histories. But
the resolution of the dating was insufficient
to tell whether the earthquakes recorded from
southern Oregon to Washington represented
one or several closely spaced events. A solution
to this uncertainty came from across the Pacific,
where historical records of tsunamis in Japan
indicate that a large tsunami with no known
local source inundated the east coast in January
1700 (Satake
et al.
, 1996). The timing of the
tsunami and the pattern of coastal inundation
in Japan suggest the Cascadian subduction
zone as a likely source. In addition, numerical
models have been developed to describe how
tsunami waves dissipate as a function of
original size and distance while traversing the
Pacific. By comparing the model results for
different lengths of Cascadian ruptures with
different sizes of waves and run-up on
the Japanese coast, it appears quite possible
that nearly the entire locked portion of the
Cascadian subduction zone ruptured 300 years
ago. A similar rupture today could produce a
magnitude 9 earthquake perhaps analogous to
the 2004 Sumatra earthquake.
Although the coastal stratigraphic record
clearly records as many as three paleoearthquakes,
a longer time series is needed to examine
recurrence intervals for the Cascadian subduction
zone. The violent shaking from large earthquakes
also triggers large-scale failures on the edge of
the continental shelf. Such failures generate
turbidity flows that rush down submarine
canyons and produce recognizable turbidite
beds as they come to rest (Adams, 1990). Over
the past 15 years, extensive coring of numerous
channels that drain the continental shelf along
the Cascadia margin has revealed an extended
sequence of Holocene turbidites (Goldfinger
et al.
, 2003). Radiocarbon dating, identification
