Bulletin
of
the
Seismological
Society
of
America,
Vol.
76,
No.3,
pp.
675-708,
June
1986
SOURCE
CHARACTERISTICS
OF
HYPOTHETICAL
SUBDUCTION
EARTHQUAKES
IN
THE
NORTHWESTERN
UNITED
STATES
BY
THOMAS
H.
HEATON
AND
STEPHEN
H.
HARTZELL
ABSTRACT
Historic
earthquake
sequences
on
subduction
zones
that
are
similar
to
the
Cascadia
subduction
zone
are
used
to
hypothesize
the
nature
of
shallow
sub-
duction
earthquakes
that
might
occur
in
the
northwestern
United
States.
Based
on
systematic
comparisons
of
several
physical
characteristics,
including
phy-
siography
and
seismicity,
subduction
zones
that
are
deemed
most
similar
to
the
Cascadia
subduction
zone
are
those
in
southern
Chile,
southwestern
Japan,
and
Colombia.
These
zones
have
all
experienced
very
large
earthquake
sequences,
and
if the
Cascadia
subduction
zone
is
also
capable
of
storing
elastic
strain
energy
along
its
greater
than
1000
km
length,
then
earthquakes
of
very
large
size
(Mw
>
8~)
must
be
considered.
Circumstantial
evidence
is
presented
that
suggests
(but
does
not
prove)
that
large
subduction
earthquakes
along
the
Cascadia
subduction
zone
may
have
an
average
repeat
time
of
400
to
500
yr.
INTRODUCTION
This
is the
third
in
a series
of
four
papers
that
lead
to
an
estimation
ofthe
seismic
hazard
associated
with
the
subduction
of
the
Juan
de
Fuca
and
Gorda
plates
beneath
North
America.
In
the
first
paper,
Heaton
and
Kanamori
(1984)
compared
physical
characteristics
of
the
Cascadia
subduction
zone
(also
referred
to
as
the
Juan
de
Fuca
subduction
zone)
with
those
of
other
subduction
zones
and
concluded
that
the
Cascadia
subduction
zone
is similar
to other
subduction
zones
with
strong
seismic
coupling,
and
thus
may
be
capable
of
producing
great
shallow
subduction
earth-
quakes.
In
the
second
paper,
Hartzell
and
Heaton
(1985)
compared
the
nature
of
the
time
history
of
seismic
energy
release
for
60
of
the
largest
subduction
earth-
quakes
to
occur
in
the
past
50
yr.
In
this
paper,
we
extend
the
results
of
the
previous
studies
to
estimate
the
nature
of
shallow
subduction
earthquakes
that
could
be
postulated
if
the
Cascadia
subduction
zone
is
assumed
to
have
strong
seismic
coupling.
In
the
fourth
paper,
Heaton
and
Hartzell
(1986)
estimate
the
nature
of
strong
ground
motions
that
may
result
from
earthquakes
hypothesized
in
this
study.
Key
issues
addressed
here
are:
(1)
the
dimensions
and
geometry
of
hypothetical
Cascadia
subduction
earthquakes;
(2)
plausible
repeat
time;
(3)
identification
of
analogous
historic
earthquakes;
and
(
4)
estimation
of
possible
tsunami
amplitudes.
Subduction
of
the
Juan
de
Fuca
and
Gorda
plates
has
presented
earth
scientists
with
a dilemma.
Despite
compelling
evidence
of
active
plate
convergence
(Riddi-
hough,
1977;
Snavely
et
al.,
1980;
Hyndman
and
Wiechert,
1983;
Adams,
1984a;
Nishimura
et
al.,
1985),
subduction
on
the
Cascadia
zone
has
often
been
viewed
as
a relatively
benign
tectonic
process
(Riddihough,
1978;
Ando
and
Balazs,
1979;
Acharya,
1981,
1985).
There
is
no
deep
oceanic
trench
off
the
coast;
there
is
no
extensive
Benioff-
Wadati
seismicity
zone;
and
most
puzzling
of
all,
there
have
not
been
any
historic
low-angle
thrust
earthquakes
between
the
continental
and
sub-
ducted
plates.
The
two
simplest
interpretations
of
these
observations
are:
(1)
the
Cascadia
subduction
zone
is
completely
decoupled
and
subduction
is
occurring
aseismically,
or
(2)
the
Cascadia
subduction
zone
is
uniformly
locked
and
storing
elastic
energy
to
be
released in
future
great
earthquakes.
Full
resolution
of
this
issue
675
676
THOMAS
H.
HEATON
AND
STEPHEN
H.
HARTZELL
may
prove
elusive.
Although
it
is
somewhat
surprising
that
no
shallow
subduction
earthquakes
have
been
documented
in
this
region,
the
duration
of
written
history
is relatively
short.
It
seems
certain
that
great
shallow
subduction
earthquakes
have
not
occurred
in
this
region
since
the
1850's
and
highly
probable
that
they
have
not
occurred
since
the
1790's.
If
large
shallow
subduction
earthquakes
do
occur
on
the
Cascadia
subduction
zone,
we
can
infer
their
characteristics
only
by
studying
the
nature
of
subduction
earthquakes
that
have
occurred
on
other
subduction
zones
that
can
be
considered
as
analogous.
Unfortunately,
no
subduction
zone
is
exactly
the
same
as
the
Cascadia
subduction
zone,
and
thus,
the
search
for
analogs
will
always
lead
to
fundamental
ambiguities.
Systematic
variations
in
the
nature
of
seismic
energy
release
with
physical
characteristics
of
subduction
zones
have
been
discussed
by
Kelleher
et
al.
(1974),
Kanamori
(1977),
Uyeda
and
Kanamori
(1979),
Ruff
and
Kanamori
(1980),
Lay
et
al.
(1982),
Peterson
and
Seno
(1984),
and
Uyeda
(1984).
Many
types
of
correlations
have
been
suggested,
but
since
quantification
of
the
physical
characteristics
of
subduction
zones
and
their
earthquakes
is
often
rather
subjective,
not
all
of
these
studies
seem
to
reach
compatible
conclusions.
Perhaps
the
most
consistent
obser-
vation
is
that
subduction
of
young
lithosphere
is
associated
with
strong
seismic
coupling
whereas
subduction
of
old
lithosphere
is
associated
with
weak
seismic
coupling.
Ruff
and
Kanamori
(1980)
demonstrate
that
the
seismic
coupling
is well-
correlated
with
a simple
linear
function
of
subducted
plate
age
and
convergence
velocity
in
which
decreasing
age
and
increasing
convergence
velocity
imply
stronger
seismic
coupling.
Heaton
and
Kanamori
(1984)
show
that
the
Cascadia
subduction
zone
is clearly
different
from
the
class
of
aseismic
subduction
zones
that
is charac-
terized
by
the
subduction
of
old
lithosphere.
They
further
show
that
it
shares
many
characteristics
with
subduction
zones
that
have
experienced
very
large
shallow
subduction
earthquakes.
In
this
report,
we
do
not
restate
all
of
the
arguments
given
by
Heaton
and
Kanamori
(1984),
but
instead
we
extend
that
study
by
systematically
comparing
trench
bathymetry
and
shallow
seismicity
for
a world-wide
sampling
of
subduction
zones.
TRENCH
BATHYMETRY
AND
GRAVITY
Uyeda
and
Kanamori
(1979)
suggest
that
strongly
coupled
subduction
zones
are
accompanied
by
relatively
shallow
trenches,
whereas
weakly
coupled
subduction
zones
are
accompanied
by
deep
oceanic
trenches.
Similarly,
they
conclude
that
free-
air
gravity
anomalies
tend
to
be
larger
for
those trenches
with
weak
seismic
coupling.
The
Cascadia
subduction
zone
is
somewhat
unusual
in
that
it
has
virtually
no
bathymetric
trench.
In
order
to
assess
just
how
anomalous
this
is,
we
have
con-
structed
profiles
of
bathymetry
and
free-air
gravity
for
many
circum-Pacific
con-
vergent
boundaries.
Figure
1 shows
shiptrack
segments
that
were
used
to
construct
the
profiles
shown
in
Figure
2.
The
shiptrack
data
were
obtained
in
digital
form
from
NOAA
(1981).
All
data
are
plotted
on
a common
scale,
and
the
subducted
plate
is always
on
the
right-hand
side
of
the
profile.
Bathymetry
and
free-air
gravity
are
plotted
together.
Unfortunately,
gravity
data
were
not
available
for
all
the
shiptracks
selected,
and
thus,
only
bathymetry
is plotted
for
some
of
the
profiles.
Comparing
the
profiles
from
differing
subduction
zones,
we
see
that
the
Cascadia
subduction
zone
(profiles
38-43)
is
indeed
remarkable
for
its
lack
of
a bathymetric
trench,
very
shallow
ocean
floor,
and
very
small
gravity
anomalies.
However,
there
are
several
factors
that
may
help
to
explain
these
characteristics.
The
shallow
3
km
depth
of
the
Juan
de
Fuca
plate
before
subduction
is
near
the
average
depth
of
very
~
0
<;;>
~70
71'-f:;/-
+90
+120
±ISO
-60
-30
FIG.
1.
Shiptracks
used
to
construct
profiles
of
bathymetry
and
free-air
gravity
for
world-wide
subduction
zones
shown
in
Figure
2.
00
c:::
o:l
t:::'
c:::
C')
j
0
z
t<:l
~
:I:
JO
c:::
>
::-::
t<:l
00
z
--3
:I:
t<:l
z
0
~
:I:
~
t<:l
00
--3
t<:l
::c
z
c:::
00
a;,
-l
-l
a
9
0
678
THOMAS
H.
HEATON
AND
STEPHEN
H.
HARTZELL
young
oceanic
lithosphere
(Parsons
and
Sclater,
1977).
Furthermore,
reflection
profiles
(Snavely
et
al.,
1980;
Kulm
et
al.,
1984)
reveal
the
presence
of
a shallow
trench
that
has
been
completely
inundated
by
approximately
2 km
of
sediments.
Grellet
and
Dubois
(1982)
show
that
there
is
a good
correlation
between
the
age
of
subducted
plate
and
both
the
absolute
trench
depth
and
the
trench
depth
relative
0
3
km
5
1
4
500
Tonga
200
100
~-~-~
Omgal
-100
-200
200
100
~;:__:~-...."('-~~-...j
0 mgal
IOOOkm
New
Hebrides
200
100
omgal
-100
1~200
3
.
100
km
...
._,_......,
........
-
5
•
"\
0
mgal
3
km
5
7
9
7
·..
!
-100
9
\/'
-200
6
,.
"
200
100
Omgal
3
100
km
1r4200
5
o
mgal
7
-100
500
200
100
omgal
-100
IOOOkm
0
200
100
omgal
-100
200
100
omgal
500
1000
Bonin
100
200
100
~~
....
...,
Omgal
-200
.::~
2
:
•
::gal
7
-100
9
-200
0
500
IOOOkm
200
100
omgal
-100
1500km
FIG.
2.
(a)
to
(g)
Profiles
of
bathymetry
(light
solid
line)
and
free-air
gravity
(dotted
heavy
line)
for
shiptracks
shown
in
Figure
1.
All
profiles
are
plotted
on
a common
scale
and
are
aligned
approximately
perpendicular
to
the
trench
axes.
The
subducting
ocean
plate
is
always
on
the
right-hand
side
of
each
profile.
The
number
in
the
upper
left
corner
of
each
profile
corresponds
to
the
shiptrack
numbers
in
Figure
1.
to
the
adjacent
abyssal
plain.
Once
again,
the
very
shallow
depth
of
the
Cascadia
trench
is
consistent
with
the
subduction
of
very
young
oceanic
lithosphere.
Searching
through
the
profiles
in
Figure
2,
we
see
that
there
are
other
subduction
zones
that
have
features
similar
to
those
observed
for
the
Cascadia
subduction
zone.
New
Zealand,
Nankai
Trough,
Alaska,
Colombia,
and
southern
Chile
are
all
notable
SUBDUCTION
EARTHQUAKES
IN
THE
NORTHWESTERN
U.S.
679
for
the
absence
of
a
well-developed
bathymetric
trench.
Hilde
(1984)
reports
extensive
trench
sediments
for
all
of
these
areas,
and
with
the
possible
exception
of
New
Zealand,
these
trenches
have
experienced
very
large
shallow
subduction
earth-
quakes.
We
particularly
note
striking
similarities
in
both
bathymetry
and
gravity
profiles
between
the
Cascadia
subduction
zone,
Colombia,
and
southern
Chile.
SEISMICITY
One
of
the
most
striking
features
of
the
Cascadia
subduction
zone
is the
remark-
able
paucity
of
shallow
earthquake
activity
between
the
trench
axis
and
the
coastal
b
19
1
"
20
3
)
\
km
·,
0
5
:
9
3
•
km
5
7
18
s\y
Kuriles
200
100
f""'----\:l~~~~~
Omgal
-100
-200
200
100
r--'--~::..::,._:]
o
mgal
-100
-200
200
100
omgal
-100
-200
200
100
{,----"---------.:.-.:::""i
0
mgal
-300
500
IOOOkm
North
Honshu
200
100
f,,.-~-----~~~....,,_...j
0 mgal
200
100
1.-----~
o
mgal
-100
1
14
200
3
•
100
-100
-200
Japan
km
1
\
5
'
rJ
J
--o!~""'"'~
0
mgal
7
v
-100
9
-200
0
500
Kamchatka
km:~
j
::gal
7f~Y~
~-100
:.
:M"
.
:::,
7
'
:
-tOO
9
\./
-200
km:~l
~j~
:~:gal
7
-100
9
,.,
-200
0
500km
IOOOkm
FIG.
2.
Continued.
mountain
ranges
(Taber
and
Smith,
1985).
Despite
the
continuous
monitoring
of
this
region
with
a local
seismometer
network
since
1970,
very
few
earthquakes
of
any
size
have
been
observed
in
the
coastal
regions
of
Washington
and
Oregon.
This
seems
somewhat
puzzling
for
a major
plate
boundary
that
is
undergoing
3 to
4
em/
yr
of
convergence.
Heaton
and
Kanamori
(1984)
suggest
that
this
behavior
may
be
analogous
to
that
observed
on
the
San
Andreas
fault
in
California.
That
is,
while
the
central
section
of
the
San
Andreas
fault,
which
appears
to
slip
principally
through
aseismic
creep,
experiences
numerous
small
earthquakes,
the
sections
of
680
THOMAS
H.
HEATON
AND
STEPHEN
H.
HARTZELL
c
d
3
km
5
West
Aleutians
..
:~·~~
;:,.,
7[
:r
1-100
••
:((
A.r.
f~~'"'
1~
\yrv·1-1oo
.::~
"
)
~~:.,
7r~
~-100
..
:·M~
::::00
7
"1
-100
9
·~
-200
0
Alaska
500km
100
omgal
-100
1
33
1964
earthquake
200
3
km
100
Omgal
5
1
3p~
7
~
200
100
km
5
:/
Omgal
7
-100
13~~
6
200
100
km
5
'J.
0
mgal
7
-100
0
500km
3
km
5
0
1rv9-~
200
3
100
km
5
omgal
7
-100
km
:r.~~~~~
~~~
East
:~
··.i\
..
:.
-~:
0
gal
Aleutians
9~
\/
-200
0
..
:~
_j
;:,.
7~
._y-l-100
500
Washington
km:~:::gal
7~~-100
40
500
200
100
Omgal
IOOOkm
200
100
Omgal
-100
IOOOkm
1
rs
3
Oregon~
200
3
100
km
5
>1"\1
.,..
•
-~:
0
mgal
7
-100
3
km
5
0
42
FIG.
2.
Continued.
500
200
100
omgal
-100
200
100
omgal
-100
1000
km