in the rate of crustal uplift with time since the earthquake. There is ... Space Flight Center, Greenbelt, ..... An enigma is the recent GPS result at site VAN DUSEN.
CrustalUplift in the SouthcentralAlaskaSubductionZone:A NewAnalysisandInterpretationof TideGaugeObservations StevenC. CohenI and Popular
measures
sea level height seismically
the ocean
can be of either
active
tectonic
at appropriately
have examined
height crustal
relative
located
the tide gauge
the long term tidal gauge
indicates
provide
at the seven
rates
origin.
are often
tide gauges records
to a land datum;
or oceanic
plate boundaries
by the massive (magnitude 9) 1964 Prince records to determine the rate of postseismic pared
T. Freymueller
2
Summary:
A tide gauge
ments
Jeffrey
therefore,
Because
more rapid
insight
changes
the rates
in apparent
of crustal
uplift
than sea level changes,
into crustal
at
measure-
deformation
processes.
Alaska
that were affected
sites in south-central
We
William Sound, AK earthquake. We have used these (after the earthquake) crustal uplift and have com-
with more
recent
GPS rates
at nearby
sites. Our analysis
the following:
(1) The ongoing
the coseismic
(earth-
quake) motion in the sites that subsided during the earthquake are now moving visa-versa. In detail, however, the correlation between the amount of coseismic
up and motion
and the current (2) At Kodiak earthquake. at Valdez uous.
pattern
there
There
mirrors
is a distinct
decrease
to some
extent
in the rate of crustal
is also a mathematically
and Cordova,
rapid,
motion
rate of uplift is poor.
(3) On the western is quite
of vertical
although
the behavior
side of the Kenai
-1 cm/yr.
Such
significant
change
at Valdez,
Peninsula,
uplift
in particular,
i.e. at Seldovia
a rapid uplift cannot
with time since the
in the rate of vertical
be maintained
is somewhat
and Nikiski
motion ambig-
the uplift
over the several
rate
hundred
year recurrence interval between great earthquake; however, it is consistent with the "anomalous" southeast motion of GPS sites on the western side of the Kenai Peninsula that we have previously reported. (4) Vertical motion at some other boundary. Peninsula, strain
In particular, is consistent
accumulation
I Geodynamics -Geophyslcal
Branch, Institute,
sites is consistent
with strain
accumulation
at a plate
the vertical motion at Seward, on the eastern side of the Kenai with our interpretation of the horizontal GPS velocities in terms
due to steady-state
NASA's University
Goddard
plate convergence.
Space
of Alaska,
Flight
Fairbanks
Center
of
Created March, 1999. Current version November 3, 1999 2:20 pm
Crustal
Uplift
in the Southcentral
New Analysis
1Geodynamics
(phone:
Branch,
C. Cohen
Goddard
301-614-6466;
2Geophysical
Alaska
and Interpretation Steven
Space
Institute,
University
of Alaska,
T. Freymueller
Center,
FAX: 301-614-6522;
Subduction
of Tide Gauge
1 and Jeffrey
Flight
l
Greenbelt,
email:
scohen
Fairbanks,
Zone: A
Observations 2
MD 20771
@carnoustie.gsfc.nasa.gov)
AK 99709
Abstract
We have examined
the sea level height
southcentral
Alaska
the history
of crustal
behavior
that were affected
depending
uplift
ern side of the Kenai lation
while
uplift
of about
quake,
uplift,
at Seldovia 1 cm/yr
but which
of several
hundred
which
Peninsula
on the western
likely
be sustained Further
is still several
mm/yr,
represents
William There
earthquake
is considerable
side of the Kenai a long term transient
to the southwest,
sites in the region
Sound
that is consistent
over the expected
has slowed
tide gauge
to the 1964 rupture.
we find a slow uplift
that most
years.
to the earthquake.
of the site relative
and Nikiski
cannot
at seven
by the 1964 Prince
subsequent
on the location
tide records
recurrence
at Kodiak,
significantly
At Seward,
with elastic
response
interval
strain
for a great
we find evidence
accumu-
over the past three
earthquake
and a half decades.
and an uncertain
Valdez.
sites there
is a mathematically
significant
time-dependence
are not as convincing
rapid
that the rate of
at Cordova
this time dependence
on the east-
to the earth-
we find subsidence
but the data confirming
in the
we find a persistent
Peninsula
behavior,
to determine
variation
To the east of the Kenai At both of these
of
behavior
at
to the uplift
as at Kodiak.
At
Created March,1999. Current version November 3,19992:20pm
2
Anchorage,to the north thereis little evidenceof verticalmotionsincethe earthquake.Wecomparetheselong termtide gaugerecordsto recentGPSobservations.In generalthereis reasonable consistencyexceptat AnchorageandCordovawheretheGPSmeasurementindicatesomewhat morerapid uplift andsubsidence,respectively.
Created Ma,'ch, 1999. Current version November 3, 1999 2:20 [_m
3
Introduction
Among
the several
W_liam record
geodetic
data sets that have
Sound,
AK earthquake
of ongoing
postseismic
first researchers
to exploit
to 10 yrs long collected Seward,
Kodiak
uplift
occurred
postseismic
(M w - 9.2), tide gauge and interseismic
these
(St. Paul Harbor),
of postseismic
motion
partially
during
the earthquake
showed
uplift
the magnitude The
uplift
rate of 88 + 38 mm/yr
Plafker tions
be sustained
[1991],
herein
and others
slower although between
(sometimes
rates
(Figure
referred
and visa-versa.
to as SP91,
from
the entire
a decade
derived
make
In detail,
(through motion
than
in the uplift rates
since
only the post-1973 argument the SP91
1988).
that
The spatial
the correlation uplift
except
rate was
with a maximum
at Cordova, Savage
speeds and
at these
In general
did Brown
loca-
they found
et al. [1977],
to distinguish
to 1973 (i.e. -9 years
data were
Having
showed
observations
for re-examining
analysis.
up
at Anchorage,
however,
of years.
data set and from data subsequent using
the
records
analysis
10 mm/yr,
the tide gauge
of vertical
analyzed
of postseisrnic
of hundreds
duration
variation
a compelling
has passed
cycle
were
in that sites that subsided
of - 12 + 2 mrn/yr
re-examined
rates
any temporal
The rates
At least two factors
slower)
They
at Cordova.
motion
at all sites exceeded
earthquake
et al. [1977]
1). Their
and the magnitude
the most continuous
sites located
occurred
the coseismic
and a minimum
over the entire
considerably
derived
at this time. First,
at Kodiak
rates.
and Cordova
motion rates
Brown
tide gauge
thereafter
of the inferred
they did not report
movement. uplift
of the 1964 Prince
provide
at permanent
with a data set that was of longer
after the earthquake). slower.
value
observations
postseismic
mirrored
of the coseismic
weak.
that cannot
absolute
in the aftermath
at the first 4 sites, but subsidence
pattern
between
crustal
data to extract
after the earthquake
Seldovia,
been studied
found
to be slightly
the tide gauge data spanning
record
another
ten
Created March, 1999. Current version November 3, 1999 2:20 pm
years
is important
deduced
because
from annual
long fluctuations
due to non-tectonic
is now sufficiently
period
and (at somesites)
Data
robust
for comparing
to determine
sites located
to both year-to-year
As the data presented both average
variations.
Second,
near to the tide gauge
locales
short term to long term
vertical
monthly
mean
in south-central
web. These
sea level Alaska
sites are: Anchorage,
Paul Harbor
and Women's
Bay).
and Juneau)
below.
data were supplemented
The
data we formed
annual
following
the procedure
outlined
effects
at the individual
east Alaska
locales,
ence between
and several-yearindicates,
for the elapsed
contemporary
GPS
and provide
and the rates
in this paper
uplift rates
the
postseismic
data are becoming
a complementary
set of obser-
movements.
based
a region
our procedure
to correct
means
well removed
means, values.
to minimize
Seward,
for three
southeast
between
the residuals
residuals
from
corrects
whereas Since
some
the possibility
From
any annual
at the three
that the analysis
means
means
southdiffer-
that were
in SP91
oscillation would
and
tide gauge
The only significant
annual
these
sea level change
to the detrended
of the annual
there is a strong
sites (Sitka,
oceanographic
rates
(St.
as discussed
of apparent
for local
zone.
is that we rejected
Alaska
tide gauge.
from detrended
1964 rupture
and Kodiak
effects
rates
.tide gauge
over the world-wide
Nikiski,
for the Seward we derived
permanent
available
for local oceanographic
The procedure
and that of SP91
from less than six monthly
records
records
and the average
derived
Valdez,
from which
in SP91.
for the seven
data archives
Cordova,
by paper
from less than nine monthly
we sought
the NOAA
on the coherence
stations
determinations
We also extracted
derived
readings,
from
that are used
monthly
atmospheric
height
Seldovia,
Ketchikan,
gauge
occurred
Analysis
We extracted
rates
since the earthquake
are subject
processes.
temporal
at sites located
vations
has passed
sea level determinations
record
available
only 35 years
4
were
in the tide
be biased
by sea-
Created March,1999. Current version November 3,19993:25pm
5
sonaleffectsevenif it meantsomedecimatingthedatasetsomewhat.Nevertheless,the apparent sealevel rateswe obtainedby analyzingthe datathrough1988closelyreproducedthe rates reportedin SP91.
We estimated
eustatic
sea level change
change,
whereas,
mm/yr
is somewhat
SP91
uplift rates
and postglacial added
arbitrary
seven
are summarized
Constant Figure
2 shows
the uncorrected corrected
and corrected
observations.
is given
Rate
the apparent
in Table
is 85, 80, 42, 47 and 82 percent but nearly seven
zero at Kodiak
locales.
1998),
suggesting
correlated apparent
that the southeast
at Kodiak
Valdez,
to the apparent
are several
Valdez
the data set is too sparse)
the corresponding
than linear)
uncertainty
the value
of 2
uncertainty
in
at each of the
the residuals
correlated
shape
at Kodiak,
sea level correction
Anchorage,
Valdez,
respectively, fit at each
(e.g. between
of the
1996 and
in removing
all
are the systematic
residuals
of these residuals
suggests
a least squares
regression
in time for each site except
and found
on the
rate of sea level change
is not fully successful
The
both
regression
to the linear
residuals
great importance
polynomial
a linear
Alaska
Seldovia,
sea level rate. We performed
Anchorage above
Seward,
and Cordova.
(rather
from
by the southeast
sea level correction
data using a quadratic (for which
any additional
line, i.e. the apparent
3 shows
Of potentially
and possibly
Although
of the data record
and the line derived
achieved
Figure
that there
sea level fluctuations.
time-dependence
reduction at Cordova,
It is noteworthy
rate of sea level
1998 for all 7 sites with each plot showing
of each regression
and Nikiski.
for
Analysis
sea level heights
2. The variance
aspects
(to account
1.
sea level through
The slope
for the same effects.
The site-specific
2 mm/yr
of the apparent
level, we did not attribute
in Table
and Time-Dependent
to the negative
to account
at the mm/yr
due to this choice.
tide gauges
rebound)
2.5 mm/yr
our rate estimates
at each of the sites, by adding
that the rate change
and Cordova,
Nikiski
a
on the and
was formally
but was indistinguishable
Credited March,1999. Current version November 3,t9992:20pm
6
from zeroat SewardandSeldovia.The apparentsealevel ratesderivedfrom thequadraticregressionsarealsosummarizedon Table2. In Figure4 we comparethe linearandquadraticfits to the data.Sincetheerrorsin the tide gaugeobservationsandmonthly andannualmeansarenot known,we estimatedrateandratechangeerrorsfrom the misfit of theregressioncurveto the data,asoutlinedin Press
et al. [1986].
may underestimate
the true error.
subject
of oceanographic
to a variety
addition
it is well known
yet fully understood. yr even
when
Thus
are likely
robustness
in the tide gauge
There there
this question
questions
provided
to a variety
Conversely,
in several
observations
of error,
are also
which
to rate differences differences
data sets spanning
this consistency
in this manor
in the data processing.
the data set is sufficiently
requires
a more
results
with other
motions
in terms
Apparent
sea level change studies
mation
following
leveling
surveys
there
we wish to discuss
in the rate of crustal
by taking
observed
In previous
tide gauge
significance error.
obtained
robust.
a decade
In
are not
of a few mm/ exceeding
sev-
Achieving
such
or longer
whereas
years.
Interpretation
any change
the tide gauge
data is subject
the reported
observations
can achieve
are two important
has been
exceed
from
that are not fully removed
little physical
to be significant
the GPS measurements and
noises
we ascribe
these differences
that error estimates
The rate determinations
that the GPS vertical
eral mm/yr
Discussion
It is well known
critical
data. The second
of the underlying
tectonic
uplift
has been evidence
the 1964 earthquake. and concluded
Brown
that there
The
first is the whether
since the 1964 earthquake.
look at the tide gauge
geodetic
and crustal
uplift
in this section.
data itself question
processes
We will examine
and then we will compare
is how to interpret
the
of the region.
rates both for and against et al. [1977]
time-dependent
and Cohen
was a time dependent
decay
[1998] over
crustal examined
a period
deforseveral
of several
Crealed March, 1999. Current version November 3, 1999 2:20 pm
years
in the rate of crustal
papers
[Cohen
cumulative process
postseismic
as well.
uplift
However,
used than when larger
recently Kenai
might
1998)
the results
recurrence
interval
edge. Where
of the analysis
Since
be operating.
we have
argued
rates
when
postseismic
were
et al., [1999])
of the western
to much
prohave
portions
of the
slip on deep portions
and not subject
was
that a relaxation
We [Freymueller
postseismic
uplift
all the data
of the rates they observed
motion
of
that the
change;-i_n the postseismic
they speculated
horizontal
In a series
with a transient
slower
some
averages,
that is unlocked
in Freymueller
moment
rate of 9.7 _+0.5 ram/yr.
years;
hundred
following
regression
+ c (At) -, where
the SP91 analysis
cannot
gives
At = 1998.5
(data
analysis
The nearly be sustained
thus it is not entirely analysis
shows
a rate change,
- t, of2c
the site located
to other
they are from
et al. [1999]
from the linear
of several
at Cordova,
and progressing
we cite GPS results,
change
The quadratic
sites starting
rupture,
of sea level height
data for 10 years
h = a + bAt
for some
or delayed
at the individual
to the SP91-determined estimate
change.
be due to continuing
1997]
is consistent
1988 produced
trenchward
the rate of sea level change
the earlier
porating
of 102 might
a region
and Freymueller,
evidence
through
end of the 1964 coseismic
At Cordova
pares
some
the earthquake.
of the release
earthquakes.
from the rupture
data from
Cohen
and interseismic
that lie below
We will discuss
tance
data analyzed
Arm following
the late 1990s
found
that the unexpected
Peninsula
the downdip
SP91
on a time scale
argued
great
1996,
data prior to 1973 was excluded.
plate boundary during
Turnagain
through
than the preseismic
cess occurring
along
et al., 1995, Cohen,
rate as their tide-gauge
much
uplift
7
sites at increasing
an update
through
1 cm/yr
of
1997). This com-
subsidence
over the entire surprising
to
dis-
(inclusion
is 6.7 _+0.4 mrn/yr.
a deceleration
closest
implied
by
earthquake
that our data incorin the sea level height
2c, in the sea height
equation,
= 0.39 + 0.06 mm/yr 2. However,
an
Created March, 1999. Current version November 3, 1999 2:20 pm
examination
of the height
dependent
signal
skeptical
comes
as to whether
ically
real. Nevertheless,
uplift
rates
height
shown
since
residuals passes
residuals
rate solution
from the early
and later epochs
the mathematically
significant
we must
in Figure
the F test [Mikhail,
that there
We applied
with the quadratic
curve 1976]
is meaningful.
at the i% level
(SSR 1 - SSR2)
SSR is the sum of the squared
number
of parameters.
The subsidence
the average quadratic
rate from
analysis
surveys.
that can increase
the actual
whether
in apparent
sea level
the improvement
in the
The improvement that the
et al., 1995]
(1)
/ (DF 2)
provide
than the nearly
a weak
GPS vertical
of data points
argument
which
against
is in better
measurements
well beyond
the formal
error
11 years
after the Prince
the
a rate change. with
predicted
bars because are subject
minus
agreement
zero rate of 1.8 + 1.0 mm/yr
rate has large error
uncertainty
rate is phys-
of the corrected
significant.
and DF is the number
The GPS-derived
In addition,
evaluation
that the 99% probability
the GPS data is - 13.2 _+7.1 mm/yr
for 1996.5.
we are somewhat
/ (DF 1 - DF2)
GPS observations
rate of - 4.7 + 0.4 mm/yr
only a few recent errors
Recent
residuals
of the time
in the uplift
little if any change
indicating
is Zhao
Thus
variation
case is statistically
( SSR2)
where
temporal
has been
over the linear
that most
of the record.
the F test to determine
The F statistic
F --
shows
point out that even a qualitative
2 shows
the late 1980's.
improvement
to the constant
8
by the
it is based to a variety
estimates
on of
[Mao
et al.,
1999] At Valdez
the data record
quake.
The rate of apparent
cantly
different
mathematically
dates
back
to 1975,
sea level change
since
from zero,
but is less than the SP91
significant
change
that time,
William
- 0.2 + 0.7 mm/yr,
rate of 5.5 + 0.9 ram/yr.
Sound
Earth-
is not signifi-
There
is a
in the sea level rate of 0.77 + 0.14 mm/yr 2 that produces
a
Created March, 1999. Current version November 3, 1999 2:20 pm
rapid
transition
from crustal
subsidence
the F test for an improvement rate in 1996.5 3 again
result
feature
in many
found
particular
away, do marginally
about
Our qualitative
in view of the residual
confirm
the recent
and yields
ascribing
of the plots However,
The quadratic
an estimate
assessment
a great offset
between
albeit
of the residuals
passes uplift
in Figure
significance
to the
1995 and 1996 is a correlated
that come
with a large
solution
for the crustal
deal of physical
GPS measurements
uplift
sea height
from a site about
uncertainty.
The
5 km
GPS-derived
rate is 8.7 _+8.2 ram/yr.
At Seward
nificantly
the apparent
different
cent TI9, installed
At Kodiak
appears
lines of argument
height
residuals
i.e., it appears
9.5 + 2.7 mm/yr
located
shows
by SP91
from measurements
from measurements
change
the entire
variation domain
that the improvement
the rate that we derived from the data through
from
from the linear
is more
since
begun Marine
analysis
in the rate of - 0.66
over the linear
the data through
1998
First,
less
1993 at the adja-
in 1995 at the newly Facility
in Seward.
an examination
than in the previous solution
fit is statistical
is slightly
The crustal
sea level (for 1996.5,
is slightly
+ 0.11 mm/yr 2. Several
is real.
extensive
is insig-
is - 14.5 + 0.7 mm/yr;
of the data set. The quadratic
1988, - 17.5 + 0.8 mm/yr.
in apparent
of Alaska
that the time dependence
that the temporal
change
obtained
and the time dependence
uplift rate of 3.2 + 0.7 mm/yr
obtained
sea level change
also suggest
to span almost
the time-dependent
crustal
at the pier of the University
to be a significant
test with a 99% probability thermore,
The implied
the rate of apparent
there
is - 1.2 _+0.7 mm/yr
one, 7.6 + 3.5 mm/yr,
and the other, site UAMF,
however,
sea level change
from zero.
than two GPS results,
other
skeptical
uplift.
solution
9.2 _+ 1.6 ram/yr.
us somewhat
mathematical
uplift
over the linear
that is a rapid
makes
to crustal
9
see Table
lower
uplift
of the cases,
passes
significant.
the F Fur-
than that derived
rate we predict
2) is 7.4 mm/y
which
from is
Created March,1999. Current version November 3,19992:20pm
10
goodagreementwith the GPSobservedvalueof 9.2 _ 1.6 mm/yr.Thesevaluesare lessthanthe 14.8_+7.2 mm/yr observedby VeryLong BaselineInterferometry(VLBI) in the late 1980'sand 1990[Ryan,et for the period
al., 1993], (1985-1998)
- 8.1 _+ 1.1 mm/yr, 1981 when
between
rate change change
uplift.
estimated
in Cohen
and elsewhere
rate in about
[1998]
rate of - 7.2 _ 1.4 mm/yr.
between rently
installing
crustal
12 years.
the somewhat
confirms
a "permanent"
Bay is
was associated there
of sites along
to suggest
change
in the speed
faster
decreased
Turnagain
Arm,
the eastern
in good
agreement
GPS site at Seldovia.
the GPS uplift
side of the Kenai comparison
Kenai,
sites.
with SP91's
0.16 + 0.18 mrrdyr 2, is not significant.
a quantitative
of
than the 3-6 year decay
is - 9.3 _+0.8 mnfyr
uplift on the western
rate
to half its
of the various
rapid
that the
of
rate there
Although
in
check
the observed
settings
25 km to the northeast.
that the
with the change
tectonic
term,
1967 to
was a consistency
evidence
physical
This is somewhat
from
to the suggestion
In fact, we consider
that the uplift
about
is to far to make
credence
rate at Kodiak
different
The sea level rate
for the period
no substantive
for an actual uplift
we estimate
The rate change
GPS site is in Homer,
the two locales
candidate
at Seldovia
derived
1 indicate,
of the tide gauge.
for an amalgamation
and may reflect
of 18.7 ± 2.4 mm/yr
likely
- t) mm/yr,
The rate of sea level change
nearest
in Table
significant.
at Women's
that the rate change
of the relocation
the time dependent
postseismic
located
This adds further
as the comments
to be the most
From
has been
at the two sites so we have
r = 6.1 + 0.66 ( 1998.5 immediate
the possibility
However,
is an artifact
at Kodiak
crustal
was at St. Paul Harbor.
the observations
is not statistically
than the rate of - 21.1 ± 1.4 mm/yr
but raises
location.
the difference
that the tide gauge
slower
the gauge
rate has changed, observing
although
Peninsula,
of the two rates.
There
rate at Homer the distance We are cur-
Created March,1999. Current version November 3,19992:20pm At Anchorageweobtaineda sealevel rateof 0.8 + with the SP91
result
at Anchorage.
Our analysis
incomplete.
However,
data set goes back 1980's.
southwest
uplift
for the next nearest
suggested
rate of 12.4 + 5 ram/yr.
The apparent
sea level rate at Nikiski
The GPS crustal
uplift
located
was derived
Given
obtained
set of monthly
heights
indicate
about
again
by observations
movement
records
are quite
tidal heights.
This
and early
little
vertical
located
only 8 km to the
crustal
uplift
be dismissed
10 km southeast
from mean
consistent
crustal
suggesting
a rapid
of
as a single
site
of the tide gauge,
tide heights
rate of - 18.7 _+1.7 mm/yr
rate has been determined
is quite
the early mean
the fact that no data was collected
a higher
which
for most of the 1970's
by the GPS data cannot
GPS site, 1000,
and is - 9.9 + 0.8 ram/yr.
in 1984 since
at site VAN DUSEN
The GPS measurements
The rapid
1997, the fact that SP91
robust
yearly
GPS result
an uplift
level
more
a result
little, if any vertical
a rate of - 0.7 + 0.9 mm/yr,
is the recent
of the tide gauge.
very
on data beginning
is a somewhat
produces
1.3 mm/yr,
and suggests
to 1965 but still is missing
An enigma
16.1 + 6.5 ram/yr. anomaly
was based
there
The tidal heights
movement.
gives
of 1.9 + 1.9 mm/yr
11
rather
at Nikiski
is probably
that were begun
than mean from
sea
1979 to
not significant. in 1993 and is
13.3 + 3.2 mm/yr.
As we discussed have experienced
above,
postseismic
died off near Seward any similar
a portion
relaxation
to the south,
short term relaxation
puted
the quadratic
tions
were strongly
regression effected
of the Kenai
curves
and the Turnagain
with a characteristic
and perhaps influenced
Peninsula
time of a few years.
near Anchorage our estimates
excluding
by this perturbation.
Arm
to the north.
of changes
1973. None
At Kodiak
the rate change
That
To determine
in uplift
data through
region
rate,
may
relaxation whether
we recom-
of the quadratic
solu-
with the decimated
Created M:u'ch, 1999. Current version November 3,19992:20pm
12
datasetwas- 0.59_+0.16mm/yr2 which agreesvery well with theratechangeof - 0.66 +
0.11 mm/yr 2 derived
from
the entire
change
is 0.33 + 0.09 mm/yr 2 compared
Valdez
the datd'set
begins
data set. Similarly
to the 1965-1998
at Cordova,
rate change
in 1975, so it does not contain
the post-1973
rate
of 0.39 + 0.11 mm/y_ 2. At
any observations
from the first decade
after ihe earthquake. Physical
Significance
We now turn to a discussion apparent
sea level data. Table
uplift
rates
sites,
with the exception
during
deduced
perhaps,
motion
in the NE portion
et al., 1996]. The 55 mm/yr
yr at N28W rupture
observations.
lie landward
velocity
in infrequent
where great
the Kodiak
Island
region
yrs) large
earthquakes
Given
the considerable
vertical
As previously
coseismic To some
varies
Plate relative
is less than 5 degrees
earthquake
with recurrence
variation
both in infrequent and Jacob, in trench
at Kodiak
America
obliquity. intervals
great
1990; Perez orientation
1, all of the
lies over
pattern
except, N44W
increases
there
The eastern
region
earthquakes and SchoIz,
hundred
and more
a
Johnson
et al., 1994]
of 25 degrees
of several
in the from
et al, 1994;
[Demets
a por-
of postseis-
uplift
but is about
[Brocher,
and
that slipped
Cordova
plate boundary
with an obliquity
there
[Nishenko
with locale,
to North
1971]
in Figure
the observed
locked
of the figure to 10 degrees
ruptures
shown
subsidence.
by these
[Plafker,
with all the sites exhibiting
trench
at Seward,
implied
of the megathrust
extent
of the interseismically
N17W
rates
motion
of that portion
pattern
of the offshore
of the Pacific
in a direction
at Kodiak
uplift.
the coseismic
The dip angle
of the uplift
both the coseismic
and thus underwent
The strike
of the figure.
significance
and thus experienced
mirrors
Cordova.
few degrees
about
of Cordova,
the 1964 earthquake
mic vertical
3 shows
from the tide gauge
tion of the slip plane
center
of the physical
is
and 59 mrrd tends
years, frequent
to
while (50-60
1997].
and slab dip in the region
under
study,
we
Created March,1999. Current version November 3,L9992:20pm
13
havefoundit moreinstructiveto
plot the uplift data as a function
maximum
1964 earthquake
coseismic
tion of distance detic
from the trench.
data for it lies directly
dislocation cally
subsidence
model
locked
the locked
region.
parameters,
The data
1985,
the effects rates
Valdez,
where
and Kodiak
agreement
tectonic
and Frohlich, Kodiak
because
ate for the great coseismic observed buttressed
crustal
sustained
The
an uplift
of less than a meter for the several
for Seward,
period
there
is consistent
GPS measurements
in Figure
cautiously,
too great
accumulation.
and Freymueller
of 5nor-
of these the
since
it lies
here [Pulpan
is sustainable
recurrence
thought
5 and Table
strain
of Cohen
For all three
rate. At Kodiak
earthquake
be much
depth
for Cordova,
sites considered
year interval
would
with elastic
value.
disloca-
to the trench
except
rate of 16.5 mm/yr
for a large
hundred
shown
of distance
representative
relative
particularly
than the other
1996]. However,
results
by the contemporary
vector
end of
of an elastic
assuming
than the 1996.5
must be viewed
to the east, then the uplift
movement
as a function
postseismic
better
of the plate boundary
and interseismi-
the downdip
are the results
velocity
of .geo-
slip, elastic
ruptured
both the mean rate and the 1996.5 model
than as a func-
dip angle of 5 deg, and locking
for the entire
dislocation
it give a net uplift
subsidence.
values
in a uniform
uplift,
the axis of
in the interpretation
motions
in the figure
of the plate
rate with the model
earthquakes
vertical
rate of 55 mm/yr,
we show
1985; Lu and Wyss,
50 yrs. If such a rate were
the interseismic
from [1971])
this axis also lies over
of the rate of interseismic
are mean
segment
interest
that the coseismically
are the same,
of the obliquity
rate fits the elastic
of the average
in a different
estimate
by Plafker
end of the rupture
5. Also shown
convergence
mal. The tide gauge
sites the mean
downdip Assuming
showing
in Figure
1992]
i.e. a plate
25 km. We ignore
0ve_.!he
of the megathrust
from the axis are plotted tion [Okada,
This axis is of considerable
of the earthquake.
portions
(as estimated
of distance
at
time of
to be approprito be relieved
3, suggest
that the
This argument [1997]
by
and
is
Created March, 1999. Current Freymtteller
et al. [1999]
of the Kenai
Peninsula,
may have
been rapid
version November 3, 1999 2:20 pm
that show that the pattern including
transient
gain Arm in the decade
agreement
and Valdez. expected
for elastic
et al. [1999] throughout
show
motion
rust conditions forward steeply
slip on the shallow
tonically release deep
results
different
downdip
slip model
ory results
segment.
during
of the parameters
observations
on the western
do not claim
this interpretation
Conclusions
that Seldovia
side of the Kenai
depths
peninsula
is the only possibility.
constrained
than would
the locked
that the megath-
with continuous slip on a more
do in fact lie in a tec-
matter 6 shows
coseismic
moment
to find a transient the dislocation
of 40 km [Oleskevich geometry
so we can claim
are consistent
be
of sites
speculated
by the small
fit to the plate interface
are uniquely
motion
postseismic
Figure
are also is
in Freymueller
a model
then it is a simple
between
faster
from
and Nikiski
sites (as suggested
at 25 degrees
is a reasonable
southeast
proposed
is such
are used for Cordova
are much
and transient
there. For example,
and Anchorage
velocities
different
Turna-
of that uplift
et al. [1999}
quite
interface
the 1964 earthquake),
1999] and 80 kin. The geometry
cm/yr)
They
pattern
side
There
and along
GPS measurements
Freymueller
Peninsula.
If we assume
dipping
(-2
conditions
of the plate
than fits the observations
but none
rapid
than the other
for slip on a plane
et al. [1999],
the horizontal
Peninsula.
side of Kenai
environment
in this region
and Seldovia
accumulation.
Peninsula
Valdez,
at Nikiski
from tectonic
portion
at Cordova,
on the eastern
strain
but the spatial
the average
Again
side of the Kenai
there
with elastic
provided
is an anomalously
on the eastern
dipping
uplifts
deformation
side of the Kenai
motions
model
accumulation.
that there
the western
horizontal
The vertical
the crustal
strain
are consistent
the 1964 event,
with the elastic
By contrast,
of horizontal
uplift on the eastern
or so following
that it died out near Seward. reasonable
Seward,
14
found
theet al.,
by Doser
only that the
with deep transient
slip. We
Created March, 1999. Current version November 3, 1999 2:20 pm
The tide gauge temporal Sound
records
distribution Earthquake.
in southcentral
of vertical
crustal
The longer
data
and we find considerable decay
time on the order
the western L
span has enabled
data
to the great
us to search
the record
10 years
and
William
of observations
and recent
GPS mea-
for time-dependent
behavior
in the rate of crustal
has persisted
the spatial
1964 Prince
these obser.-y, ations
We also find that the rapid
Peninsula
set for studying
incorporates
between
for a decrease
of a decade.
side of the Kenai
subsequent
of the records
a comparison
evidence
are a useful
motion
This new examination
since the last study and provides surements.
Alaska
15
uplift
uplift
at Kodiak
at Seldovia
with a
and Nikiski
on
for at least three
and a half decades.
There
is
on the eastern
side of the rupture,
to deter-
.
a need for further mine
whether
observations
the apparent
at Cordova
rate of change
tent with the idea that postseismic dependent quake
on both location
at these
relaxation
relative
sites is real. The data presented
may be a multiphase
to the coseismic
process
here
are consis-
with the details
rupture
features
and time since
for her insights
through
an ongoing
the earth-
occurrence.
Acknowledgments: Alaska
and Valdez,
tectonics.
and Kodiak.
We thank
Jeanne
We also thank
We also express
leveling
records
NASA's
Solid Earth
Sauber
her for providing
appreciation
that were not available and Natural
Hazards
some
of the GPS data, particular
to Scott Duncan through
the internet.
Program.
dialogue
for providing This research
tide gauge
on
from Cordova records
was funded,
and
in part, by
Created March, 1999. Current version November 3, 1999 2:20 pm
16
References
Brocher,
T.M.G.S.
megathrust Res.,
Fuis,
beneath
M.A.
the northern
99, 11663-11685,
Brown,
L.D.,
Cohen,
Alaska,
J.J. Tabor,
Gulf of Alaska
using
and N.J. Chfistensen,
wide-angle
seismic
Mapping
the
data, J. Geophys. ;_
S.R. Holdahl,
J. Geophys.
S.C., Time-dependent
Alaska
G. Plafker,
1994
R.E. Reilinger,
Anchorage,
Fisher,
Res.,
uplift
since the 1964 Prince
and E.I. Balazs,
82, 3369-3378,
of the Kenai
William
Sound
Postseismic
crustal
and adjacent
areas
uplift
near
1977.
Peninsula
Earthquake,
J. Geophys.
of southcentral
Res.,
101,
8595-8604,
1996.
Cohen,
S.C., On the rapid
Prince
William
Cohen,
S.C.,
insula,
S.Holdahl,
Alaska
2038,
Sound
since
postseismic
uplift
along
Turnagain
Earthquake,
Geophys.
Res. Lett.,
D. Caprette,
S. Hilla,
R. Safford,
the 1964 Prince
William
Sound
Arm,
Alaska
following
25, 1213-1215,
1998.
and D. Schultz,
Uplift
Earthquake,
J. Geophys.
the 1964
of the Kenai Res.,
Pen-
100, 2031-
1995.
Cohen,
S.C.,
and J.T. Freymueller,
102, 20,479-20,487, Demets,
on the Kenai
Peninsula,
Alaska,
J. Geophys.
Res.,
1997.
C., R.G. Gordon,
reversal
Deformation
D.F. Argus,
time scale on estimates
and S. Stein,
of current
plate
Effects motions,
of recent
revisions
to the geomagnetic
Geophys.
Res. Lett.,
of the Prince
William
21, 2191-2194,
1994.
Doser,
D.I., A.M.
over thirty Freymuetler, insula,
years
Veileux,
and M. Velasquez,
following
J.T., S.C. Cohen,
Alaska,
the 1964 Great
Seismicity Alaskan
and H. Fletcher,
and their implications,
earthquake,
Variations
J. Geophys.
Res.,
PAGEOPH,
in present-day (submitted
Sound
region
in press,
1999.
deformation,
to JGR,
3/23/99).
Kenai
for
Pen-
Created March,1999. Current version November 3,19992:20pm
17
Johnson,J.M., K. Satake,S.R.Holdahl,andJ. Sauber,The 1964PrinceWilliam Soundearthquake:Jointinversionof tsunamiandgeodeticdata,J. Lu, Z., and M. Wyss, estimates
based
Mao, A. C.G.A.
on fault
E.M.,
Nishenko,
and Least
J. Geophys.'Res,
Am.,
75, 1135-1054,
boundary
Res.,
101,522-532,
derived
from
101, 803-816,
in GPS coordinate
1996.
stress
direction
1996.
time series,
J. Geophys.
Res.,
D.A.,
tion earthquake:
deformation
Seismic
pp26,
potential
Harper
and Row, New
of the Queen
York,
1976.
Charlotte-Alaska-Aleutian
Seis-
1990.
due to shear
and tensile
faults
in a half-space,
Bull Seismo.
Soc.
due to shear
and tensile
faults
in a half-space,
Bull Seismo.
Soc.
1992.
R.D. Hyndmann, Thermal
J. Geophys.
Squares,
1985.
Am., 82, 1018-1040,
Res.,
during
and K. Wang,
and structural 104,
O.J., and C.H. Scholz,
earthquakes
Noise
95, 2511-2532,
deformation
Y., Internal
Oleskevich,
J. Geophys.
and T.H. Dixon,
Observations
Y., Surface
Perez.
solutions,
plate
Res.,
1999,
Okada,
Chile,
of the Aleutian
S.E, and K.H. Jacobs,
mic Zone,
Okada,
plane
Harrison,
104, 2797-2816, Mikhail,
Segmentation
Geophys.
models
14965-14991, Long-term
the time between
The updip
and downdip
of Cascadia,
south
limits
Alaska,
to great
SW Japan,
subducand
1999.
seismic
behavior
successive
shocks,
of the focal J. Geophys.
adjacent Res.,
102,
regions
of great
8203-8216,
1997.
Plafker,
G. Tectonics,
Academy Pulpan, Inlet,
in The Great Alaska
of Sciences,
Washington,
H., and C. Frohlich, Alaska,
determined
Geometry
DC,
Earthquake
Geology,
pp 47-122,
National
1971.
of the subducted
from relocated
of 1964,
earthquake
plate near Kodiak hypocenters,
Bull.
Island Seismo.
and lower
Cook
Soc. Ant., 75,
Created March, 1999. Current version November 3, 1999 2:20 pm
791-810, Press,
Ryan, NASA
Flannery,
Computing,
S.A. Teukolsky,
Tech. Memo.
104572,
J.C. and G. Plafker,
Geophys.
University
NASA
space
Press, geodesy
Numerical
Cambridge,
Recipes,
The Art of
1986.
program-GSFC
data
analysis-1992,
1993. Tide gage measurements
Res., 96, 4325-4335,
D., D. Christensen,
Geophys.
and W.T. Vetterling,
pp. 507, Cambridge
J.W., C. Ma, and D.S. Caprette.
Savage,
Zhao,
1985.
W.H., B.E
Scientific
18
along
the south
coast
of Alaska,
J.
1991.
and H. Palpan,
Res., 100, 6487-6504,
of uplift
1995.
Tomographic _-
imaging
of the Alaska
subduction
zone, J.
Created March,1999. Current version November 3,19992:20pm Table
1: Tide
19
Gauge
Site
Sites
Comment
Cordova
Record middle
is among the most complete to late 1960's.
Valdez
No data
Seward
Some
Kodiak
The operational
with significant
temporal
gaps
only in the
< prior
to 1975, but from
missing
moved
data, but the record tide gauge
a few kilometers
sites and the leveling gauges
was located
to Women's result
postseismic
at St. Paul Harbor Bay. A leveling
was verified
at both sites for two months to within
over the entire
complete.
until
survey
the two
operation
in 1984. The tide gauge analyses
1984, then was
connects
by the simultaneous
1.5 cm. Separate
interval.
of tide
and leveling
data
in the offset
(1967 faster
- 1982) and Women's Bay (1985-present) records reveal a substantially uplift rate when the gauge was located at St. Paul Harbor, than at
of the St. Paul Harbor
Bay.
Strong record with only a few missing data points. There is a significant aly in the late 1970's which is not removed by the correction procedures employed
Anchorage
extends
is fairly
agree
Women's Seldovia
then on the record
herein;
Interpretation tors. First, lesser obvious annual
the subsequent
of the tide gauge as Brown
extent
eral years
however
et al. [1977]
Seward)
may exhibit
in the first decade at Anchorage
means
signal
record
is well behaved.
at Anchorage
is handicapped
and SP91 point out Anchorage a sinusoidal
following
oscillation
the earthquake.
on less than 9 monthly
by two fac(and to a much
with a period This oscillation
in our data set that of SP91 because
that are based
anom-
means.
we have Second,
of sevis less
deleted there
the are
very few data available from the early 1970's until the mid 1980's. After that the record becomes more robust. An alternative sea level record has been derived from Nikiski
the more complete
The tide gauge 1997.
operated
set of monthly through
much
mean tide data. of the 1970's,
then
was reactivated
in
Created March,1999. Current version November 3,1999 2:20 Table
2: Rate
of Apparent
Sea Level
Change,
Tide Gauge
Time Period
sea
(mm/yr) 1965-98
"
Valdez
Seward
Kodiak
Seldovia
Anchorage
Nikiski
Recent
6.7 + 0.4
analysis,
r = b + 2cAt
b(mm/yr);
2c (mm/yr2);
-0.6+1.0
to Oldest
Determinations
sea level
level analysis, At
Cordova
20
r - Most
r, quadratic r, linear
pm
Ref.
Comment
= 1998.5-t ; 0.39+0.06
(1)
1974-88
9.4
(2)
1964-88
9.7 + 0.5
(2)
1964-74
12.7 _+2.4
(3)
1975-98
- 0.2 _+0.7
1974-88
5.5 + 0.9
1965-98
- 1.2 + 0.7
1974-88
2.9
(2)
1964-88
0.1 + 1.0
(2)
1964-73
- 11.3 +3.7
(3)
1985-98
-8.1+
(1)
Women's
1967-98
- 14.5 + 0.7
(1)
WB and SP
1974-88
-14.8
(2)
1967-98
- 17.5 +0.8
(2)
1967-81
-21.1+_1.4
(1)
St. Paul Harbor
1965-74
- 86.8 + 38.4
(3)
SP
1966-98
-9.3+0.8
1974-88
-7.0
(2)
1964-88
-7.2+1.4
(2)
1964-74
-28.1_+8.1
(3)
1984-98
0.8+
(1)
1965-98
-0.7+0.9
1964-88
1.9 _+ 1.9
(2)
1964-74
- 14.3+7.6
(3)
1972-98
- 9.9 + 0.8
(1)
- 8.7 + 1.6 ; 0.77 + 0.14
(1) (2)
- 2.7 _+2.7 ; 0.09 _+0.15
1.1 -4.1_+2.1
; -0.66+0.11
-11.8+2.9;
0.16+0.18
1.3
(1)
(1) This study;
(2) Savage
(2)
- 18.7 + 1.7 andPlafker
[1991];
(3) Brown
et al. [1977].
(SP)
(1)
mean mean
sea level tide level
no data from through
1971-79
Bay (WB)
no data
1981
1996 after
1984
Created March,1999. Current version November 3,19992:20pm
Table
Tide Gauge Site (TGS)
3: Tide Gauge 1 and GPS Rates
TGS Mean
TGS 1996.5
Uplift
Uplift Rate 2 ._(mm/yr)
Rate
(mm/yr)
21
of Uplift
GPS Uplift
GPS Site
Rate
(mm/yr)
comment
Cordova
- 4.7 +0.4
1.8 + 1.0
4050
- 13.2 + 7.1 8
GPS: dates
Valdez
2.2 + 0.2
9.2 + 1.6
POWE
8.7 + 8.2 8
GPS: dates
Seward
3.2 + 0.7
T19
7.6 + 3.5 5
GPS:
1993-98
UAMF
9.5 + 27 5
GPS:
1995-98
KDK1
9.2 + 1.6 8
GPS: dates
VLBI6
14.8 + 7.2
VLBI:
2.7+0.9
VANDUSEN
16.1 + 6.5 5
GPS:
1995-97
11.9_+0.8
NIK/NIK2
13.5+3.2
GPS:
1993-98
Kodiak 3
Anchorage
16.5 + 0.7
4
Nikiski
1. Assumes combined derived rates. 2. Cordova, 3. Combined 4. Mean
Valdez,
effects
sea level rise and post-glacial
assumed
to have time-dependent
rebound
5
= 2 mm/y
rate.
Bay and St. Paul Harbor.
Analysis.
5. Update of Freymueller 6. VLBI measurement. 7. Ryan
of eustatic
and Kodiak
Women's
Tide Level
7.4 _+2.1
et al. [1999]
to include
1998 data.
et al. [1993].
8. Previously
unreported
GPS data processed
with update
to Freymueller
et al. [1999].
1984_90
for the tide gauge-
Created March,1999. Current version November 3,19992:20pm
22
Figure Captions Figure 1:Map of southcentralAlaskashowingmajorcities,tide gaugelocations(n.b.Nikiski is immediatelynorthwestof Kenai),andcoseismic.upliftpatternfor 1964PrinceWilliam Sound Earthquake.From Plafker Figure squares
2: Uncorrected
3: Sea level
Figure
4: Corrected
Figure
5: Uplift
dislocation Figure creep
(circles)
line with slope
Figure
at depth
as given
height
++
and corrected
(corrected
sea level heights
annual
sea level heights
sea level heights
and linear
and quadratic
from axis of maximum
for elastic
deformation
rate of western
Kenai
in an elastic
(squares)
and linear
least
on Table 2.
residuals
rate vs distance
model
6: Uplift
[1971].
halfspace
due to a locked
Peninsula (solid
line).
minus
least
least squares
coseismic megathrust
sites (data points)
squares
line).
fits.
subsidence'(data (dashed and predicted
points)
and
line). uplift
rate due to
/
Figure
1
Cordova
Valdez
4.1
4.3
E4.0
99
oo ,'
o ,,',
E4.2
,
> _J
-u3.9
-
,_ Q
"4.1
CO
P9
co
E3.8
-E 4.0.
P
o
o
_
_.3.7
4)
"4.1
[]
4) CO
_4.0 _3.9
3.8 1960
= 1970
, 1980 Time, yr
i 1990
i 2000
Figure
4
Elastic
Dislocation
Model:
vel=55
mm/yr,
dip=5
deg; depth=5-25
20
_- Kodiak-avg _Seldovial_ _;J_likiski
!.__
//"
10._.
_.
/
{ Kokiak-199_.5 eward
#: ¢::
F Valdez-1996.5
4-"_/_nch°rage
" Valdez-avg
% "_ _ _"-
_---
.
Cordova-av,
I
-50 Distance from Axis
I
0 of Maximum
I
I
50 Coseismic
Figure
100 Subsidence,
5
150 km
k
Deep Slip Model: vel=50 mm/yr, dip=25 deg; depth--40-80 km Updip Tip of Fault Midway between Seldovia and Nikiski
20 ._
// _'kiski 10 E E
\\
Seldovia o_\
\ \ \ \ \
e_
_
0
-100
-50 0 50 Distance from Axis of Maximum Coseismic
100 Subsidence,
Figure 6
150 km